* reads no longer cached since kernel 4.19
[not found] <b039d510-9b03-e6a3-499a-1dbe72764cbe@postgarage.at>
@ 2020-02-12 6:02 ` Postgarage Graz IT
2020-02-12 9:33 ` Matthias Ferdinand
2020-02-17 14:31 ` Nix
0 siblings, 2 replies; 7+ messages in thread
From: Postgarage Graz IT @ 2020-02-12 6:02 UTC (permalink / raw)
To: linux-bcache
On 10.02.20 17:10, Ville Aakko wrote:
> Hi,
>
> A fellow user responding here.
>
> I've noticed similar behavior and have asked on this same mailing list
> previously. See:
> https://www.spinics.net/lists/linux-bcache/msg07859.html
>
> Also seems there are other users with this issue on the Arch Forum,
> where I have also started a discussion:
> https://bbs.archlinux.org/viewtopic.php?id=250525
> There is yet to be a single user to reply there (or on this mailing
> list) claiming they have a working setup (for caching reads).
>
> Judging from the Arch Linux thread, I have a hunch there were some
> changes ~4.18, which broke read caching for many (all?) desktop users
> (as anything which is flagged as readahed will not be cached, despite
> setting sequential_cutoff). Also (again from the Arch thread) a
> planned patch might enable expected read caching: "[PATCH 3/5] bcache:
> add readahead cache policy options via sysfs interface" / see:
> https://www.spinics.net/lists/linux-bcache/msg08074.html
Indeed that patch works.
Now I'm using the 5.6-rc1 kernel and the performance gain is huge.
I only wonder, why the used cache number doesn't go up anymore like it
did for pre-4.19 kernels.
Linux kkb 5.6.0-050600rc1-generic #202002092032 SMP Mon Feb 10 01:36:50
UTC 2020 x86_64 x86_64 x86_64 GNU/Linux
--- bcache ---
Device /dev/md0 (9:0)
UUID d4f0e4cd-c2dc-4cec-bf5b-96f1f87ff0b8
Block Size 0.50KiB
Bucket Size 512.00KiB
Congested? False
Read Congestion 0.0ms
Write Congestion 0.0ms
Total Cache Size 111.79GiB
Total Cache Used 1.12GiB (0%)
Total Cache Unused 110.67GiB (99%)
Evictable Cache 111.79GiB (100%)
Replacement Policy [lru] fifo random
Cache Mode writethrough [writeback] writearound none
Total Hits 0
Total Misses 0
Total Bypass Hits 0
Total Bypass Misses 0
Total Bypassed 0B
1st pass
0.00user 0.60system 0:08.66elapsed 6%CPU (0avgtext+0avgdata
2220maxresident)k
2097632inputs+0outputs (1major+115minor)pagefaults 0swaps
2nd pass
0.00user 0.47system 0:03.34elapsed 14%CPU (0avgtext+0avgdata
2216maxresident)k
2097128inputs+0outputs (1major+113minor)pagefaults 0swaps
3rd pass
0.00user 0.45system 0:02.58elapsed 17%CPU (0avgtext+0avgdata
2096maxresident)k
2097296inputs+0outputs (1major+110minor)pagefaults 0swaps
--- bcache ---
Device /dev/md0 (9:0)
UUID d4f0e4cd-c2dc-4cec-bf5b-96f1f87ff0b8
Block Size 0.50KiB
Bucket Size 512.00KiB
Congested? False
Read Congestion 0.0ms
Write Congestion 0.0ms
Total Cache Size 111.79GiB
Total Cache Used 1.12GiB (0%)
Total Cache Unused 110.67GiB (99%)
Evictable Cache 110.67GiB (99%)
Replacement Policy [lru] fifo random
Cache Mode writethrough [writeback] writearound none
Total Hits 6352 (51%)
Total Misses 6075
Total Bypass Hits 0
Total Bypass Misses 0
As you can see, the reads must come from the SSD in the 2nd and 3rd
pass, still "Total Cache Used" stays the same.
>
> However this is highly speculative from someone not understanding file
> systems or insides of bcache or the code at all.
>
> Perhaps someone more involved can reply: is the current behavior
> expected (reads are not getting cached practically at all). Also, is
> the patch I've linked possibly going to fix the current issues?
>
> Kind Regards,
> Ville Aakko
>
>
>
> su 9. helmik. 2020 klo 23.37 Postgarage Graz IT (it@postgarage.at) kirjoitti:
>>
>> Hello!
>>
>> I noticed, that bcache is no longer caching reads on my system which
>> makes it behave like if there were only hdds.
>>
>> I'm using two hdds in a raid 1 as the backing device and a single ssd as
>> cache device:
>>
>> sda 8:0 0 111,8G 0 disk
>> └─bcache0 252:0 0 921,9G 0 disk /
>> sdb 8:16 0 931,5G 0 disk
>> ├─sdb1 8:17 0 922G 0 part
>> │ └─md0 9:0 0 921,9G 0 raid1
>> │ └─bcache0 252:0 0 921,9G 0 disk /
>> ├─sdb2 8:18 0 1K 0 part
>> ├─sdb5 8:21 0 1,9G 0 part
>> │ └─md1 9:1 0 1,9G 0 raid1 /boot
>> └─sdb6 8:22 0 7,6G 0 part [SWAP]
>> sdc 8:32 0 931,5G 0 disk
>> ├─sdc1 8:33 0 922G 0 part
>> │ └─md0 9:0 0 921,9G 0 raid1
>> │ └─bcache0 252:0 0 921,9G 0 disk /
>> ├─sdc2 8:34 0 1K 0 part
>> ├─sdc5 8:37 0 1,9G 0 part
>> │ └─md1 9:1 0 1,9G 0 raid1 /boot
>> └─sdc6 8:38 0 7,6G 0 part [SWAP]
>>
>>
>> For benchmarking every time I detach the cache device, stop the bcache
>> device, do a wipefs on the cache device, then make-bcache -C /dev/sda
>> and finally reattach the cache.
>> After that, I'm using the following script to repeatedly read a 1gb file:
>>
>> #!/bin/sh
>> echo 0 > /sys/block/bcache0/bcache/sequential_cutoff
>> echo 0 > /sys/block/bcache0/bcache/cache/congested_read_threshold_us
>> echo 0 > /sys/block/bcache0/bcache/cache/congested_write_threshold_us
>> uname -a
>> echo
>> bcache-status
>> echo
>> echo "1st pass"
>> sync; echo 3 > /proc/sys/vm/drop_caches
>> (time cat 1GB.bin > /dev/null)
>> echo
>> echo "2nd pass"
>> sync; echo 3 > /proc/sys/vm/drop_caches
>> (time cat 1GB.bin > /dev/null)
>> echo
>> echo "3rd pass"
>> sync; echo 3 > /proc/sys/vm/drop_caches
>> (time cat 1GB.bin > /dev/null)
>> bcache-status
>>
>>
>>
>> As you can see from the results below, kernel 4.18.20 is the last
>> kernel, where the cache grows and the performance goes up.
>>
>> I also compiled 4.19.0 with the bcache files from 4.18.20 and much to my
>> suprise, that didn't change 4.19's behavior - still no caching. So some
>> other changes must be the culprit or I did something wrong.
>> I'm not that much into compiling the kernel, but I checked out the
>> 4.19.0 and 4.18.20 commits and replaced the 4.19.0 drivers/md/bcache
>> directory with the one from 4.18.20 - then recompiled and installed the
>> new kernel.
>>
>> So i am at my wits end. Any help would be appreciated.
>> Thanks
>> Flo
>>
>>
>> Linux kkb 4.18.20-041820-generic #201812030624 SMP Mon Dec 3 11:25:55
>> UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
>>
>> --- bcache ---
>> Device /dev/md0 (9:0)
>> UUID 8275bf01-f0b3-423e-87fa-48336ce33068
>> Block Size 0.50KiB
>> Bucket Size 512.00KiB
>> Congested? False
>> Read Congestion 0.0ms
>> Write Congestion 0.0ms
>> Total Cache Size 111.79GiB
>> Total Cache Used 1.12GiB (0%)
>> Total Cache Unused 110.67GiB (99%)
>> Evictable Cache 111.79GiB (100%)
>> Replacement Policy [lru] fifo random
>> Cache Mode writethrough [writeback] writearound none
>> Total Hits 0 (0%)
>> Total Misses 6
>> Total Bypass Hits 0
>> Total Bypass Misses 0
>> Total Bypassed 0B
>>
>> 1st pass
>> 0.00user 0.36system 0:08.58elapsed 4%CPU (0avgtext+0avgdata
>> 2196maxresident)k
>> 2097608inputs+0outputs (1major+113minor)pagefaults 0swaps
>>
>> 2nd pass
>> 0.00user 0.32system 0:03.29elapsed 9%CPU (0avgtext+0avgdata
>> 2100maxresident)k
>> 2097184inputs+0outputs (1major+110minor)pagefaults 0swaps
>>
>> 3rd pass
>> 0.00user 0.32system 0:02.64elapsed 12%CPU (0avgtext+0avgdata
>> 2092maxresident)k
>> 2097280inputs+0outputs (1major+111minor)pagefaults 0swaps
>> --- bcache ---
>> Device /dev/md0 (9:0)
>> UUID 8275bf01-f0b3-423e-87fa-48336ce33068
>> Block Size 0.50KiB
>> Bucket Size 512.00KiB
>> Congested? False
>> Read Congestion 0.0ms
>> Write Congestion 0.0ms
>> Total Cache Size 111.79GiB
>> Total Cache Used 2.24GiB (2%)
>> Total Cache Unused 109.55GiB (98%)
>> Evictable Cache 110.67GiB (99%)
>> Replacement Policy [lru] fifo random
>> Cache Mode writethrough [writeback] writearound none
>> Total Hits 5 (0%)
>> Total Misses 4079
>> Total Bypass Hits 0 (0%)
>> Total Bypass Misses 615
>> Total Bypassed 2.40MiB
>>
>>
>>
>> Linux kkb 4.19.0-041900-generic #201810221809 SMP Mon Oct 22 22:11:45
>> UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
>>
>> --- bcache ---
>> Device /dev/md0 (9:0)
>> UUID 67269654-92e8-4c3b-a524-8e8910082146
>> Block Size 0.50KiB
>> Bucket Size 512.00KiB
>> Congested? False
>> Read Congestion 0.0ms
>> Write Congestion 0.0ms
>> Total Cache Size 111.79GiB
>> Total Cache Used 1.12GiB (0%)
>> Total Cache Unused 110.67GiB (99%)
>> Evictable Cache 111.79GiB (100%)
>> Replacement Policy [lru] fifo random
>> Cache Mode writethrough [writeback] writearound none
>> Total Hits 0 (0%)
>> Total Misses 1
>> Total Bypass Hits 0
>> Total Bypass Misses 0
>> Total Bypassed 0B
>>
>> 1st pass
>> 0.00user 0.33system 0:09.29elapsed 3%CPU (0avgtext+0avgdata
>> 2280maxresident)k
>> 2097624inputs+0outputs (1major+113minor)pagefaults 0swaps
>>
>> 2nd pass
>> 0.00user 0.33system 0:08.47elapsed 4%CPU (0avgtext+0avgdata
>> 2248maxresident)k
>> 2097280inputs+0outputs (1major+111minor)pagefaults 0swaps
>>
>> 3rd pass
>> 0.00user 0.37system 0:10.46elapsed 3%CPU (0avgtext+0avgdata
>> 2220maxresident)k
>> 2097616inputs+0outputs (1major+114minor)pagefaults 0swaps
>> --- bcache ---
>> Device /dev/md0 (9:0)
>> UUID 67269654-92e8-4c3b-a524-8e8910082146
>> Block Size 0.50KiB
>> Bucket Size 512.00KiB
>> Congested? False
>> Read Congestion 0.0ms
>> Write Congestion 0.0ms
>> Total Cache Size 111.79GiB
>> Total Cache Used 1.12GiB (0%)
>> Total Cache Unused 110.67GiB (99%)
>> Evictable Cache 111.79GiB (100%)
>> Replacement Policy [lru] fifo random
>> Cache Mode writethrough [writeback] writearound none
>> Total Hits 132 (23%)
>> Total Misses 436
>> Total Bypass Hits 51 (0%)
>> Total Bypass Misses 17399
>> Total Bypassed 43.50MiB
>>
>>
>>
>>
>> Linux kkb 5.5.2-050502-generic #202002041931 SMP Tue Feb 4 19:33:15 UTC
>> 2020 x86_64 x86_64 x86_64 GNU/Linux
>>
>> --- bcache ---
>> Device /dev/md0 (9:0)
>> UUID 38a8b675-e332-4076-b0cf-44e4be72c300
>> Block Size 0.50KiB
>> Bucket Size 512.00KiB
>> Congested? False
>> Read Congestion 0.0ms
>> Write Congestion 0.0ms
>> Total Cache Size 111.79GiB
>> Total Cache Used 1.12GiB (0%)
>> Total Cache Unused 110.67GiB (99%)
>> Evictable Cache 111.79GiB (100%)
>> Replacement Policy [lru] fifo random
>> Cache Mode writethrough [writeback] writearound none
>> Total Hits 0 (0%)
>> Total Misses 1
>> Total Bypass Hits 0 (0%)
>> Total Bypass Misses 3
>> Total Bypassed 52.00KiB
>>
>> 1st pass
>> 0.00user 0.42system 0:09.21elapsed 4%CPU (0avgtext+0avgdata
>> 2216maxresident)k
>> 2097608inputs+0outputs (1major+112minor)pagefaults 0swaps
>>
>> 2nd pass
>> 0.00user 0.42system 0:09.62elapsed 4%CPU (0avgtext+0avgdata
>> 2248maxresident)k
>> 2097280inputs+0outputs (1major+112minor)pagefaults 0swaps
>>
>> 3rd pass
>> 0.00user 0.43system 0:08.75elapsed 5%CPU (0avgtext+0avgdata
>> 2220maxresident)k
>> 2097224inputs+0outputs (1major+114minor)pagefaults 0swaps
>> --- bcache ---
>> Device /dev/md0 (9:0)
>> UUID 38a8b675-e332-4076-b0cf-44e4be72c300
>> Block Size 0.50KiB
>> Bucket Size 512.00KiB
>> Congested? False
>> Read Congestion 0.0ms
>> Write Congestion 0.0ms
>> Total Cache Size 111.79GiB
>> Total Cache Used 1.12GiB (0%)
>> Total Cache Unused 110.67GiB (99%)
>> Evictable Cache 111.79GiB (100%)
>> Replacement Policy [lru] fifo random
>> Cache Mode writethrough [writeback] writearound none
>> Total Hits 121 (32%)
>> Total Misses 246
>> Total Bypass Hits 15 (0%)
>> Total Bypass Misses 12811
>> Total Bypassed 39.70MiB
>
>
>
> --
> --
> Ville Aakko - ville.aakko@gmail.com
>
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: reads no longer cached since kernel 4.19
2020-02-12 6:02 ` reads no longer cached since kernel 4.19 Postgarage Graz IT
@ 2020-02-12 9:33 ` Matthias Ferdinand
2020-02-12 10:03 ` Postgarage Graz IT
2020-02-17 14:31 ` Nix
1 sibling, 1 reply; 7+ messages in thread
From: Matthias Ferdinand @ 2020-02-12 9:33 UTC (permalink / raw)
To: Postgarage Graz IT; +Cc: linux-bcache
Sorry for being late to reply. Your bcache is set to "writeback", and
your test is doing reads from a bcache with a still empty cache device;
while the "1GB" file resides initially only on the backing device.
