From: "heming.zhao@suse.com" <heming.zhao@suse.com>
To: David Teigland <teigland@redhat.com>,
Martin Wilck <martin.wilck@suse.com>
Cc: "rogerheflin@gmail.com" <rogerheflin@gmail.com>,
"zkabelac@redhat.com" <zkabelac@redhat.com>,
"linux-lvm@redhat.com" <linux-lvm@redhat.com>
Subject: Re: [linux-lvm] Discussion: performance issue on event activation mode
Date: Wed, 9 Jun 2021 00:18:17 +0800 [thread overview]
Message-ID: <0b83d419-8569-2ba0-5f5b-2c3df4e2eb21@suse.com> (raw)
In-Reply-To: <20210607213003.GA8181@redhat.com>
On 6/8/21 5:30 AM, David Teigland wrote:
> On Mon, Jun 07, 2021 at 10:27:20AM +0000, Martin Wilck wrote:
>> Most importantly, this was about LVM2 scanning of physical volumes. The
>> number of udev workers has very little influence on PV scanning,
>> because the udev rules only activate systemd service. The actual
>> scanning takes place in lvm2-pvscan@.service. And unlike udev, there's
>> no limit for the number of instances of a given systemd service
>> template that can run at any given time.
>
> Excessive device scanning has been the historical problem in this area,
> but Heming mentioned dev_cache_scan() specifically as a problem. That was
> surprising to me since it doesn't scan/read devices, it just creates a
> list of device names on the system (either readdir in /dev or udev
> listing.) If there are still problems with excessive scannning/reading,
> we'll need some more diagnosis of what's happening, there could be some
> cases we've missed.
>
the dev_cache_scan doesn't have direct disk IOs, but libudev will scan/read
udev db which issue real disk IOs (location is /run/udev/data).
we can see with combination "obtain_device_list_from_udev=0 &
event_activation=1" could largely reduce booting time from 2min6s to 40s.
the key is dev_cache_scan() does the scan device by itself (scaning "/dev").
I am not very familiar with systemd-udev, below shows a little more info
about libudev path. the top function is _insert_udev_dir, this function:
1. scans/reads /sys/class/block/. O(n)
2. scans/reads udev db (/run/udev/data). may O(n)
udev will call device_read_db => handle_db_line to handle every
line of a db file.
3. does qsort & deduplication the devices list. O(n) + O(n)
4. has lots of "memory alloc" & "string copy" actions during working.
it takes too much memory, from the host side, use 'top' can see:
- direct activation only used 2G memory during boot
- event activation cost ~20G memory.
I didn't test the related udev code, and guess the <2> takes too much time.
And there are thousand scanning job parallel in /run/udev/data, meanwhile
there are many devices need to generate udev db file in the same dir. I am
not sure if the filesystem can perfect handle this scenario.
the another code path, obtain_device_list_from_udev=0, which triggers to
scan/read "/dev", this dir has less write IOs than /run/udev/data.
