mkfs and mount tips?

mkfs and mount tips?

[Leon Kolchinsky]

Hello All, 

I have Pentium II (Deschutes) with first 10GB (/dev/hda) and second
60GB(/dev/hdc) disk. 
After reading gentoo xfs threads on their forum and some SGI docs and FAQs I
came with this options for creating FS and mounting the disks: 

1) To create XFS on hda: 

Code:	
# mkfs.xfs -l internal,size=128m -d agcount=2 /dev/hda	


I've also seen "–d unwritten=0" option: 

So my question: 
Is it safe to add –d unwritten=0 option to increase performance like this
(or will I lose some essential functionality)?: 

Is this how the code should look?:	
# mkfs.xfs -l internal,size=128m -d agcount=2 –d unwritten=0 /dev/hda	


2) To prevent data lost in case of power outage(Disabling the write back
cache): 
Add the following to local.start: 

Code:	
# hdparm -W0 /dev/hda
# hdparm -W0 /dev/hdc
# blktool /dev/hda wcache off 
# blktool /dev/hdc wcache off	


Right? 

3) Mount options: 

On gentoo xfs thread it's suggested that the mount options should be
"noatime,logbufs=8" 

But what about "osyncisdsync" mount option. 

"• osyncisdsync
– Writes to files opened with the O_SYNC flag set will behave as if the
O_DSYNC flag had been used instead. 
– This can result in better performance without compromising data safety. 
– However timestamp updates from O_SYNC writes can be lost if the system
crashes. 
Use osyncisosync to disable this setting."

So do you think it is safe to add "osyncisdsync" mount option to fstab? 


I'd appreciate any comments/tips/answers.


Best Regards,
Leon Kolchinsky

RE: mkfs and mount tips?

[Leon Kolchinsky]

> -----Original Message-----
> From: Timothy Shimmin [mailto:tes@sgi.com]
> Sent: Thursday, February 15, 2007 3:19 AM
> To: Leon Kolchinsky; linux-xfs@oss.sgi.com
> Subject: Re: mkfs and mount tips?
> 
> Hi Leon,
> 
> --On 13 February 2007 10:48:15 AM +0200 Leon Kolchinsky
> <leonk@construct.haifa.ac.il> wrote:
> 
> > I've also seen "–d unwritten=0" option:
> >
> > So my question:
> > Is it safe to add –d unwritten=0 option to increase performance like
> this
> > (or will I lose some essential functionality)?:
> >
> 
> My understanding (although I'm not familiar with that code),
> is that unwritten extents are used in space preallocation.
> So unless you reserve space for a file it will not have an effect.
> And if you do, then setting "unwritten=0" will speed up writes because it
> doesn't
> need to flag the unwritten extents and write out the extra transactions
> for this.
> If the unwritten extents aren't flagged as such then there can be a
> security issue
> where one can read old data (other's data) for these unwritten parts.
> In fact, the security issue on preallocation (1997-98 sgi-pv#705217) was
> what motivated
> the idea of flagging extents as unwritten in the first place.
> 

Thanks for clearing that out :)

> --Tim

Re: mkfs and mount tips?

[Timothy Shimmin]

Hi Leon,

--On 13 February 2007 10:48:15 AM +0200 Leon Kolchinsky <leonk@construct.haifa.ac.il> wrote:

> I've also seen "–d unwritten=0" option:
>
> So my question:
> Is it safe to add –d unwritten=0 option to increase performance like this
> (or will I lose some essential functionality)?:
>

My understanding (although I'm not familiar with that code),
is that unwritten extents are used in space preallocation.
So unless you reserve space for a file it will not have an effect.
And if you do, then setting "unwritten=0" will speed up writes because it doesn't
need to flag the unwritten extents and write out the extra transactions for this.
If the unwritten extents aren't flagged as such then there can be a security issue
where one can read old data (other's data) for these unwritten parts.
In fact, the security issue on preallocation (1997-98 sgi-pv#705217) was what motivated
the idea of flagging extents as unwritten in the first place.

--Tim

Re: mkfs and mount tips?

[Eric Sandeen]

Leon Kolchinsky wrote:

> What about "osyncisdsync" mount option, can someone recommend this based on
> its own experience?

[esandeen@neon linux-2.6.20]$ grep -r osyncisdsync fs/xfs/xfs_vfsops.c
                 } else if (!strcmp(this_char, "osyncisdsync")) {
         "XFS: osyncisdsync is now the default, option is deprecated.");
[esandeen@neon linux-2.6.20]$

-Eric

RE: mkfs and mount tips?