My understanding so far was that in writeback mode only writes will ever
add data to the caching device (first as dirty, and post-writeback as
clean for as long as as space restrictions / usage patterns allow).
Therefore I would not have expected any piece of your "1GB" file to ever
be added to your caching device, and reading it would have to resort to
the backing device, even when repeatedly reading it.
I have seen discussion about caching of readahead requests for newer
kernels, and I thought (but never really checked) this would only
concern read-caching modes. Am I wrong in this assumption?
Matthias
On Wed, Feb 12, 2020 at 07:02:28AM +0100, Postgarage Graz IT wrote:
> On 10.02.20 17:10, Ville Aakko wrote:
> > Hi,
> >
> > A fellow user responding here.
> >
> > I've noticed similar behavior and have asked on this same mailing list
> > previously. See:
> > https://www.spinics.net/lists/linux-bcache/msg07859.html
> >
> > Also seems there are other users with this issue on the Arch Forum,
> > where I have also started a discussion:
> > https://bbs.archlinux.org/viewtopic.php?id=250525
> > There is yet to be a single user to reply there (or on this mailing
> > list) claiming they have a working setup (for caching reads).
> >
> > Judging from the Arch Linux thread, I have a hunch there were some
> > changes ~4.18, which broke read caching for many (all?) desktop users
> > (as anything which is flagged as readahed will not be cached, despite
> > setting sequential_cutoff). Also (again from the Arch thread) a
> > planned patch might enable expected read caching: "[PATCH 3/5] bcache:
> > add readahead cache policy options via sysfs interface" / see:
> > https://www.spinics.net/lists/linux-bcache/msg08074.html
>
> Indeed that patch works.
> Now I'm using the 5.6-rc1 kernel and the performance gain is huge.
>
> I only wonder, why the used cache number doesn't go up anymore like it
> did for pre-4.19 kernels.
>
>
> Linux kkb 5.6.0-050600rc1-generic #202002092032 SMP Mon Feb 10 01:36:50
> UTC 2020 x86_64 x86_64 x86_64 GNU/Linux
>
> --- bcache ---
> Device /dev/md0 (9:0)
> UUID d4f0e4cd-c2dc-4cec-bf5b-96f1f87ff0b8
> Block Size 0.50KiB
> Bucket Size 512.00KiB
> Congested? False
> Read Congestion 0.0ms
> Write Congestion 0.0ms
> Total Cache Size 111.79GiB
> Total Cache Used 1.12GiB (0%)
> Total Cache Unused 110.67GiB (99%)
> Evictable Cache 111.79GiB (100%)
> Replacement Policy [lru] fifo random
> Cache Mode writethrough [writeback] writearound none
> Total Hits 0
> Total Misses 0
> Total Bypass Hits 0
> Total Bypass Misses 0
> Total Bypassed 0B
>
> 1st pass
> 0.00user 0.60system 0:08.66elapsed 6%CPU (0avgtext+0avgdata
> 2220maxresident)k
> 2097632inputs+0outputs (1major+115minor)pagefaults 0swaps
>
> 2nd pass
> 0.00user 0.47system 0:03.34elapsed 14%CPU (0avgtext+0avgdata
> 2216maxresident)k
> 2097128inputs+0outputs (1major+113minor)pagefaults 0swaps
>
> 3rd pass
> 0.00user 0.45system 0:02.58elapsed 17%CPU (0avgtext+0avgdata
> 2096maxresident)k
> 2097296inputs+0outputs (1major+110minor)pagefaults 0swaps
> --- bcache ---
> Device /dev/md0 (9:0)
> UUID d4f0e4cd-c2dc-4cec-bf5b-96f1f87ff0b8
> Block Size 0.50KiB
> Bucket Size 512.00KiB
> Congested? False
> Read Congestion 0.0ms
> Write Congestion 0.0ms
> Total Cache Size 111.79GiB
> Total Cache Used 1.12GiB (0%)
> Total Cache Unused 110.67GiB (99%)
> Evictable Cache 110.67GiB (99%)
> Replacement Policy [lru] fifo random
> Cache Mode writethrough [writeback] writearound none
> Total Hits 6352 (51%)
> Total Misses 6075
> Total Bypass Hits 0
> Total Bypass Misses 0
>
>
> As you can see, the reads must come from the SSD in the 2nd and 3rd
> pass, still "Total Cache Used" stays the same.
>
>
> >
> > However this is highly speculative from someone not understanding file
> > systems or insides of bcache or the code at all.
> >
> > Perhaps someone more involved can reply: is the current behavior
> > expected (reads are not getting cached practically at all). Also, is
> > the patch I've linked possibly going to fix the current issues?
> >
> > Kind Regards,
> > Ville Aakko
> >
> >
> >
> > su 9. helmik. 2020 klo 23.37 Postgarage Graz IT (it@postgarage.at) kirjoitti:
> >>
> >> Hello!
> >>
> >> I noticed, that bcache is no longer caching reads on my system which
> >> makes it behave like if there were only hdds.
> >>
> >> I'm using two hdds in a raid 1 as the backing device and a single ssd as
> >> cache device:
> >>
> >> sda 8:0 0 111,8G 0 disk
> >> └─bcache0 252:0 0 921,9G 0 disk /
> >> sdb 8:16 0 931,5G 0 disk
> >> ├─sdb1 8:17 0 922G 0 part
> >> │ └─md0 9:0 0 921,9G 0 raid1
> >> │ └─bcache0 252:0 0 921,9G 0 disk /
> >> ├─sdb2 8:18 0 1K 0 part
> >> ├─sdb5 8:21 0 1,9G 0 part
> >> │ └─md1 9:1 0 1,9G 0 raid1 /boot
> >> └─sdb6 8:22 0 7,6G 0 part [SWAP]
> >> sdc 8:32 0 931,5G 0 disk
> >> ├─sdc1 8:33 0 922G 0 part
> >> │ └─md0 9:0 0 921,9G 0 raid1
> >> │ └─bcache0 252:0 0 921,9G 0 disk /
> >> ├─sdc2 8:34 0 1K 0 part
> >> ├─sdc5 8:37 0 1,9G 0 part
> >> │ └─md1 9:1 0 1,9G 0 raid1 /boot
> >> └─sdc6 8:38 0 7,6G 0 part [SWAP]
> >>
> >>
> >> For benchmarking every time I detach the cache device, stop the bcache
> >> device, do a wipefs on the cache device, then make-bcache -C /dev/sda
> >> and finally reattach the cache.
> >> After that, I'm using the following script to repeatedly read a 1gb file:
> >>
> >> #!/bin/sh
> >> echo 0 > /sys/block/bcache0/bcache/sequential_cutoff
> >> echo 0 > /sys/block/bcache0/bcache/cache/congested_read_threshold_us
> >> echo 0 > /sys/block/bcache0/bcache/cache/congested_write_threshold_us
> >> uname -a
> >> echo
> >> bcache-status
> >> echo
> >> echo "1st pass"
> >> sync; echo 3 > /proc/sys/vm/drop_caches
> >> (time cat 1GB.bin > /dev/null)
> >> echo
> >> echo "2nd pass"
> >> sync; echo 3 > /proc/sys/vm/drop_caches
> >> (time cat 1GB.bin > /dev/null)
> >> echo
> >> echo "3rd pass"
> >> sync; echo 3 > /proc/sys/vm/drop_caches
> >> (time cat 1GB.bin > /dev/null)
> >> bcache-status
> >>
> >>
> >>
> >> As you can see from the results below, kernel 4.18.20 is the last
> >> kernel, where the cache grows and the performance goes up.
> >>
> >> I also compiled 4.19.0 with the bcache files from 4.18.20 and much to my
> >> suprise, that didn't change 4.19's behavior - still no caching. So some
> >> other changes must be the culprit or I did something wrong.
> >> I'm not that much into compiling the kernel, but I checked out the
> >> 4.19.0 and 4.18.20 commits and replaced the 4.19.0 drivers/md/bcache
> >> directory with the one from 4.18.20 - then recompiled and installed the
> >> new kernel.
> >>
> >> So i am at my wits end. Any help would be appreciated.
