[Gluster-devel] Report ESTALE as ENOENT

Raghavendra Gowdappa rgowdapp at redhat.com
Thu Mar 24 05:01:59 UTC 2016



----- Original Message -----
> From: "Soumya Koduri" <skoduri at redhat.com>
> To: "Raghavendra Gowdappa" <rgowdapp at redhat.com>
> Cc: "Poornima Gurusiddaiah" <pgurusid at redhat.com>, "Raghavendra Talur" <rtalur at redhat.com>, "Shyamsundar Ranganathan"
> <srangana at redhat.com>, "Vijay Bellur" <vbellur at redhat.com>, "Niels de Vos" <ndevos at redhat.com>, "Ira Cooper"
> <icooper at redhat.com>, "Nithya Balachandran" <nbalacha at redhat.com>, "Gluster Devel" <gluster-devel at gluster.org>
> Sent: Wednesday, March 23, 2016 9:16:19 PM
> Subject: Re: Report ESTALE as ENOENT
> 
> Hi Raghavendra,
> 
> In [1], its mentioned that "when the inode/gfid is missing, brick report back
> as an ESTALE error". Could you please list the possible cases which shall
> result in this behavior.

This can be a racing rm. The test case which uncovered this issue was rm -rf parallely from two mounts on the same directory.

> If it occurs only when the file/dir is not actually
> present at the back-end, shouldn't we fix the server to send ENOENT then?

Ravi has already answered this on another mail. As the link mentions, its to aid interface layers (fuse, gfapi) to retrigger a resolution.

> 
> Thanks,
> Soumya
> 
> ----- Original Message -----
> > From: "Raghavendra Gowdappa" <rgowdapp at redhat.com>
> > To: "Soumya Koduri" <skoduri at redhat.com>, "Poornima Gurusiddaiah"
> > <pgurusid at redhat.com>, "Raghavendra Talur"
> > <rtalur at redhat.com>
> > Cc: "Shyamsundar Ranganathan" <srangana at redhat.com>, "Vijay Bellur"
> > <vbellur at redhat.com>, "Niels de Vos"
> > <ndevos at redhat.com>, "Ira Cooper" <icooper at redhat.com>, "Nithya
> > Balachandran" <nbalacha at redhat.com>, "Gluster Devel"
> > <gluster-devel at gluster.org>
> > Sent: Wednesday, March 23, 2016 2:03:34 PM
> > Subject: Report ESTALE as ENOENT
> > 
> > Hi all,
> > 
> > rm has a different behavior for ESTALE when compared to ENOENT. So, we
> > shouldn't be reporting ENOENT errors as ESTALE. I've a fix for fuse [1].
> > Similar fix is necessary for NFS and gfapi (samba too?).
> > 
> > Also, reviews on [1] is much appreciated as this patch unconditionally
> > converts all ESTALE to ENOENT, which might not be correct behavior always.
> > Sometimes ESTALE might be a valid errno. If you point me when it is
> > necessary to report ESTALE errors unchanged, I'll accommodate the comments
> > in the patch.
> > 
> > @Soumya/Poornima/Raghavendra,
> > 
> > Is it possible to send an analogous patch to NFS and gfapi?
> > 
> > [1] http://review.gluster.org/#/c/13816/
> > 
> > regards,
> > Raghavendra
> > 
> 


More information about the Gluster-devel mailing list