[Leon Kolchinsky]

> (...)
> > > When I was trying out different optimizations (and what I currently
> use on
> > > a couple volumes is):
> > >
> > > logbufs=8,logbsize=262144,biosize=16,noatime,nodiratime,nobarrier
> > >
> >
> > I think that "nobarrier" option is bad, cause power failure would
> corrupt
> > your FS in this case.
> It depends on hardware. If you use SCSI with write cache turned off I
> see no reason to loss data. You might use RAID with write cache (battery
> powered) safely too.
> I'm not sure but even (S)ATA with write cache disabled should be fine.
> 

Thanks for the tips.

1) Is "–d unwritten=0" option is pretty safe?

2) What's the recommended value of "-d agcount=" for?:

10GB disk for web/mail/streaming server accordingly
50GB disk for web/mail/streaming server accordingly
300GB disk for web/mail/streaming server accordingly

Or how can I calculate these recommended values?

> Regards,
> 
> Olaf
> --
> Olaf Fr?czyk <olaf@cbk.poznan.pl>


Regards,
Leon Kolchinsky

Re: mkfs and mount tips?

[Andi Kleen]

"Leon Kolchinsky" <leonk@construct.haifa.ac.il> writes:
> 
> What about "osyncisdsync" mount option, can someone recommend this based on
> its own experience?

Not sync metadata on O_SYNC writes.

That's also dangerous -- might cause data corruption after crash. It is mostly
a hack if you know your programs very well, but can't change them
to use O_DSYNC when they only need that.

Also his other options basically all just use more memory. If you
have a P90 you're likely low on memory for modern standards and 
shouldn't use them.

-Andi

RE: mkfs and mount tips?

[Olaf Frączyk]

On Tue, 2007-02-13 at 12:33 +0200, Leon Kolchinsky wrote:
> 
> > -----Original Message-----
> > From: Justin Piszcz [mailto:jpiszcz@lucidpixels.com]
> > Sent: Tuesday, February 13, 2007 11:30 AM
> > To: Leon Kolchinsky
> > Cc: linux-xfs@oss.sgi.com
> > Subject: Re: mkfs and mount tips?
(...)
> > When I was trying out different optimizations (and what I currently use on
> > a couple volumes is):
> > 
> > logbufs=8,logbsize=262144,biosize=16,noatime,nodiratime,nobarrier
> > 
> 
> I think that "nobarrier" option is bad, cause power failure would corrupt
> your FS in this case.
It depends on hardware. If you use SCSI with write cache turned off I
see no reason to loss data. You might use RAID with write cache (battery
powered) safely too.
I'm not sure but even (S)ATA with write cache disabled should be fine.

Regards,

Olaf
-- 
Olaf Frączyk <olaf@cbk.poznan.pl>

RE: mkfs and mount tips?

[Leon Kolchinsky]

> -----Original Message-----
> From: Justin Piszcz [mailto:jpiszcz@lucidpixels.com]
> Sent: Tuesday, February 13, 2007 11:30 AM
> To: Leon Kolchinsky
> Cc: linux-xfs@oss.sgi.com
> Subject: Re: mkfs and mount tips?
> 
> 
> 
> On Tue, 13 Feb 2007, Leon Kolchinsky wrote:
> 
> > Hello All,
> >
> > I have Pentium II (Deschutes) with first 10GB (/dev/hda) and second
> > 60GB(/dev/hdc) disk.
> > After reading gentoo xfs threads on their forum and some SGI docs and
> FAQs I
> > came with this options for creating FS and mounting the disks:
> >
> > 1) To create XFS on hda:
> >
> > Code:
> > # mkfs.xfs -l internal,size=128m -d agcount=2 /dev/hda
> >
> >
> > I've also seen "?d unwritten=0" option:
> >
> > So my question:
> > Is it safe to add ?d unwritten=0 option to increase performance like
> this
> > (or will I lose some essential functionality)?:
> >
> > Is this how the code should look?:
> > # mkfs.xfs -l internal,size=128m -d agcount=2 ?d unwritten=0 /dev/hda
> >
> >
> > 2) To prevent data lost in case of power outage(Disabling the write back
> > cache):
> > Add the following to local.start:
> >
> > Code:
> > # hdparm -W0 /dev/hda
> > # hdparm -W0 /dev/hdc
> > # blktool /dev/hda wcache off
> > # blktool /dev/hdc wcache off
> >
> >
> > Right?
> >
> > 3) Mount options:
> >
> > On gentoo xfs thread it's suggested that the mount options should be
> > "noatime,logbufs=8"
> >
> > But what about "osyncisdsync" mount option.
> >
> > "? osyncisdsync
> > ? Writes to files opened with the O_SYNC flag set will behave as if the
> > O_DSYNC flag
> > had been used instead.
> > ? This can result in better performance without compromising data
> safety.
> > ? However timestamp updates from O_SYNC writes can be lost if the system
> > crashes.
> > Use osyncisosync to disable this setting."
> >
> > So do you think it is safe to add "osyncisdsync" mount option to fstab?
> >
> >
> > I'd appreciate any comments/tips/answers.
> >
> >
> > Best Regards,
> > Leon Kolchinsky
> >
> >
> >
> >
> >
> 
> When I was trying out different optimizations (and what I currently use on
> a couple volumes is):
> 
> logbufs=8,logbsize=262144,biosize=16,noatime,nodiratime,nobarrier
> 