> >> Thanks
> >> Flo
> >>
> >>
> >> Linux kkb 4.18.20-041820-generic #201812030624 SMP Mon Dec 3 11:25:55
> >> UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
> >>
> >> --- bcache ---
> >> Device /dev/md0 (9:0)
> >> UUID 8275bf01-f0b3-423e-87fa-48336ce33068
> >> Block Size 0.50KiB
> >> Bucket Size 512.00KiB
> >> Congested? False
> >> Read Congestion 0.0ms
> >> Write Congestion 0.0ms
> >> Total Cache Size 111.79GiB
> >> Total Cache Used 1.12GiB (0%)
> >> Total Cache Unused 110.67GiB (99%)
> >> Evictable Cache 111.79GiB (100%)
> >> Replacement Policy [lru] fifo random
> >> Cache Mode writethrough [writeback] writearound none
> >> Total Hits 0 (0%)
> >> Total Misses 6
> >> Total Bypass Hits 0
> >> Total Bypass Misses 0
> >> Total Bypassed 0B
> >>
> >> 1st pass
> >> 0.00user 0.36system 0:08.58elapsed 4%CPU (0avgtext+0avgdata
> >> 2196maxresident)k
> >> 2097608inputs+0outputs (1major+113minor)pagefaults 0swaps
> >>
> >> 2nd pass
> >> 0.00user 0.32system 0:03.29elapsed 9%CPU (0avgtext+0avgdata
> >> 2100maxresident)k
> >> 2097184inputs+0outputs (1major+110minor)pagefaults 0swaps
> >>
> >> 3rd pass
> >> 0.00user 0.32system 0:02.64elapsed 12%CPU (0avgtext+0avgdata
> >> 2092maxresident)k
> >> 2097280inputs+0outputs (1major+111minor)pagefaults 0swaps
> >> --- bcache ---
> >> Device /dev/md0 (9:0)
> >> UUID 8275bf01-f0b3-423e-87fa-48336ce33068
> >> Block Size 0.50KiB
> >> Bucket Size 512.00KiB
> >> Congested? False
> >> Read Congestion 0.0ms
> >> Write Congestion 0.0ms
> >> Total Cache Size 111.79GiB
> >> Total Cache Used 2.24GiB (2%)
> >> Total Cache Unused 109.55GiB (98%)
> >> Evictable Cache 110.67GiB (99%)
> >> Replacement Policy [lru] fifo random
> >> Cache Mode writethrough [writeback] writearound none
> >> Total Hits 5 (0%)
> >> Total Misses 4079
> >> Total Bypass Hits 0 (0%)
> >> Total Bypass Misses 615
> >> Total Bypassed 2.40MiB
> >>
> >>
> >>
> >> Linux kkb 4.19.0-041900-generic #201810221809 SMP Mon Oct 22 22:11:45
> >> UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
> >>
> >> --- bcache ---
> >> Device /dev/md0 (9:0)
> >> UUID 67269654-92e8-4c3b-a524-8e8910082146
> >> Block Size 0.50KiB
> >> Bucket Size 512.00KiB
> >> Congested? False
> >> Read Congestion 0.0ms
> >> Write Congestion 0.0ms
> >> Total Cache Size 111.79GiB
> >> Total Cache Used 1.12GiB (0%)
> >> Total Cache Unused 110.67GiB (99%)
> >> Evictable Cache 111.79GiB (100%)
> >> Replacement Policy [lru] fifo random
> >> Cache Mode writethrough [writeback] writearound none
> >> Total Hits 0 (0%)
> >> Total Misses 1
> >> Total Bypass Hits 0
> >> Total Bypass Misses 0
> >> Total Bypassed 0B
> >>
> >> 1st pass
> >> 0.00user 0.33system 0:09.29elapsed 3%CPU (0avgtext+0avgdata
> >> 2280maxresident)k
> >> 2097624inputs+0outputs (1major+113minor)pagefaults 0swaps
> >>
> >> 2nd pass
> >> 0.00user 0.33system 0:08.47elapsed 4%CPU (0avgtext+0avgdata
> >> 2248maxresident)k
> >> 2097280inputs+0outputs (1major+111minor)pagefaults 0swaps
> >>
> >> 3rd pass
> >> 0.00user 0.37system 0:10.46elapsed 3%CPU (0avgtext+0avgdata
> >> 2220maxresident)k
> >> 2097616inputs+0outputs (1major+114minor)pagefaults 0swaps
> >> --- bcache ---
> >> Device /dev/md0 (9:0)
> >> UUID 67269654-92e8-4c3b-a524-8e8910082146
> >> Block Size 0.50KiB
> >> Bucket Size 512.00KiB
> >> Congested? False
> >> Read Congestion 0.0ms
> >> Write Congestion 0.0ms
> >> Total Cache Size 111.79GiB
> >> Total Cache Used 1.12GiB (0%)
> >> Total Cache Unused 110.67GiB (99%)
> >> Evictable Cache 111.79GiB (100%)
> >> Replacement Policy [lru] fifo random
> >> Cache Mode writethrough [writeback] writearound none
> >> Total Hits 132 (23%)
> >> Total Misses 436
> >> Total Bypass Hits 51 (0%)
> >> Total Bypass Misses 17399
> >> Total Bypassed 43.50MiB
> >>
> >>
> >>
> >>
> >> Linux kkb 5.5.2-050502-generic #202002041931 SMP Tue Feb 4 19:33:15 UTC
> >> 2020 x86_64 x86_64 x86_64 GNU/Linux
> >>
> >> --- bcache ---
> >> Device /dev/md0 (9:0)
> >> UUID 38a8b675-e332-4076-b0cf-44e4be72c300
> >> Block Size 0.50KiB
> >> Bucket Size 512.00KiB
> >> Congested? False
> >> Read Congestion 0.0ms
> >> Write Congestion 0.0ms
> >> Total Cache Size 111.79GiB
> >> Total Cache Used 1.12GiB (0%)
> >> Total Cache Unused 110.67GiB (99%)
> >> Evictable Cache 111.79GiB (100%)
> >> Replacement Policy [lru] fifo random
> >> Cache Mode writethrough [writeback] writearound none
> >> Total Hits 0 (0%)
> >> Total Misses 1
> >> Total Bypass Hits 0 (0%)
> >> Total Bypass Misses 3
> >> Total Bypassed 52.00KiB
> >>
> >> 1st pass
> >> 0.00user 0.42system 0:09.21elapsed 4%CPU (0avgtext+0avgdata
> >> 2216maxresident)k
> >> 2097608inputs+0outputs (1major+112minor)pagefaults 0swaps
> >>
> >> 2nd pass
> >> 0.00user 0.42system 0:09.62elapsed 4%CPU (0avgtext+0avgdata
> >> 2248maxresident)k
> >> 2097280inputs+0outputs (1major+112minor)pagefaults 0swaps
> >>
> >> 3rd pass
> >> 0.00user 0.43system 0:08.75elapsed 5%CPU (0avgtext+0avgdata
> >> 2220maxresident)k
> >> 2097224inputs+0outputs (1major+114minor)pagefaults 0swaps
> >> --- bcache ---
> >> Device /dev/md0 (9:0)
> >> UUID 38a8b675-e332-4076-b0cf-44e4be72c300
> >> Block Size 0.50KiB
> >> Bucket Size 512.00KiB
> >> Congested? False
> >> Read Congestion 0.0ms
> >> Write Congestion 0.0ms
> >> Total Cache Size 111.79GiB
> >> Total Cache Used 1.12GiB (0%)
> >> Total Cache Unused 110.67GiB (99%)
> >> Evictable Cache 111.79GiB (100%)
> >> Replacement Policy [lru] fifo random
> >> Cache Mode writethrough [writeback] writearound none
> >> Total Hits 121 (32%)
> >> Total Misses 246
> >> Total Bypass Hits 15 (0%)
> >> Total Bypass Misses 12811
> >> Total Bypassed 39.70MiB
> >
> >
> >
> > --
> > --
> > Ville Aakko - ville.aakko@gmail.com
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: reads no longer cached since kernel 4.19
2020-02-12 9:33 ` Matthias Ferdinand
@ 2020-02-12 10:03 ` Postgarage Graz IT
2020-02-12 20:32 ` Ville Aakko
0 siblings, 1 reply; 7+ messages in thread
From: Postgarage Graz IT @ 2020-02-12 10:03 UTC (permalink / raw)
To: Matthias Ferdinand; +Cc: linux-bcache
On 12.02.20 10:33, Matthias Ferdinand wrote:
> Sorry for being late to reply. Your bcache is set to "writeback", and
> your test is doing reads from a bcache with a still empty cache device;
> while the "1GB" file resides initially only on the backing device.
>
> My understanding so far was that in writeback mode only writes will ever
> add data to the caching device (first as dirty, and post-writeback as
> clean for as long as as space restrictions / usage patterns allow).
> Therefore I would not have expected any piece of your "1GB" file to ever
> be added to your caching device, and reading it would have to resort to
> the backing device, even when repeatedly reading it.
I would have said, that in writeback mode bcache acts as a read/write
cache caching both reads and writes.
Furthermore I also did the benchmark with all three caching modes and
all behaved the same.
As you can see, with kernel 5.6 I've got back the behavior like with
kernels >= 4.18, the read is cached in writeback mode.
But perhaps I'm wrong and this shouldn't happen?
>
> I have seen discussion about caching of readahead requests for newer
> kernels, and I thought (but never really checked) this would only
> concern read-caching modes. Am I wrong in this assumption?
>
> Matthias
>
>
> On Wed, Feb 12, 2020 at 07:02:28AM +0100, Postgarage Graz IT wrote:
>> On 10.02.20 17:10, Ville Aakko wrote:
>>> Hi,
>>>
>>> A fellow user responding here.
>>>
>>> I've noticed similar behavior and have asked on this same mailing list
>>> previously. See:
>>> https://www.spinics.net/lists/linux-bcache/msg07859.html
>>>
>>> Also seems there are other users with this issue on the Arch Forum,
>>> where I have also started a discussion:
>>> https://bbs.archlinux.org/viewtopic.php?id=250525
>>> There is yet to be a single user to reply there (or on this mailing
>>> list) claiming they have a working setup (for caching reads).
>>>
>>> Judging from the Arch Linux thread, I have a hunch there were some
>>> changes ~4.18, which broke read caching for many (all?) desktop users
>>> (as anything which is flagged as readahed will not be cached, despite
>>> setting sequential_cutoff). Also (again from the Arch thread) a
>>> planned patch might enable expected read caching: "[PATCH 3/5] bcache:
>>> add readahead cache policy options via sysfs interface" / see:
>>> https://www.spinics.net/lists/linux-bcache/msg08074.html
>>
>> Indeed that patch works.
>> Now I'm using the 5.6-rc1 kernel and the performance gain is huge.
>>
>> I only wonder, why the used cache number doesn't go up anymore like it
>> did for pre-4.19 kernels.
>>
>>
>> Linux kkb 5.6.0-050600rc1-generic #202002092032 SMP Mon Feb 10 01:36:50
>> UTC 2020 x86_64 x86_64 x86_64 GNU/Linux
>>
>> --- bcache ---
>> Device /dev/md0 (9:0)
>> UUID d4f0e4cd-c2dc-4cec-bf5b-96f1f87ff0b8
>> Block Size 0.50KiB
>> Bucket Size 512.00KiB
>> Congested? False
>> Read Congestion 0.0ms
>> Write Congestion 0.0ms
>> Total Cache Size 111.79GiB
>> Total Cache Used 1.12GiB (0%)
>> Total Cache Unused 110.67GiB (99%)
>> Evictable Cache 111.79GiB (100%)
>> Replacement Policy [lru] fifo random
>> Cache Mode writethrough [writeback] writearound none
>> Total Hits 0
>> Total Misses 0
>> Total Bypass Hits 0
>> Total Bypass Misses 0
>> Total Bypassed 0B
>>
>> 1st pass
>> 0.00user 0.60system 0:08.66elapsed 6%CPU (0avgtext+0avgdata
>> 2220maxresident)k
>> 2097632inputs+0outputs (1major+115minor)pagefaults 0swaps
>>
>> 2nd pass
>> 0.00user 0.47system 0:03.34elapsed 14%CPU (0avgtext+0avgdata
>> 2216maxresident)k
>> 2097128inputs+0outputs (1major+113minor)pagefaults 0swaps
>>
>> 3rd pass
>> 0.00user 0.45system 0:02.58elapsed 17%CPU (0avgtext+0avgdata
>> 2096maxresident)k
>> 2097296inputs+0outputs (1major+110minor)pagefaults 0swaps
>> --- bcache ---
>> Device /dev/md0 (9:0)
>> UUID d4f0e4cd-c2dc-4cec-bf5b-96f1f87ff0b8
>> Block Size 0.50KiB
>> Bucket Size 512.00KiB
>> Congested? False
>> Read Congestion 0.0ms
>> Write Congestion 0.0ms
>> Total Cache Size 111.79GiB
>> Total Cache Used 1.12GiB (0%)
>> Total Cache Unused 110.67GiB (99%)
>> Evictable Cache 110.67GiB (99%)
>> Replacement Policy [lru] fifo random
>> Cache Mode writethrough [writeback] writearound none
>> Total Hits 6352 (51%)
>> Total Misses 6075
>> Total Bypass Hits 0
>> Total Bypass Misses 0
>>
>>
>> As you can see, the reads must come from the SSD in the 2nd and 3rd
>> pass, still "Total Cache Used" stays the same.
>>
>>
>>>
>>> However this is highly speculative from someone not understanding file
>>> systems or insides of bcache or the code at all.
>>>
>>> Perhaps someone more involved can reply: is the current behavior
>>> expected (reads are not getting cached practically at all). Also, is
>>> the patch I've linked possibly going to fix the current issues?
>>>
>>> Kind Regards,
>>> Ville Aakko
>>>
>>>
>>>
>>> su 9. helmik. 2020 klo 23.37 Postgarage Graz IT (it@postgarage.at) kirjoitti:
>>>>
>>>> Hello!
>>>>
>>>> I noticed, that bcache is no longer caching reads on my system which
>>>> makes it behave like if there were only hdds.