Regards
heming
_______________________________________________
linux-lvm mailing list
linux-lvm@redhat.com
https://listman.redhat.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/
next prev parent reply other threads:[~2021-06-08 16:19 UTC|newest]
Thread overview: 86+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-06-06 6:15 [linux-lvm] Discussion: performance issue on event activation mode heming.zhao
2021-06-06 16:35 ` Roger Heflin
2021-06-07 10:27 ` Martin Wilck
2021-06-07 15:30 ` heming.zhao
2021-06-07 15:45 ` Martin Wilck
2021-06-07 20:52 ` Roger Heflin
2021-06-07 21:30 ` David Teigland
2021-06-08 8:26 ` Martin Wilck
2021-06-08 15:39 ` David Teigland
2021-06-08 15:47 ` Martin Wilck
2021-06-08 16:02 ` Zdenek Kabelac
2021-06-08 16:05 ` Martin Wilck
2021-06-08 16:03 ` David Teigland
2021-06-08 16:07 ` Martin Wilck
2021-06-15 17:03 ` David Teigland
2021-06-15 18:21 ` Zdenek Kabelac
2021-06-16 16:18 ` heming.zhao
2021-06-16 16:38 ` David Teigland
2021-06-17 3:46 ` heming.zhao
2021-06-17 15:27 ` David Teigland
2021-06-08 16:49 ` heming.zhao
2021-06-08 16:18 ` heming.zhao [this message]
2021-06-09 4:01 ` heming.zhao
2021-06-09 5:37 ` Heming Zhao
2021-06-09 18:59 ` David Teigland
2021-06-10 17:23 ` heming.zhao
2021-06-07 15:48 ` Martin Wilck
2021-06-07 16:31 ` Zdenek Kabelac
2021-06-07 21:48 ` David Teigland
2021-06-08 12:29 ` Peter Rajnoha
2021-06-08 13:23 ` Martin Wilck
2021-06-08 13:41 ` Peter Rajnoha
2021-06-08 13:46 ` Zdenek Kabelac
2021-06-08 13:56 ` Peter Rajnoha
2021-06-08 14:23 ` Zdenek Kabelac
2021-06-08 14:48 ` Martin Wilck
2021-06-08 15:19 ` Peter Rajnoha
2021-06-08 15:39 ` Martin Wilck
2021-09-09 19:44 ` David Teigland
2021-09-10 17:38 ` Martin Wilck
2021-09-12 16:51 ` heming.zhao
2021-09-27 10:00 ` Peter Rajnoha
2021-09-27 15:38 ` David Teigland
2021-09-28 6:34 ` Martin Wilck
2021-09-28 14:42 ` David Teigland
2021-09-28 15:16 ` Martin Wilck
2021-09-28 15:31 ` Martin Wilck
2021-09-28 15:56 ` David Teigland
2021-09-28 18:03 ` Benjamin Marzinski
2021-09-28 17:42 ` Benjamin Marzinski
2021-09-28 19:15 ` Martin Wilck
2021-09-29 22:06 ` Peter Rajnoha
2021-09-30 7:51 ` Martin Wilck
2021-09-30 8:07 ` heming.zhao
2021-09-30 9:31 ` Martin Wilck
2021-09-30 11:41 ` Peter Rajnoha
2021-09-30 15:32 ` heming.zhao
2021-10-01 7:41 ` Martin Wilck
2021-10-01 8:08 ` Peter Rajnoha
2021-09-30 11:29 ` Peter Rajnoha
2021-09-30 16:04 ` David Teigland
2021-09-30 14:41 ` Benjamin Marzinski
2021-10-01 7:42 ` Martin Wilck
2021-09-29 21:53 ` Peter Rajnoha
2021-09-30 7:45 ` Martin Wilck
2021-09-29 21:39 ` Peter Rajnoha
2021-09-30 7:22 ` Martin Wilck
2021-09-30 14:26 ` David Teigland
2021-09-30 15:55 ` David Teigland
2021-10-01 8:00 ` Peter Rajnoha
2021-10-18 6:24 ` Martin Wilck
2021-10-18 15:04 ` David Teigland
2021-10-18 16:56 ` heming.zhao
2021-10-18 21:51 ` Zdenek Kabelac
2021-10-19 17:18 ` David Teigland
2021-10-20 14:40 ` Martin Wilck
2021-10-20 14:50 ` David Teigland
2021-10-20 14:54 ` Martin Wilck
2021-10-20 15:12 ` David Teigland
2021-06-07 16:40 ` David Teigland
2021-07-02 21:09 ` David Teigland
2021-07-02 21:22 ` Martin Wilck
2021-07-02 22:02 ` David Teigland
2021-07-03 11:49 ` heming.zhao
2021-07-08 10:10 ` Tom Yan
2021-07-02 21:31 ` Tom Yan
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=0b83d419-8569-2ba0-5f5b-2c3df4e2eb21@suse.com \
--to=heming.zhao@suse.com \
--cc=linux-lvm@redhat.com \
--cc=martin.wilck@suse.com \
--cc=rogerheflin@gmail.com \
--cc=teigland@redhat.com \
--cc=zkabelac@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).