I think that "nobarrier" option is bad, cause power failure would corrupt
your FS in this case.
"nodiratime" is useless here, cause you're already using noatime which
includes nodiratime AFAIK.

What about "osyncisdsync" mount option, can someone recommend this based on
its own experience?

What are other optimization options during mkfs?

Regards,
Leon

Re: mkfs and mount tips?

[Justin Piszcz]

On Tue, 13 Feb 2007, Leon Kolchinsky wrote:

> Hello All,
>
> I have Pentium II (Deschutes) with first 10GB (/dev/hda) and second
> 60GB(/dev/hdc) disk.
> After reading gentoo xfs threads on their forum and some SGI docs and FAQs I
> came with this options for creating FS and mounting the disks:
>
> 1) To create XFS on hda:
>
> Code:
> # mkfs.xfs -l internal,size=128m -d agcount=2 /dev/hda
>
>
> I've also seen "?d unwritten=0" option:
>
> So my question:
> Is it safe to add ?d unwritten=0 option to increase performance like this
> (or will I lose some essential functionality)?:
>
> Is this how the code should look?:
> # mkfs.xfs -l internal,size=128m -d agcount=2 ?d unwritten=0 /dev/hda
>
>
> 2) To prevent data lost in case of power outage(Disabling the write back
> cache):
> Add the following to local.start:
>
> Code:
> # hdparm -W0 /dev/hda
> # hdparm -W0 /dev/hdc
> # blktool /dev/hda wcache off
> # blktool /dev/hdc wcache off
>
>
> Right?
>
> 3) Mount options:
>
> On gentoo xfs thread it's suggested that the mount options should be
> "noatime,logbufs=8"
>
> But what about "osyncisdsync" mount option.
>
> "? osyncisdsync
> ? Writes to files opened with the O_SYNC flag set will behave as if the
> O_DSYNC flag
> had been used instead.
> ? This can result in better performance without compromising data safety.
> ? However timestamp updates from O_SYNC writes can be lost if the system
> crashes.
> Use osyncisosync to disable this setting."
>
> So do you think it is safe to add "osyncisdsync" mount option to fstab?
>
>
> I'd appreciate any comments/tips/answers.
>
>
> Best Regards,
> Leon Kolchinsky
>
>
>
>
>

When I was trying out different optimizations (and what I currently use on 
a couple volumes is):

logbufs=8,logbsize=262144,biosize=16,noatime,nodiratime,nobarrier

Justin.

mkfs and mount tips?

[Leon Kolchinsky]

Hello All, 

I have Pentium II (Deschutes) with first 10GB (/dev/hda) and second
60GB(/dev/hdc) disk. 
After reading gentoo xfs threads on their forum and some SGI docs and FAQs I
came with this options for creating FS and mounting the disks: 

1) To create XFS on hda: 

Code:	
# mkfs.xfs -l internal,size=128m -d agcount=2 /dev/hda	


I've also seen "–d unwritten=0" option: 

So my question: 
Is it safe to add –d unwritten=0 option to increase performance like this
(or will I lose some essential functionality)?: 

Is this how the code should look?:	
# mkfs.xfs -l internal,size=128m -d agcount=2 –d unwritten=0 /dev/hda	


2) To prevent data lost in case of power outage(Disabling the write back
cache): 
Add the following to local.start: 

Code:	
# hdparm -W0 /dev/hda 
# hdparm -W0 /dev/hdc 
# blktool /dev/hda wcache off 
# blktool /dev/hdc wcache off	


Right? 

3) Mount options: 

On gentoo xfs thread it's suggested that the mount options should be
"noatime,logbufs=8" 

But what about "osyncisdsync" mount option. 

"• osyncisdsync 
– Writes to files opened with the O_SYNC flag set will behave as if the
O_DSYNC flag 
had been used instead. 
– This can result in better performance without compromising data safety. 
– However timestamp updates from O_SYNC writes can be lost if the system
crashes. 
Use osyncisosync to disable this setting."

So do you think it is safe to add "osyncisdsync" mount option to fstab? 


I'd appreciate any comments/tips/answers.


Best Regards,
Leon Kolchinsky