>>>>
>>>> I'm using two hdds in a raid 1 as the backing device and a single ssd as
>>>> cache device:
>>>>
>>>> sda 8:0 0 111,8G 0 disk
>>>> └─bcache0 252:0 0 921,9G 0 disk /
>>>> sdb 8:16 0 931,5G 0 disk
>>>> ├─sdb1 8:17 0 922G 0 part
>>>> │ └─md0 9:0 0 921,9G 0 raid1
>>>> │ └─bcache0 252:0 0 921,9G 0 disk /
>>>> ├─sdb2 8:18 0 1K 0 part
>>>> ├─sdb5 8:21 0 1,9G 0 part
>>>> │ └─md1 9:1 0 1,9G 0 raid1 /boot
>>>> └─sdb6 8:22 0 7,6G 0 part [SWAP]
>>>> sdc 8:32 0 931,5G 0 disk
>>>> ├─sdc1 8:33 0 922G 0 part
>>>> │ └─md0 9:0 0 921,9G 0 raid1
>>>> │ └─bcache0 252:0 0 921,9G 0 disk /
>>>> ├─sdc2 8:34 0 1K 0 part
>>>> ├─sdc5 8:37 0 1,9G 0 part
>>>> │ └─md1 9:1 0 1,9G 0 raid1 /boot
>>>> └─sdc6 8:38 0 7,6G 0 part [SWAP]
>>>>
>>>>
>>>> For benchmarking every time I detach the cache device, stop the bcache
>>>> device, do a wipefs on the cache device, then make-bcache -C /dev/sda
>>>> and finally reattach the cache.
>>>> After that, I'm using the following script to repeatedly read a 1gb file:
>>>>
>>>> #!/bin/sh
>>>> echo 0 > /sys/block/bcache0/bcache/sequential_cutoff
>>>> echo 0 > /sys/block/bcache0/bcache/cache/congested_read_threshold_us
>>>> echo 0 > /sys/block/bcache0/bcache/cache/congested_write_threshold_us
>>>> uname -a
>>>> echo
>>>> bcache-status
>>>> echo
>>>> echo "1st pass"
>>>> sync; echo 3 > /proc/sys/vm/drop_caches
>>>> (time cat 1GB.bin > /dev/null)
>>>> echo
>>>> echo "2nd pass"
>>>> sync; echo 3 > /proc/sys/vm/drop_caches
>>>> (time cat 1GB.bin > /dev/null)
>>>> echo
>>>> echo "3rd pass"
>>>> sync; echo 3 > /proc/sys/vm/drop_caches
>>>> (time cat 1GB.bin > /dev/null)
>>>> bcache-status
>>>>
>>>>
>>>>
>>>> As you can see from the results below, kernel 4.18.20 is the last
>>>> kernel, where the cache grows and the performance goes up.
>>>>
>>>> I also compiled 4.19.0 with the bcache files from 4.18.20 and much to my
>>>> suprise, that didn't change 4.19's behavior - still no caching. So some
>>>> other changes must be the culprit or I did something wrong.
>>>> I'm not that much into compiling the kernel, but I checked out the
>>>> 4.19.0 and 4.18.20 commits and replaced the 4.19.0 drivers/md/bcache
>>>> directory with the one from 4.18.20 - then recompiled and installed the
>>>> new kernel.
>>>>
>>>> So i am at my wits end. Any help would be appreciated.
>>>> Thanks
>>>> Flo
>>>>
>>>>
>>>> Linux kkb 4.18.20-041820-generic #201812030624 SMP Mon Dec 3 11:25:55
>>>> UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
>>>>
>>>> --- bcache ---
>>>> Device /dev/md0 (9:0)
>>>> UUID 8275bf01-f0b3-423e-87fa-48336ce33068
>>>> Block Size 0.50KiB
>>>> Bucket Size 512.00KiB
>>>> Congested? False
>>>> Read Congestion 0.0ms
>>>> Write Congestion 0.0ms
>>>> Total Cache Size 111.79GiB
>>>> Total Cache Used 1.12GiB (0%)
>>>> Total Cache Unused 110.67GiB (99%)
>>>> Evictable Cache 111.79GiB (100%)
>>>> Replacement Policy [lru] fifo random
>>>> Cache Mode writethrough [writeback] writearound none
>>>> Total Hits 0 (0%)
>>>> Total Misses 6
>>>> Total Bypass Hits 0
>>>> Total Bypass Misses 0
>>>> Total Bypassed 0B
>>>>
>>>> 1st pass
>>>> 0.00user 0.36system 0:08.58elapsed 4%CPU (0avgtext+0avgdata
>>>> 2196maxresident)k
>>>> 2097608inputs+0outputs (1major+113minor)pagefaults 0swaps
>>>>
>>>> 2nd pass
>>>> 0.00user 0.32system 0:03.29elapsed 9%CPU (0avgtext+0avgdata
>>>> 2100maxresident)k
>>>> 2097184inputs+0outputs (1major+110minor)pagefaults 0swaps
>>>>
>>>> 3rd pass
>>>> 0.00user 0.32system 0:02.64elapsed 12%CPU (0avgtext+0avgdata
>>>> 2092maxresident)k
>>>> 2097280inputs+0outputs (1major+111minor)pagefaults 0swaps
>>>> --- bcache ---
>>>> Device /dev/md0 (9:0)
>>>> UUID 8275bf01-f0b3-423e-87fa-48336ce33068
>>>> Block Size 0.50KiB
>>>> Bucket Size 512.00KiB
>>>> Congested? False
>>>> Read Congestion 0.0ms
>>>> Write Congestion 0.0ms
>>>> Total Cache Size 111.79GiB
>>>> Total Cache Used 2.24GiB (2%)
>>>> Total Cache Unused 109.55GiB (98%)
>>>> Evictable Cache 110.67GiB (99%)
>>>> Replacement Policy [lru] fifo random
>>>> Cache Mode writethrough [writeback] writearound none
>>>> Total Hits 5 (0%)
>>>> Total Misses 4079
>>>> Total Bypass Hits 0 (0%)
>>>> Total Bypass Misses 615
>>>> Total Bypassed 2.40MiB
>>>>
>>>>
>>>>
>>>> Linux kkb 4.19.0-041900-generic #201810221809 SMP Mon Oct 22 22:11:45
>>>> UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
>>>>
>>>> --- bcache ---
>>>> Device /dev/md0 (9:0)
>>>> UUID 67269654-92e8-4c3b-a524-8e8910082146
>>>> Block Size 0.50KiB
>>>> Bucket Size 512.00KiB
>>>> Congested? False
>>>> Read Congestion 0.0ms
>>>> Write Congestion 0.0ms
>>>> Total Cache Size 111.79GiB
>>>> Total Cache Used 1.12GiB (0%)
>>>> Total Cache Unused 110.67GiB (99%)
>>>> Evictable Cache 111.79GiB (100%)
>>>> Replacement Policy [lru] fifo random
>>>> Cache Mode writethrough [writeback] writearound none
>>>> Total Hits 0 (0%)
>>>> Total Misses 1
>>>> Total Bypass Hits 0
>>>> Total Bypass Misses 0
>>>> Total Bypassed 0B
>>>>
>>>> 1st pass
>>>> 0.00user 0.33system 0:09.29elapsed 3%CPU (0avgtext+0avgdata
>>>> 2280maxresident)k
>>>> 2097624inputs+0outputs (1major+113minor)pagefaults 0swaps
>>>>
>>>> 2nd pass
>>>> 0.00user 0.33system 0:08.47elapsed 4%CPU (0avgtext+0avgdata
>>>> 2248maxresident)k
>>>> 2097280inputs+0outputs (1major+111minor)pagefaults 0swaps
>>>>
>>>> 3rd pass
>>>> 0.00user 0.37system 0:10.46elapsed 3%CPU (0avgtext+0avgdata
>>>> 2220maxresident)k
>>>> 2097616inputs+0outputs (1major+114minor)pagefaults 0swaps
>>>> --- bcache ---
>>>> Device /dev/md0 (9:0)
>>>> UUID 67269654-92e8-4c3b-a524-8e8910082146
>>>> Block Size 0.50KiB
>>>> Bucket Size 512.00KiB
>>>> Congested? False
>>>> Read Congestion 0.0ms
>>>> Write Congestion 0.0ms
>>>> Total Cache Size 111.79GiB
>>>> Total Cache Used 1.12GiB (0%)
>>>> Total Cache Unused 110.67GiB (99%)
>>>> Evictable Cache 111.79GiB (100%)
>>>> Replacement Policy [lru] fifo random
>>>> Cache Mode writethrough [writeback] writearound none
>>>> Total Hits 132 (23%)
>>>> Total Misses 436
>>>> Total Bypass Hits 51 (0%)
>>>> Total Bypass Misses 17399
>>>> Total Bypassed 43.50MiB
>>>>
>>>>
>>>>
>>>>
>>>> Linux kkb 5.5.2-050502-generic #202002041931 SMP Tue Feb 4 19:33:15 UTC
>>>> 2020 x86_64 x86_64 x86_64 GNU/Linux
>>>>
>>>> --- bcache ---
>>>> Device /dev/md0 (9:0)
>>>> UUID 38a8b675-e332-4076-b0cf-44e4be72c300
>>>> Block Size 0.50KiB
>>>> Bucket Size 512.00KiB
>>>> Congested? False
>>>> Read Congestion 0.0ms
>>>> Write Congestion 0.0ms
>>>> Total Cache Size 111.79GiB
>>>> Total Cache Used 1.12GiB (0%)
>>>> Total Cache Unused 110.67GiB (99%)
>>>> Evictable Cache 111.79GiB (100%)
>>>> Replacement Policy [lru] fifo random
>>>> Cache Mode writethrough [writeback] writearound none
>>>> Total Hits 0 (0%)
>>>> Total Misses 1
>>>> Total Bypass Hits 0 (0%)
>>>> Total Bypass Misses 3
>>>> Total Bypassed 52.00KiB
>>>>
>>>> 1st pass
>>>> 0.00user 0.42system 0:09.21elapsed 4%CPU (0avgtext+0avgdata
>>>> 2216maxresident)k
>>>> 2097608inputs+0outputs (1major+112minor)pagefaults 0swaps
>>>>
>>>> 2nd pass
>>>> 0.00user 0.42system 0:09.62elapsed 4%CPU (0avgtext+0avgdata
>>>> 2248maxresident)k
>>>> 2097280inputs+0outputs (1major+112minor)pagefaults 0swaps
>>>>
>>>> 3rd pass
>>>> 0.00user 0.43system 0:08.75elapsed 5%CPU (0avgtext+0avgdata
>>>> 2220maxresident)k
>>>> 2097224inputs+0outputs (1major+114minor)pagefaults 0swaps
>>>> --- bcache ---
>>>> Device /dev/md0 (9:0)
>>>> UUID 38a8b675-e332-4076-b0cf-44e4be72c300
>>>> Block Size 0.50KiB
>>>> Bucket Size 512.00KiB
>>>> Congested? False
>>>> Read Congestion 0.0ms
>>>> Write Congestion 0.0ms
>>>> Total Cache Size 111.79GiB
>>>> Total Cache Used 1.12GiB (0%)
>>>> Total Cache Unused 110.67GiB (99%)
>>>> Evictable Cache 111.79GiB (100%)
>>>> Replacement Policy [lru] fifo random
>>>> Cache Mode writethrough [writeback] writearound none
>>>> Total Hits 121 (32%)
>>>> Total Misses 246
>>>> Total Bypass Hits 15 (0%)
>>>> Total Bypass Misses 12811
>>>> Total Bypassed 39.70MiB
>>>
>>>
>>>
>>> --
>>> --
>>> Ville Aakko - ville.aakko@gmail.com
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: reads no longer cached since kernel 4.19
2020-02-12 10:03 ` Postgarage Graz IT
@ 2020-02-12 20:32 ` Ville Aakko
2020-02-13 11:00 ` Matthias Ferdinand
0 siblings, 1 reply; 7+ messages in thread
From: Ville Aakko @ 2020-02-12 20:32 UTC (permalink / raw)
To: linux-bcache
Hi,
(Resending a second time as previous email was, again, sent as html -
Android gmail client sucks!). My reply inline:
ke 12. helmik. 2020 klo 12.05 Postgarage Graz IT (it@postgarage.at) kirjoitti:
>
> On 12.02.20 10:33, Matthias Ferdinand wrote:
> > Sorry for being late to reply. Your bcache is set to "writeback", and
> > your test is doing reads from a bcache with a still empty cache device;
> > while the "1GB" file resides initially only on the backing device.
> >
> > My understanding so far was that in writeback mode only writes will ever
> > add data to the caching device (first as dirty, and post-writeback as
> > clean for as long as as space restrictions / usage patterns allow).
> > Therefore I would not have expected any piece of your "1GB" file to ever
> > be added to your caching device, and reading it would have to resort to
> > the backing device, even when repeatedly reading it.
>
> I would have said, that in writeback mode bcache acts as a read/write
> cache caching both reads and writes.
> Furthermore I also did the benchmark with all three caching modes and
> all behaved the same.
>
> As you can see, with kernel 5.6 I've got back the behavior like with
> kernels >= 4.18, the read is cached in writeback mode.
>
> But perhaps I'm wrong and this shouldn't happen?
As per previous discussions (links in the end; (also linked in my
reply which I accidentally sent only to Flo (Postgarage Graz IT):
Should not all of the 1gb be cached if sequentual_cutoff is set to 0?
