question regarding caching

question regarding caching

[Aastha Mehta]

Hello,

I have some questions regarding caching in BTRFS. When a file system
is unmounted and mounted again, would all the previously cached
content be removed from the cache after flushing to disk? After
remounting, would the initial requests always be fetched from the
disk?

Rather than a local disk, I have a remote device to which my IO
requests are sent and from which the data is fetched. I need certain
data to be fetched from the remote device after a remount. But somehow
I do not see any request appearing at the device. I even tried to do
drop_caches after remounting the file system, but that does not seem
to help.

I guess my problem is not related to BTRFS, but since I am working
with BTRFS, I wanted to ask here for help. Could any one tell me how I
can ensure that requests are fetched from the (remote) device,
especially after file system remount, without having to use
drop_caches?

Please let me know if I described the problem too vaguely and should
give some more details.

Wishing everyone a happy new year.

Thanks and regards,
Aastha.

-- 
Aastha Mehta
MPI-SWS, Germany
E-mail: aasthakm@mpi-sws.org

Re: question regarding caching

[Kai Krakow]

Aastha Mehta <aasthakm@gmail.com> schrieb:

> Rather than a local disk, I have a remote device to which my IO
> requests are sent and from which the data is fetched. I need certain
> data to be fetched from the remote device after a remount. But somehow
> I do not see any request appearing at the device. I even tried to do
> drop_caches after remounting the file system, but that does not seem
> to help.

Maybe you or your distribution deployed cachefilesd and uses it for the 
remote fs?

HTH
Kai

Re: question regarding caching

[Alex Elsayed]

Kai Krakow wrote:

> Aastha Mehta <aasthakm@gmail.com> schrieb:
> 
>> Rather than a local disk, I have a remote device to which my IO
>> requests are sent and from which the data is fetched. I need certain
>> data to be fetched from the remote device after a remount. But somehow
>> I do not see any request appearing at the device. I even tried to do
>> drop_caches after remounting the file system, but that does not seem
>> to help.
> 
> Maybe you or your distribution deployed cachefilesd and uses it for the
> remote fs?

The FS-Cache infrastructure only applies to filesystems which opt in to 
using it (NFS, recent work on Ceph, etc), so mounting btrfs over a remote 
block device wouldn't use it at all.

Re: question regarding caching

[Austin S Hemmelgarn]

On 12/30/2013 09:24 PM, Aastha Mehta wrote:
> Hello,
> 
> I have some questions regarding caching in BTRFS. When a file
> system is unmounted and mounted again, would all the previously
> cached content be removed from the cache after flushing to disk?
> After remounting, would the initial requests always be fetched from
> the disk?
> 
> Rather than a local disk, I have a remote device to which my IO 
> requests are sent and from which the data is fetched. I need
> certain data to be fetched from the remote device after a remount.
> But somehow I do not see any request appearing at the device. I
> even tried to do drop_caches after remounting the file system, but
> that does not seem to help.
> 
> I guess my problem is not related to BTRFS, but since I am working 
> with BTRFS, I wanted to ask here for help. Could any one tell me
> how I can ensure that requests are fetched from the (remote)
> device, especially after file system remount, without having to
> use drop_caches?
> 
> Please let me know if I described the problem too vaguely and
> should give some more details.
The data is probably still cached in the block layer, so after
unmounting, you could try 'echo 1 > /proc/sys/vm/drop_caches' before
mounting again, but make sure to run sync right before doing that,
otherwise you might lose data.

Re: question regarding caching

[Sander]

Austin S Hemmelgarn wrote (ao):
> The data is probably still cached in the block layer, so after
> unmounting, you could try 'echo 1 > /proc/sys/vm/drop_caches' before
> mounting again, but make sure to run sync right before doing that,
> otherwise you might lose data.

Lose data? Where you get this from?

	Sander

Re: question regarding caching

[Austin S Hemmelgarn]

On 2014-01-03 03:39, Sander wrote:
> Austin S Hemmelgarn wrote (ao):
>> The data is probably still cached in the block layer, so after 
>> unmounting, you could try 'echo 1 > /proc/sys/vm/drop_caches'
>> before mounting again, but make sure to run sync right before
>> doing that, otherwise you might lose data.
> 
> Lose data? Where you get this from?
> 
> Sander
> 
Sorry, misread the documentation, thought it said destructive where it
really said non-destructive.
It's still a good idea to run sync before trying to clear the caches
though, cause dirty objects aren't freeable.

Re: question regarding caching

[Aastha Mehta]

On 3 January 2014 13:12, Austin S Hemmelgarn <ahferroin7@gmail.com> wrote:
>
> On 2014-01-03 03:39, Sander wrote:
> > Austin S Hemmelgarn wrote (ao):
> >> The data is probably still cached in the block layer, so after
> >> unmounting, you could try 'echo 1 > /proc/sys/vm/drop_caches'
> >> before mounting again, but make sure to run sync right before
> >> doing that, otherwise you might lose data.
> >
> > Lose data? Where you get this from?
> >
> > Sander
> >
> Sorry, misread the documentation, thought it said destructive where it
> really said non-destructive.
> It's still a good idea to run sync before trying to clear the caches
> though, cause dirty objects aren't freeable.

Hello,

I figured out what the problem was. After remounting, I was doing
fseek to the end of file to calculate the file size. It seems that
fseek does prefetching of file data, which is why my reads were being
served from the cache rather than from the disk. I could not find
where this prefetching is called from. Anyways, using fstat instead of
fseek to find the file size ensures that there is no unwanted
prefetching.

Thanks,
Aastha.