Previously this used to be the case (I've used to set it to 0 since I
do not believe my ssd will ever wear out in a sensible timespan on
desktop usage), and my SSD it is a lot faster than my mechanical HDDs
even for larger, sequential data reads.
Also, on my setup there is no performance gain whatsoever currently in
boot times or loading times of any applications, no matter what
sequential_cutoff or cache_mode is set to. Previously, this used not
to be the case; e.g. with bcache boot and app loading times (I've used
libreoffice and firefox starts as benchmarks) were on par with ssd
performance.
Indeed, the behavior is same no matter what is the cache mode.
Also, some reports in Arch forums thread (despite it being an Arch
forum, there was at least one user who has the same issue on Ubuntu).
Regards,
Ville
*) https://www.spinics.net/lists/linux-bcache/msg07859.html
**) https://bbs.archlinux.org/viewtopic.php?id=250525
>
> >
> > I have seen discussion about caching of readahead requests for newer
> > kernels, and I thought (but never really checked) this would only
> > concern read-caching modes. Am I wrong in this assumption?
> >
> > Matthias
> >
> >
> > On Wed, Feb 12, 2020 at 07:02:28AM +0100, Postgarage Graz IT wrote:
> >> On 10.02.20 17:10, Ville Aakko wrote:
> >>> Hi,
> >>>
> >>> A fellow user responding here.
> >>>
> >>> I've noticed similar behavior and have asked on this same mailing list
> >>> previously. See:
> >>> https://www.spinics.net/lists/linux-bcache/msg07859.html
> >>>
> >>> Also seems there are other users with this issue on the Arch Forum,
> >>> where I have also started a discussion:
> >>> https://bbs.archlinux.org/viewtopic.php?id=250525
> >>> There is yet to be a single user to reply there (or on this mailing
> >>> list) claiming they have a working setup (for caching reads).
> >>>
> >>> Judging from the Arch Linux thread, I have a hunch there were some
> >>> changes ~4.18, which broke read caching for many (all?) desktop users
> >>> (as anything which is flagged as readahed will not be cached, despite
> >>> setting sequential_cutoff). Also (again from the Arch thread) a
> >>> planned patch might enable expected read caching: "[PATCH 3/5] bcache:
> >>> add readahead cache policy options via sysfs interface" / see:
> >>> https://www.spinics.net/lists/linux-bcache/msg08074.html
> >>
> >> Indeed that patch works.
> >> Now I'm using the 5.6-rc1 kernel and the performance gain is huge.
> >>
> >> I only wonder, why the used cache number doesn't go up anymore like it
> >> did for pre-4.19 kernels.
> >>
> >>
> >> Linux kkb 5.6.0-050600rc1-generic #202002092032 SMP Mon Feb 10 01:36:50
> >> UTC 2020 x86_64 x86_64 x86_64 GNU/Linux
> >>
> >> --- bcache ---
> >> Device /dev/md0 (9:0)
> >> UUID d4f0e4cd-c2dc-4cec-bf5b-96f1f87ff0b8
> >> Block Size 0.50KiB
> >> Bucket Size 512.00KiB
> >> Congested? False
> >> Read Congestion 0.0ms
> >> Write Congestion 0.0ms
> >> Total Cache Size 111.79GiB
> >> Total Cache Used 1.12GiB (0%)
> >> Total Cache Unused 110.67GiB (99%)
> >> Evictable Cache 111.79GiB (100%)
> >> Replacement Policy [lru] fifo random
> >> Cache Mode writethrough [writeback] writearound none
> >> Total Hits 0
> >> Total Misses 0
> >> Total Bypass Hits 0
> >> Total Bypass Misses 0
> >> Total Bypassed 0B
> >>
> >> 1st pass
> >> 0.00user 0.60system 0:08.66elapsed 6%CPU (0avgtext+0avgdata
> >> 2220maxresident)k
> >> 2097632inputs+0outputs (1major+115minor)pagefaults 0swaps
> >>
> >> 2nd pass
> >> 0.00user 0.47system 0:03.34elapsed 14%CPU (0avgtext+0avgdata
> >> 2216maxresident)k
> >> 2097128inputs+0outputs (1major+113minor)pagefaults 0swaps
> >>
> >> 3rd pass
> >> 0.00user 0.45system 0:02.58elapsed 17%CPU (0avgtext+0avgdata
> >> 2096maxresident)k
> >> 2097296inputs+0outputs (1major+110minor)pagefaults 0swaps
> >> --- bcache ---
> >> Device /dev/md0 (9:0)
> >> UUID d4f0e4cd-c2dc-4cec-bf5b-96f1f87ff0b8
> >> Block Size 0.50KiB
> >> Bucket Size 512.00KiB
> >> Congested? False
> >> Read Congestion 0.0ms
> >> Write Congestion 0.0ms
> >> Total Cache Size 111.79GiB
> >> Total Cache Used 1.12GiB (0%)
> >> Total Cache Unused 110.67GiB (99%)
> >> Evictable Cache 110.67GiB (99%)
> >> Replacement Policy [lru] fifo random
> >> Cache Mode writethrough [writeback] writearound none
> >> Total Hits 6352 (51%)
> >> Total Misses 6075
> >> Total Bypass Hits 0
> >> Total Bypass Misses 0
> >>
> >>
> >> As you can see, the reads must come from the SSD in the 2nd and 3rd
> >> pass, still "Total Cache Used" stays the same.
> >>
> >>
> >>>
> >>> However this is highly speculative from someone not understanding file
> >>> systems or insides of bcache or the code at all.
> >>>
> >>> Perhaps someone more involved can reply: is the current behavior
> >>> expected (reads are not getting cached practically at all). Also, is
> >>> the patch I've linked possibly going to fix the current issues?
> >>>
> >>> Kind Regards,
> >>> Ville Aakko
> >>>
> >>>
> >>>
> >>> su 9. helmik. 2020 klo 23.37 Postgarage Graz IT (it@postgarage.at) kirjoitti:
> >>>>
> >>>> Hello!
> >>>>
> >>>> I noticed, that bcache is no longer caching reads on my system which
> >>>> makes it behave like if there were only hdds.
> >>>>
> >>>> I'm using two hdds in a raid 1 as the backing device and a single ssd as
> >>>> cache device:
> >>>>
> >>>> sda 8:0 0 111,8G 0 disk
> >>>> └─bcache0 252:0 0 921,9G 0 disk /
> >>>> sdb 8:16 0 931,5G 0 disk
> >>>> ├─sdb1 8:17 0 922G 0 part
> >>>> │ └─md0 9:0 0 921,9G 0 raid1
> >>>> │ └─bcache0 252:0 0 921,9G 0 disk /
> >>>> ├─sdb2 8:18 0 1K 0 part
> >>>> ├─sdb5 8:21 0 1,9G 0 part
> >>>> │ └─md1 9:1 0 1,9G 0 raid1 /boot
> >>>> └─sdb6 8:22 0 7,6G 0 part [SWAP]
> >>>> sdc 8:32 0 931,5G 0 disk
> >>>> ├─sdc1 8:33 0 922G 0 part
> >>>> │ └─md0 9:0 0 921,9G 0 raid1
> >>>> │ └─bcache0 252:0 0 921,9G 0 disk /
> >>>> ├─sdc2 8:34 0 1K 0 part
> >>>> ├─sdc5 8:37 0 1,9G 0 part
> >>>> │ └─md1 9:1 0 1,9G 0 raid1 /boot
> >>>> └─sdc6 8:38 0 7,6G 0 part [SWAP]
> >>>>
> >>>>
> >>>> For benchmarking every time I detach the cache device, stop the bcache
> >>>> device, do a wipefs on the cache device, then make-bcache -C /dev/sda
> >>>> and finally reattach the cache.
> >>>> After that, I'm using the following script to repeatedly read a 1gb file:
> >>>>
> >>>> #!/bin/sh
> >>>> echo 0 > /sys/block/bcache0/bcache/sequential_cutoff
> >>>> echo 0 > /sys/block/bcache0/bcache/cache/congested_read_threshold_us
> >>>> echo 0 > /sys/block/bcache0/bcache/cache/congested_write_threshold_us
> >>>> uname -a
> >>>> echo
> >>>> bcache-status
> >>>> echo
> >>>> echo "1st pass"
> >>>> sync; echo 3 > /proc/sys/vm/drop_caches
> >>>> (time cat 1GB.bin > /dev/null)
> >>>> echo
> >>>> echo "2nd pass"
> >>>> sync; echo 3 > /proc/sys/vm/drop_caches
> >>>> (time cat 1GB.bin > /dev/null)
> >>>> echo
> >>>> echo "3rd pass"
> >>>> sync; echo 3 > /proc/sys/vm/drop_caches
> >>>> (time cat 1GB.bin > /dev/null)
> >>>> bcache-status
> >>>>
> >>>>
> >>>>
> >>>> As you can see from the results below, kernel 4.18.20 is the last
> >>>> kernel, where the cache grows and the performance goes up.
> >>>>
> >>>> I also compiled 4.19.0 with the bcache files from 4.18.20 and much to my
> >>>> suprise, that didn't change 4.19's behavior - still no caching. So some
> >>>> other changes must be the culprit or I did something wrong.
> >>>> I'm not that much into compiling the kernel, but I checked out the
> >>>> 4.19.0 and 4.18.20 commits and replaced the 4.19.0 drivers/md/bcache
> >>>> directory with the one from 4.18.20 - then recompiled and installed the
> >>>> new kernel.
> >>>>
> >>>> So i am at my wits end. Any help would be appreciated.
> >>>> Thanks
> >>>> Flo
> >>>>
> >>>>
> >>>> Linux kkb 4.18.20-041820-generic #201812030624 SMP Mon Dec 3 11:25:55
> >>>> UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
> >>>>
> >>>> --- bcache ---
> >>>> Device /dev/md0 (9:0)
> >>>> UUID 8275bf01-f0b3-423e-87fa-48336ce33068
> >>>> Block Size 0.50KiB
> >>>> Bucket Size 512.00KiB
> >>>> Congested? False
> >>>> Read Congestion 0.0ms
> >>>> Write Congestion 0.0ms
> >>>> Total Cache Size 111.79GiB
> >>>> Total Cache Used 1.12GiB (0%)
> >>>> Total Cache Unused 110.67GiB (99%)
> >>>> Evictable Cache 111.79GiB (100%)
> >>>> Replacement Policy [lru] fifo random
> >>>> Cache Mode writethrough [writeback] writearound none
> >>>> Total Hits 0 (0%)
> >>>> Total Misses 6
> >>>> Total Bypass Hits 0
> >>>> Total Bypass Misses 0
> >>>> Total Bypassed 0B
> >>>>
> >>>> 1st pass
> >>>> 0.00user 0.36system 0:08.58elapsed 4%CPU (0avgtext+0avgdata
> >>>> 2196maxresident)k
> >>>> 2097608inputs+0outputs (1major+113minor)pagefaults 0swaps
> >>>>
> >>>> 2nd pass
> >>>> 0.00user 0.32system 0:03.29elapsed 9%CPU (0avgtext+0avgdata
> >>>> 2100maxresident)k
> >>>> 2097184inputs+0outputs (1major+110minor)pagefaults 0swaps
> >>>>
> >>>> 3rd pass
> >>>> 0.00user 0.32system 0:02.64elapsed 12%CPU (0avgtext+0avgdata
> >>>> 2092maxresident)k
> >>>> 2097280inputs+0outputs (1major+111minor)pagefaults 0swaps
> >>>> --- bcache ---
> >>>> Device /dev/md0 (9:0)
> >>>> UUID 8275bf01-f0b3-423e-87fa-48336ce33068
> >>>> Block Size 0.50KiB
> >>>> Bucket Size 512.00KiB
> >>>> Congested? False
> >>>> Read Congestion 0.0ms
> >>>> Write Congestion 0.0ms
> >>>> Total Cache Size 111.79GiB
> >>>> Total Cache Used 2.24GiB (2%)
> >>>> Total Cache Unused 109.55GiB (98%)
> >>>> Evictable Cache 110.67GiB (99%)
> >>>> Replacement Policy [lru] fifo random
> >>>> Cache Mode writethrough [writeback] writearound none
> >>>> Total Hits 5 (0%)
> >>>> Total Misses 4079
> >>>> Total Bypass Hits 0 (0%)
> >>>> Total Bypass Misses 615
> >>>> Total Bypassed 2.40MiB
> >>>>
> >>>>
> >>>>
> >>>> Linux kkb 4.19.0-041900-generic #201810221809 SMP Mon Oct 22 22:11:45
> >>>> UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
> >>>>
> >>>> --- bcache ---
> >>>> Device /dev/md0 (9:0)
> >>>> UUID 67269654-92e8-4c3b-a524-8e8910082146
> >>>> Block Size 0.50KiB
> >>>> Bucket Size 512.00KiB
> >>>> Congested? False
> >>>> Read Congestion 0.0ms
> >>>> Write Congestion 0.0ms
> >>>> Total Cache Size 111.79GiB
> >>>> Total Cache Used 1.12GiB (0%)
> >>>> Total Cache Unused 110.67GiB (99%)
> >>>> Evictable Cache 111.79GiB (100%)
> >>>> Replacement Policy [lru] fifo random
> >>>> Cache Mode writethrough [writeback] writearound none
> >>>> Total Hits 0 (0%)
> >>>> Total Misses 1
> >>>> Total Bypass Hits 0
> >>>> Total Bypass Misses 0
> >>>> Total Bypassed 0B
> >>>>
> >>>> 1st pass
> >>>> 0.00user 0.33system 0:09.29elapsed 3%CPU (0avgtext+0avgdata
> >>>> 2280maxresident)k
> >>>> 2097624inputs+0outputs (1major+113minor)pagefaults 0swaps
> >>>>
> >>>> 2nd pass
> >>>> 0.00user 0.33system 0:08.47elapsed 4%CPU (0avgtext+0avgdata
> >>>> 2248maxresident)k
> >>>> 2097280inputs+0outputs (1major+111minor)pagefaults 0swaps
> >>>>
> >>>> 3rd pass
> >>>> 0.00user 0.37system 0:10.46elapsed 3%CPU (0avgtext+0avgdata
> >>>> 2220maxresident)k
> >>>> 2097616inputs+0outputs (1major+114minor)pagefaults 0swaps
> >>>> --- bcache ---
> >>>> Device /dev/md0 (9:0)
> >>>> UUID 67269654-92e8-4c3b-a524-8e8910082146
> >>>> Block Size 0.50KiB
> >>>> Bucket Size 512.00KiB
> >>>> Congested? False
> >>>> Read Congestion 0.0ms
> >>>> Write Congestion 0.0ms
> >>>> Total Cache Size 111.79GiB
> >>>> Total Cache Used 1.12GiB (0%)
> >>>> Total Cache Unused 110.67GiB (99%)
> >>>> Evictable Cache 111.79GiB (100%)
> >>>> Replacement Policy [lru] fifo random
> >>>> Cache Mode writethrough [writeback] writearound none
> >>>> Total Hits 132 (23%)
> >>>> Total Misses 436
> >>>> Total Bypass Hits 51 (0%)
> >>>> Total Bypass Misses 17399
> >>>> Total Bypassed 43.50MiB
> >>>>
> >>>>
> >>>>
> >>>>
> >>>> Linux kkb 5.5.2-050502-generic #202002041931 SMP Tue Feb 4 19:33:15 UTC
> >>>> 2020 x86_64 x86_64 x86_64 GNU/Linux
> >>>>
> >>>> --- bcache ---
> >>>> Device /dev/md0 (9:0)
> >>>> UUID 38a8b675-e332-4076-b0cf-44e4be72c300
> >>>> Block Size 0.50KiB
> >>>> Bucket Size 512.00KiB
> >>>> Congested? False
> >>>> Read Congestion 0.0ms
> >>>> Write Congestion 0.0ms
> >>>> Total Cache Size 111.79GiB
> >>>> Total Cache Used 1.12GiB (0%)
> >>>> Total Cache Unused 110.67GiB (99%)
> >>>> Evictable Cache 111.79GiB (100%)
> >>>> Replacement Policy [lru] fifo random
> >>>> Cache Mode writethrough [writeback] writearound none
> >>>> Total Hits 0 (0%)
> >>>> Total Misses 1
> >>>> Total Bypass Hits 0 (0%)
> >>>> Total Bypass Misses 3
> >>>> Total Bypassed 52.00KiB
> >>>>
> >>>> 1st pass
> >>>> 0.00user 0.42system 0:09.21elapsed 4%CPU (0avgtext+0avgdata
> >>>> 2216maxresident)k
> >>>> 2097608inputs+0outputs (1major+112minor)pagefaults 0swaps
> >>>>
> >>>> 2nd pass
> >>>> 0.00user 0.42system 0:09.62elapsed 4%CPU (0avgtext+0avgdata
> >>>> 2248maxresident)k
> >>>> 2097280inputs+0outputs (1major+112minor)pagefaults 0swaps
> >>>>
> >>>> 3rd pass
> >>>> 0.00user 0.43system 0:08.75elapsed 5%CPU (0avgtext+0avgdata
> >>>> 2220maxresident)k
> >>>> 2097224inputs+0outputs (1major+114minor)pagefaults 0swaps
> >>>> --- bcache ---
> >>>> Device /dev/md0 (9:0)
> >>>> UUID 38a8b675-e332-4076-b0cf-44e4be72c300
> >>>> Block Size 0.50KiB
> >>>> Bucket Size 512.00KiB
> >>>> Congested? False
> >>>> Read Congestion 0.0ms
> >>>> Write Congestion 0.0ms
> >>>> Total Cache Size 111.79GiB
> >>>> Total Cache Used 1.12GiB (0%)
> >>>> Total Cache Unused 110.67GiB (99%)
> >>>> Evictable Cache 111.79GiB (100%)
> >>>> Replacement Policy [lru] fifo random
> >>>> Cache Mode writethrough [writeback] writearound none
> >>>> Total Hits 121 (32%)
> >>>> Total Misses 246
> >>>> Total Bypass Hits 15 (0%)
> >>>> Total Bypass Misses 12811
> >>>> Total Bypassed 39.70MiB
> >>>
> >>>
> >>>
> >>> --
> >>> --
> >>> Ville Aakko - ville.aakko@gmail.com
--
--
Ville Aakko - ville.aakko@gmail.com
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: reads no longer cached since kernel 4.19
2020-02-12 20:32 ` Ville Aakko
@ 2020-02-13 11:00 ` Matthias Ferdinand
0 siblings, 0 replies; 7+ messages in thread
From: Matthias Ferdinand @ 2020-02-13 11:00 UTC (permalink / raw)
To: linux-bcache
[resending to list]
On Wed, Feb 12, 2020 at 10:32:37PM +0200, Ville Aakko wrote:
> Should not all of the 1gb be cached if sequentual_cutoff is set to 0?
Sorry my test setup is currently busy with other stuff so I can't test
myself at the moment.
sequential_cutoff=0 should lead to everything going to the cache device
(unless congested). Still I am not sure bcache should put reads into the
cache when in writeback mode. And as has been said, it looked like the
cache device was not used at all, so let's try with some writes first:
Can you try this sequence:
a) detach / recreate / reattach cache device
b) write the "1GB" file to the bcache device
c) drop vm caches
d) now read the "1GB" file from the bcache device
After b) you should see some cache device usage and perhaps even some
dirty data (newer bcache wants to flush dirty data as fast as possible
if the backing device goes idle, so by the time you look at it, it might
not be dirty anymore).
If the reads in d) do not read from the cache device, or the cache
device still shows no sign of being filled with data, that would be a
real bug.
Matthias
>
> Previously this used to be the case (I've used to set it to 0 since I
> do not believe my ssd will ever wear out in a sensible timespan on
> desktop usage), and my SSD it is a lot faster than my mechanical HDDs
> even for larger, sequential data reads.
>
> Also, on my setup there is no performance gain whatsoever currently in
> boot times or loading times of any applications, no matter what
> sequential_cutoff or cache_mode is set to. Previously, this used not
> to be the case; e.g. with bcache boot and app loading times (I've used
> libreoffice and firefox starts as benchmarks) were on par with ssd
> performance.
>
> Indeed, the behavior is same no matter what is the cache mode.
>
> Also, some reports in Arch forums thread (despite it being an Arch
> forum, there was at least one user who has the same issue on Ubuntu).
>
> Regards,
>
> Ville
> *) https://www.spinics.net/lists/linux-bcache/msg07859.html
> **) https://bbs.archlinux.org/viewtopic.php?id=250525
>
> >
> > >
> > > I have seen discussion about caching of readahead requests for newer
> > > kernels, and I thought (but never really checked) this would only
> > > concern read-caching modes. Am I wrong in this assumption?
> > >
> > > Matthias
> > >
> > >
> > > On Wed, Feb 12, 2020 at 07:02:28AM +0100, Postgarage Graz IT wrote:
> > >> On 10.02.20 17:10, Ville Aakko wrote:
> > >>> Hi,
> > >>>
> > >>> A fellow user responding here.
> > >>>
> > >>> I've noticed similar behavior and have asked on this same mailing list
> > >>> previously. See:
> > >>> https://www.spinics.net/lists/linux-bcache/msg07859.html
> > >>>
> > >>> Also seems there are other users with this issue on the Arch Forum,
> > >>> where I have also started a discussion:
> > >>> https://bbs.archlinux.org/viewtopic.php?id=250525
> > >>> There is yet to be a single user to reply there (or on this mailing
> > >>> list) claiming they have a working setup (for caching reads).
> > >>>
> > >>> Judging from the Arch Linux thread, I have a hunch there were some
> > >>> changes ~4.18, which broke read caching for many (all?) desktop users
> > >>> (as anything which is flagged as readahed will not be cached, despite
> > >>> setting sequential_cutoff). Also (again from the Arch thread) a
> > >>> planned patch might enable expected read caching: "[PATCH 3/5] bcache:
> > >>> add readahead cache policy options via sysfs interface" / see:
> > >>> https://www.spinics.net/lists/linux-bcache/msg08074.html
> > >>
> > >> Indeed that patch works.
> > >> Now I'm using the 5.6-rc1 kernel and the performance gain is huge.
> > >>
> > >> I only wonder, why the used cache number doesn't go up anymore like it
> > >> did for pre-4.19 kernels.
> > >>
> > >>
> > >> Linux kkb 5.6.0-050600rc1-generic #202002092032 SMP Mon Feb 10 01:36:50
> > >> UTC 2020 x86_64 x86_64 x86_64 GNU/Linux
> > >>
> > >> --- bcache ---
> > >> Device /dev/md0 (9:0)
> > >> UUID d4f0e4cd-c2dc-4cec-bf5b-96f1f87ff0b8
> > >> Block Size 0.50KiB
> > >> Bucket Size 512.00KiB
> > >> Congested? False
> > >> Read Congestion 0.0ms
> > >> Write Congestion 0.0ms
> > >> Total Cache Size 111.79GiB
> > >> Total Cache Used 1.12GiB (0%)
> > >> Total Cache Unused 110.67GiB (99%)
> > >> Evictable Cache 111.79GiB (100%)
> > >> Replacement Policy [lru] fifo random
> > >> Cache Mode writethrough [writeback] writearound none
> > >> Total Hits 0
> > >> Total Misses 0
> > >> Total Bypass Hits 0
> > >> Total Bypass Misses 0
> > >> Total Bypassed 0B
> > >>
> > >> 1st pass
> > >> 0.00user 0.60system 0:08.66elapsed 6%CPU (0avgtext+0avgdata
> > >> 2220maxresident)k
> > >> 2097632inputs+0outputs (1major+115minor)pagefaults 0swaps
> > >>
> > >> 2nd pass
> > >> 0.00user 0.47system 0:03.34elapsed 14%CPU (0avgtext+0avgdata
> > >> 2216maxresident)k
> > >> 2097128inputs+0outputs (1major+113minor)pagefaults 0swaps
> > >>
> > >> 3rd pass
> > >> 0.00user 0.45system 0:02.58elapsed 17%CPU (0avgtext+0avgdata
> > >> 2096maxresident)k
> > >> 2097296inputs+0outputs (1major+110minor)pagefaults 0swaps
> > >> --- bcache ---
> > >> Device /dev/md0 (9:0)
> > >> UUID d4f0e4cd-c2dc-4cec-bf5b-96f1f87ff0b8
> > >> Block Size 0.50KiB
> > >> Bucket Size 512.00KiB
> > >> Congested? False
> > >> Read Congestion 0.0ms
> > >> Write Congestion 0.0ms
> > >> Total Cache Size 111.79GiB
> > >> Total Cache Used 1.12GiB (0%)
> > >> Total Cache Unused 110.67GiB (99%)
> > >> Evictable Cache 110.67GiB (99%)
> > >> Replacement Policy [lru] fifo random
> > >> Cache Mode writethrough [writeback] writearound none
> > >> Total Hits 6352 (51%)
> > >> Total Misses 6075
> > >> Total Bypass Hits 0
> > >> Total Bypass Misses 0
> > >>
> > >>
> > >> As you can see, the reads must come from the SSD in the 2nd and 3rd
> > >> pass, still "Total Cache Used" stays the same.
> > >>
> > >>
> > >>>
> > >>> However this is highly speculative from someone not understanding file
> > >>> systems or insides of bcache or the code at all.
> > >>>
> > >>> Perhaps someone more involved can reply: is the current behavior
> > >>> expected (reads are not getting cached practically at all). Also, is
> > >>> the patch I've linked possibly going to fix the current issues?
> > >>>
> > >>> Kind Regards,
> > >>> Ville Aakko
> > >>>
> > >>>
> > >>>
> > >>> su 9. helmik. 2020 klo 23.37 Postgarage Graz IT (it@postgarage.at) kirjoitti:
> > >>>>
> > >>>> Hello!
> > >>>>
> > >>>> I noticed, that bcache is no longer caching reads on my system which
> > >>>> makes it behave like if there were only hdds.
> > >>>>
> > >>>> I'm using two hdds in a raid 1 as the backing device and a single ssd as
> > >>>> cache device:
> > >>>>
> > >>>> sda 8:0 0 111,8G 0 disk
> > >>>> └─bcache0 252:0 0 921,9G 0 disk /
> > >>>> sdb 8:16 0 931,5G 0 disk
> > >>>> ├─sdb1 8:17 0 922G 0 part
> > >>>> │ └─md0 9:0 0 921,9G 0 raid1
> > >>>> │ └─bcache0 252:0 0 921,9G 0 disk /
> > >>>> ├─sdb2 8:18 0 1K 0 part
> > >>>> ├─sdb5 8:21 0 1,9G 0 part
> > >>>> │ └─md1 9:1 0 1,9G 0 raid1 /boot
> > >>>> └─sdb6 8:22 0 7,6G 0 part [SWAP]
> > >>>> sdc 8:32 0 931,5G 0 disk
> > >>>> ├─sdc1 8:33 0 922G 0 part
> > >>>> │ └─md0 9:0 0 921,9G 0 raid1
> > >>>> │ └─bcache0 252:0 0 921,9G 0 disk /
> > >>>> ├─sdc2 8:34 0 1K 0 part
> > >>>> ├─sdc5 8:37 0 1,9G 0 part
> > >>>> │ └─md1 9:1 0 1,9G 0 raid1 /boot
> > >>>> └─sdc6 8:38 0 7,6G 0 part [SWAP]
> > >>>>
> > >>>>
> > >>>> For benchmarking every time I detach the cache device, stop the bcache
> > >>>> device, do a wipefs on the cache device, then make-bcache -C /dev/sda
> > >>>> and finally reattach the cache.
> > >>>> After that, I'm using the following script to repeatedly read a 1gb file:
> > >>>>
> > >>>> #!/bin/sh
> > >>>> echo 0 > /sys/block/bcache0/bcache/sequential_cutoff
> > >>>> echo 0 > /sys/block/bcache0/bcache/cache/congested_read_threshold_us
> > >>>> echo 0 > /sys/block/bcache0/bcache/cache/congested_write_threshold_us
> > >>>> uname -a
> > >>>> echo
> > >>>> bcache-status
> > >>>> echo
> > >>>> echo "1st pass"
> > >>>> sync; echo 3 > /proc/sys/vm/drop_caches
> > >>>> (time cat 1GB.bin > /dev/null)
> > >>>> echo
> > >>>> echo "2nd pass"
> > >>>> sync; echo 3 > /proc/sys/vm/drop_caches
> > >>>> (time cat 1GB.bin > /dev/null)
> > >>>> echo
> > >>>> echo "3rd pass"
> > >>>> sync; echo 3 > /proc/sys/vm/drop_caches
> > >>>> (time cat 1GB.bin > /dev/null)
> > >>>> bcache-status
> > >>>>
> > >>>>
> > >>>>
> > >>>> As you can see from the results below, kernel 4.18.20 is the last
> > >>>> kernel, where the cache grows and the performance goes up.
> > >>>>
> > >>>> I also compiled 4.19.0 with the bcache files from 4.18.20 and much to my
> > >>>> suprise, that didn't change 4.19's behavior - still no caching. So some
> > >>>> other changes must be the culprit or I did something wrong.
> > >>>> I'm not that much into compiling the kernel, but I checked out the
> > >>>> 4.19.0 and 4.18.20 commits and replaced the 4.19.0 drivers/md/bcache
> > >>>> directory with the one from 4.18.20 - then recompiled and installed the
> > >>>> new kernel.
> > >>>>
> > >>>> So i am at my wits end. Any help would be appreciated.
> > >>>> Thanks
> > >>>> Flo
> > >>>>
> > >>>>
> > >>>> Linux kkb 4.18.20-041820-generic #201812030624 SMP Mon Dec 3 11:25:55
> > >>>> UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
> > >>>>
> > >>>> --- bcache ---
> > >>>> Device /dev/md0 (9:0)
> > >>>> UUID 8275bf01-f0b3-423e-87fa-48336ce33068
> > >>>> Block Size 0.50KiB
> > >>>> Bucket Size 512.00KiB
> > >>>> Congested? False
> > >>>> Read Congestion 0.0ms
> > >>>> Write Congestion 0.0ms
> > >>>> Total Cache Size 111.79GiB
> > >>>> Total Cache Used 1.12GiB (0%)
> > >>>> Total Cache Unused 110.67GiB (99%)
> > >>>> Evictable Cache 111.79GiB (100%)
> > >>>> Replacement Policy [lru] fifo random
> > >>>> Cache Mode writethrough [writeback] writearound none
> > >>>> Total Hits 0 (0%)
> > >>>> Total Misses 6
> > >>>> Total Bypass Hits 0
> > >>>> Total Bypass Misses 0
> > >>>> Total Bypassed 0B
> > >>>>
> > >>>> 1st pass
> > >>>> 0.00user 0.36system 0:08.58elapsed 4%CPU (0avgtext+0avgdata
> > >>>> 2196maxresident)k
> > >>>> 2097608inputs+0outputs (1major+113minor)pagefaults 0swaps
> > >>>>
> > >>>> 2nd pass
> > >>>> 0.00user 0.32system 0:03.29elapsed 9%CPU (0avgtext+0avgdata
> > >>>> 2100maxresident)k
> > >>>> 2097184inputs+0outputs (1major+110minor)pagefaults 0swaps
> > >>>>
> > >>>> 3rd pass
> > >>>> 0.00user 0.32system 0:02.64elapsed 12%CPU (0avgtext+0avgdata
> > >>>> 2092maxresident)k
> > >>>> 2097280inputs+0outputs (1major+111minor)pagefaults 0swaps
> > >>>> --- bcache ---
> > >>>> Device /dev/md0 (9:0)
> > >>>> UUID 8275bf01-f0b3-423e-87fa-48336ce33068
> > >>>> Block Size 0.50KiB
> > >>>> Bucket Size 512.00KiB
> > >>>> Congested? False
> > >>>> Read Congestion 0.0ms
> > >>>> Write Congestion 0.0ms
> > >>>> Total Cache Size 111.79GiB
> > >>>> Total Cache Used 2.24GiB (2%)
> > >>>> Total Cache Unused 109.55GiB (98%)
> > >>>> Evictable Cache 110.67GiB (99%)
> > >>>> Replacement Policy [lru] fifo random
> > >>>> Cache Mode writethrough [writeback] writearound none
> > >>>> Total Hits 5 (0%)
> > >>>> Total Misses 4079
> > >>>> Total Bypass Hits 0 (0%)
> > >>>> Total Bypass Misses 615
> > >>>> Total Bypassed 2.40MiB
> > >>>>
> > >>>>
> > >>>>
> > >>>> Linux kkb 4.19.0-041900-generic #201810221809 SMP Mon Oct 22 22:11:45
> > >>>> UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
> > >>>>
> > >>>> --- bcache ---
> > >>>> Device /dev/md0 (9:0)
> > >>>> UUID 67269654-92e8-4c3b-a524-8e8910082146
> > >>>> Block Size 0.50KiB
> > >>>> Bucket Size 512.00KiB
> > >>>> Congested? False
> > >>>> Read Congestion 0.0ms
> > >>>> Write Congestion 0.0ms
> > >>>> Total Cache Size 111.79GiB
> > >>>> Total Cache Used 1.12GiB (0%)
> > >>>> Total Cache Unused 110.67GiB (99%)
> > >>>> Evictable Cache 111.79GiB (100%)
> > >>>> Replacement Policy [lru] fifo random
> > >>>> Cache Mode writethrough [writeback] writearound none
> > >>>> Total Hits 0 (0%)
> > >>>> Total Misses 1
> > >>>> Total Bypass Hits 0
> > >>>> Total Bypass Misses 0
> > >>>> Total Bypassed 0B
> > >>>>
> > >>>> 1st pass
> > >>>> 0.00user 0.33system 0:09.29elapsed 3%CPU (0avgtext+0avgdata
> > >>>> 2280maxresident)k
> > >>>> 2097624inputs+0outputs (1major+113minor)pagefaults 0swaps
> > >>>>
> > >>>> 2nd pass
> > >>>> 0.00user 0.33system 0:08.47elapsed 4%CPU (0avgtext+0avgdata
> > >>>> 2248maxresident)k
> > >>>> 2097280inputs+0outputs (1major+111minor)pagefaults 0swaps
> > >>>>
> > >>>> 3rd pass
> > >>>> 0.00user 0.37system 0:10.46elapsed 3%CPU (0avgtext+0avgdata
> > >>>> 2220maxresident)k
> > >>>> 2097616inputs+0outputs (1major+114minor)pagefaults 0swaps
> > >>>> --- bcache ---
> > >>>> Device /dev/md0 (9:0)
> > >>>> UUID 67269654-92e8-4c3b-a524-8e8910082146
> > >>>> Block Size 0.50KiB
> > >>>> Bucket Size 512.00KiB
> > >>>> Congested? False
> > >>>> Read Congestion 0.0ms
> > >>>> Write Congestion 0.0ms
> > >>>> Total Cache Size 111.79GiB
> > >>>> Total Cache Used 1.12GiB (0%)
> > >>>> Total Cache Unused 110.67GiB (99%)
> > >>>> Evictable Cache 111.79GiB (100%)
> > >>>> Replacement Policy [lru] fifo random
> > >>>> Cache Mode writethrough [writeback] writearound none
> > >>>> Total Hits 132 (23%)
> > >>>> Total Misses 436
> > >>>> Total Bypass Hits 51 (0%)
> > >>>> Total Bypass Misses 17399
> > >>>> Total Bypassed 43.50MiB
> > >>>>
> > >>>>
> > >>>>
> > >>>>
> > >>>> Linux kkb 5.5.2-050502-generic #202002041931 SMP Tue Feb 4 19:33:15 UTC
> > >>>> 2020 x86_64 x86_64 x86_64 GNU/Linux
> > >>>>
> > >>>> --- bcache ---
> > >>>> Device /dev/md0 (9:0)
> > >>>> UUID 38a8b675-e332-4076-b0cf-44e4be72c300
> > >>>> Block Size 0.50KiB
> > >>>> Bucket Size 512.00KiB
> > >>>> Congested? False
> > >>>> Read Congestion 0.0ms
> > >>>> Write Congestion 0.0ms
> > >>>> Total Cache Size 111.79GiB
> > >>>> Total Cache Used 1.12GiB (0%)
> > >>>> Total Cache Unused 110.67GiB (99%)
> > >>>> Evictable Cache 111.79GiB (100%)
> > >>>> Replacement Policy [lru] fifo random
> > >>>> Cache Mode writethrough [writeback] writearound none
> > >>>> Total Hits 0 (0%)
> > >>>> Total Misses 1
> > >>>> Total Bypass Hits 0 (0%)
> > >>>> Total Bypass Misses 3
> > >>>> Total Bypassed 52.00KiB
> > >>>>
> > >>>> 1st pass
> > >>>> 0.00user 0.42system 0:09.21elapsed 4%CPU (0avgtext+0avgdata
> > >>>> 2216maxresident)k
> > >>>> 2097608inputs+0outputs (1major+112minor)pagefaults 0swaps
> > >>>>
> > >>>> 2nd pass
> > >>>> 0.00user 0.42system 0:09.62elapsed 4%CPU (0avgtext+0avgdata
> > >>>> 2248maxresident)k
> > >>>> 2097280inputs+0outputs (1major+112minor)pagefaults 0swaps
> > >>>>
> > >>>> 3rd pass
> > >>>> 0.00user 0.43system 0:08.75elapsed 5%CPU (0avgtext+0avgdata
> > >>>> 2220maxresident)k
> > >>>> 2097224inputs+0outputs (1major+114minor)pagefaults 0swaps
> > >>>> --- bcache ---
> > >>>> Device /dev/md0 (9:0)
> > >>>> UUID 38a8b675-e332-4076-b0cf-44e4be72c300
> > >>>> Block Size 0.50KiB
> > >>>> Bucket Size 512.00KiB
> > >>>> Congested? False
> > >>>> Read Congestion 0.0ms
> > >>>> Write Congestion 0.0ms
> > >>>> Total Cache Size 111.79GiB
> > >>>> Total Cache Used 1.12GiB (0%)
> > >>>> Total Cache Unused 110.67GiB (99%)
> > >>>> Evictable Cache 111.79GiB (100%)
> > >>>> Replacement Policy [lru] fifo random
> > >>>> Cache Mode writethrough [writeback] writearound none
> > >>>> Total Hits 121 (32%)
> > >>>> Total Misses 246
> > >>>> Total Bypass Hits 15 (0%)
> > >>>> Total Bypass Misses 12811
> > >>>> Total Bypassed 39.70MiB
> > >>>
> > >>>
> > >>>
> > >>> --
> > >>> --
> > >>> Ville Aakko - ville.aakko@gmail.com
>
>
>
> --
> --
> Ville Aakko - ville.aakko@gmail.com
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: reads no longer cached since kernel 4.19
2020-02-12 6:02 ` reads no longer cached since kernel 4.19 Postgarage Graz IT
2020-02-12 9:33 ` Matthias Ferdinand
@ 2020-02-17 14:31 ` Nix
1 sibling, 0 replies; 7+ messages in thread
From: Nix @ 2020-02-17 14:31 UTC (permalink / raw)
To: Postgarage Graz IT; +Cc: linux-bcache
On 12 Feb 2020, Postgarage Graz stated:
> On 10.02.20 17:10, Ville Aakko wrote:
>> Hi,
>>
>> A fellow user responding here.
>>
>> I've noticed similar behavior and have asked on this same mailing list
>> previously. See:
>> https://www.spinics.net/lists/linux-bcache/msg07859.html
>>
>> Also seems there are other users with this issue on the Arch Forum,
>> where I have also started a discussion:
>> https://bbs.archlinux.org/viewtopic.php?id=250525
>> There is yet to be a single user to reply there (or on this mailing
>> list) claiming they have a working setup (for caching reads).
>>
>> Judging from the Arch Linux thread, I have a hunch there were some
>> changes ~4.18, which broke read caching for many (all?) desktop users
>> (as anything which is flagged as readahed will not be cached, despite
>> setting sequential_cutoff). Also (again from the Arch thread) a
>> planned patch might enable expected read caching: "[PATCH 3/5] bcache:
>> add readahead cache policy options via sysfs interface" / see:
>> https://www.spinics.net/lists/linux-bcache/msg08074.html
>
> Indeed that patch works.
> Now I'm using the 5.6-rc1 kernel and the performance gain is huge.
Note: 4.19 had an *extra* bug as well, which eliminated all metadata
caching on some filesystems (like XFS, but IIRC not ext4). It was fixed
in v5.1 by commit dc7292a5bcb4c878b.
So you had two problems :)
(I've just moved to a readahead-caching kernel, and while I don't see
any performance gain yet, I'm sure it will come once the cache finishes
populating. It's certainly seeing more writes, 20GiB written in only two
days where before it took a month to write that much.)
Note: the readahead fix was well-timed, since it was only in v5.4 that
the Linux NFS client stopped hardwiring a readahead size of 15 times the
optimal read size, i.e., uh, 15MiB with most servers. That really would
have filled the bcache of the NFS server with a lot of junk.)
^ permalink raw reply [flat|nested] 7+ messages in thread
* reads no longer cached since kernel 4.19
@ 2020-02-09 21:31 Postgarage Graz IT
0 siblings, 0 replies; 7+ messages in thread
From: Postgarage Graz IT @ 2020-02-09 21:31 UTC (permalink / raw)
To: linux-bcache
Hello!
I noticed, that bcache is no longer caching reads on my system which
makes it behave like if there were only hdds.
I'm using two hdds in a raid 1 as the backing device and a single ssd as
cache device:
sda 8:0 0 111,8G 0 disk
└─bcache0 252:0 0 921,9G 0 disk /
sdb 8:16 0 931,5G 0 disk
├─sdb1 8:17 0 922G 0 part
│ └─md0 9:0 0 921,9G 0 raid1
│ └─bcache0 252:0 0 921,9G 0 disk /
├─sdb2 8:18 0 1K 0 part
├─sdb5 8:21 0 1,9G 0 part
│ └─md1 9:1 0 1,9G 0 raid1 /boot
└─sdb6 8:22 0 7,6G 0 part [SWAP]
sdc 8:32 0 931,5G 0 disk
├─sdc1 8:33 0 922G 0 part
│ └─md0 9:0 0 921,9G 0 raid1
│ └─bcache0 252:0 0 921,9G 0 disk /
├─sdc2 8:34 0 1K 0 part
├─sdc5 8:37 0 1,9G 0 part
│ └─md1 9:1 0 1,9G 0 raid1 /boot
└─sdc6 8:38 0 7,6G 0 part [SWAP]
For benchmarking every time I detach the cache device, stop the bcache
device, do a wipefs on the cache device, then make-bcache -C /dev/sda
and finally reattach the cache.
After that, I'm using the following script to repeatedly read a 1gb file:
#!/bin/sh
echo 0 > /sys/block/bcache0/bcache/sequential_cutoff
echo 0 > /sys/block/bcache0/bcache/cache/congested_read_threshold_us
echo 0 > /sys/block/bcache0/bcache/cache/congested_write_threshold_us
uname -a
echo
bcache-status
echo
echo "1st pass"
sync; echo 3 > /proc/sys/vm/drop_caches
(time cat 1GB.bin > /dev/null)
echo
echo "2nd pass"
sync; echo 3 > /proc/sys/vm/drop_caches
(time cat 1GB.bin > /dev/null)
echo
echo "3rd pass"
sync; echo 3 > /proc/sys/vm/drop_caches
(time cat 1GB.bin > /dev/null)
bcache-status
As you can see from the results below, kernel 4.18.20 is the last
kernel, where the cache grows and the performance goes up.
I also compiled 4.19.0 with the bcache files from 4.18.20 and much to my
suprise, that didn't change 4.19's behavior - still no caching. So some
other changes must be the culprit or I did something wrong.
I'm not that much into compiling the kernel, but I checked out the
4.19.0 and 4.18.20 commits and replaced the 4.19.0 drivers/md/bcache
directory with the one from 4.18.20 - then recompiled and installed the
new kernel.
So i am at my wits end. Any help would be appreciated.
Thanks
Flo
Linux kkb 4.18.20-041820-generic #201812030624 SMP Mon Dec 3 11:25:55
UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
--- bcache ---
Device /dev/md0 (9:0)
UUID 8275bf01-f0b3-423e-87fa-48336ce33068
Block Size 0.50KiB
Bucket Size 512.00KiB
Congested? False
Read Congestion 0.0ms
Write Congestion 0.0ms
Total Cache Size 111.79GiB
Total Cache Used 1.12GiB (0%)
Total Cache Unused 110.67GiB (99%)
Evictable Cache 111.79GiB (100%)
Replacement Policy [lru] fifo random
Cache Mode writethrough [writeback] writearound none
Total Hits 0 (0%)
Total Misses 6
Total Bypass Hits 0
Total Bypass Misses 0
Total Bypassed 0B
1st pass
0.00user 0.36system 0:08.58elapsed 4%CPU (0avgtext+0avgdata
2196maxresident)k
2097608inputs+0outputs (1major+113minor)pagefaults 0swaps
2nd pass
0.00user 0.32system 0:03.29elapsed 9%CPU (0avgtext+0avgdata
2100maxresident)k
2097184inputs+0outputs (1major+110minor)pagefaults 0swaps
3rd pass
0.00user 0.32system 0:02.64elapsed 12%CPU (0avgtext+0avgdata
2092maxresident)k
2097280inputs+0outputs (1major+111minor)pagefaults 0swaps
--- bcache ---
Device /dev/md0 (9:0)
UUID 8275bf01-f0b3-423e-87fa-48336ce33068
Block Size 0.50KiB
Bucket Size 512.00KiB
Congested? False
Read Congestion 0.0ms
Write Congestion 0.0ms
Total Cache Size 111.79GiB
Total Cache Used 2.24GiB (2%)
Total Cache Unused 109.55GiB (98%)
Evictable Cache 110.67GiB (99%)
Replacement Policy [lru] fifo random
Cache Mode writethrough [writeback] writearound none
Total Hits 5 (0%)
Total Misses 4079
Total Bypass Hits 0 (0%)
Total Bypass Misses 615
Total Bypassed 2.40MiB
Linux kkb 4.19.0-041900-generic #201810221809 SMP Mon Oct 22 22:11:45
UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
--- bcache ---
Device /dev/md0 (9:0)
UUID 67269654-92e8-4c3b-a524-8e8910082146
Block Size 0.50KiB
Bucket Size 512.00KiB
Congested? False
Read Congestion 0.0ms
Write Congestion 0.0ms
Total Cache Size 111.79GiB
Total Cache Used 1.12GiB (0%)
Total Cache Unused 110.67GiB (99%)
Evictable Cache 111.79GiB (100%)
Replacement Policy [lru] fifo random
Cache Mode writethrough [writeback] writearound none
Total Hits 0 (0%)
Total Misses 1
Total Bypass Hits 0
Total Bypass Misses 0
Total Bypassed 0B
1st pass
0.00user 0.33system 0:09.29elapsed 3%CPU (0avgtext+0avgdata
2280maxresident)k
2097624inputs+0outputs (1major+113minor)pagefaults 0swaps
2nd pass
0.00user 0.33system 0:08.47elapsed 4%CPU (0avgtext+0avgdata
2248maxresident)k
2097280inputs+0outputs (1major+111minor)pagefaults 0swaps
3rd pass
0.00user 0.37system 0:10.46elapsed 3%CPU (0avgtext+0avgdata
2220maxresident)k
2097616inputs+0outputs (1major+114minor)pagefaults 0swaps
--- bcache ---
Device /dev/md0 (9:0)
UUID 67269654-92e8-4c3b-a524-8e8910082146
Block Size 0.50KiB
Bucket Size 512.00KiB
Congested? False
Read Congestion 0.0ms
Write Congestion 0.0ms
Total Cache Size 111.79GiB
Total Cache Used 1.12GiB (0%)
Total Cache Unused 110.67GiB (99%)
Evictable Cache 111.79GiB (100%)
Replacement Policy [lru] fifo random
Cache Mode writethrough [writeback] writearound none
Total Hits 132 (23%)
Total Misses 436
Total Bypass Hits 51 (0%)
Total Bypass Misses 17399
Total Bypassed 43.50MiB
Linux kkb 5.5.2-050502-generic #202002041931 SMP Tue Feb 4 19:33:15 UTC
2020 x86_64 x86_64 x86_64 GNU/Linux
--- bcache ---
Device /dev/md0 (9:0)
UUID 38a8b675-e332-4076-b0cf-44e4be72c300
Block Size 0.50KiB
Bucket Size 512.00KiB
Congested? False
Read Congestion 0.0ms
Write Congestion 0.0ms
Total Cache Size 111.79GiB
Total Cache Used 1.12GiB (0%)
Total Cache Unused 110.67GiB (99%)
Evictable Cache 111.79GiB (100%)
Replacement Policy [lru] fifo random
Cache Mode writethrough [writeback] writearound none
Total Hits 0 (0%)
Total Misses 1
Total Bypass Hits 0 (0%)
Total Bypass Misses 3
Total Bypassed 52.00KiB
1st pass
0.00user 0.42system 0:09.21elapsed 4%CPU (0avgtext+0avgdata
2216maxresident)k
2097608inputs+0outputs (1major+112minor)pagefaults 0swaps
2nd pass
0.00user 0.42system 0:09.62elapsed 4%CPU (0avgtext+0avgdata
2248maxresident)k
2097280inputs+0outputs (1major+112minor)pagefaults 0swaps
3rd pass
0.00user 0.43system 0:08.75elapsed 5%CPU (0avgtext+0avgdata
2220maxresident)k
2097224inputs+0outputs (1major+114minor)pagefaults 0swaps
--- bcache ---
Device /dev/md0 (9:0)
UUID 38a8b675-e332-4076-b0cf-44e4be72c300
Block Size 0.50KiB
Bucket Size 512.00KiB
Congested? False
Read Congestion 0.0ms
Write Congestion 0.0ms
Total Cache Size 111.79GiB
Total Cache Used 1.12GiB (0%)
Total Cache Unused 110.67GiB (99%)
Evictable Cache 111.79GiB (100%)
Replacement Policy [lru] fifo random
Cache Mode writethrough [writeback] writearound none
Total Hits 121 (32%)
Total Misses 246
Total Bypass Hits 15 (0%)
Total Bypass Misses 12811
Total Bypassed 39.70MiB
^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2020-02-17 14:31 UTC | newest]
Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
[not found] <b039d510-9b03-e6a3-499a-1dbe72764cbe@postgarage.at>
2020-02-12 6:02 ` reads no longer cached since kernel 4.19 Postgarage Graz IT
2020-02-12 9:33 ` Matthias Ferdinand
2020-02-12 10:03 ` Postgarage Graz IT
2020-02-12 20:32 ` Ville Aakko
2020-02-13 11:00 ` Matthias Ferdinand
2020-02-17 14:31 ` Nix
2020-02-09 21:31 Postgarage Graz IT
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.