[Gluster-devel] Glusto failures with dispersed volumes + Samba

[Ashish Pandey]

Hi Nigel, 

As Pranith has already mentioned, we are getting different gfid's in loc and loc->inode. 
It looks like issue with DHT. If a re validate fails for gfid, a fresh look up should be done. 

I don't know if it is related or not but a similar bug was fixed by Pranith 
https://review.gluster.org/#/c/16986/ 

Ashish 



----- Original Message -----

From: "Pranith Kumar Karampuri" <pkarampu at redhat.com> 
To: "Anoop C S" <anoopcs at autistici.org> 
Cc: "gluster-devel" <gluster-devel at gluster.org> 
Sent: Thursday, June 29, 2017 7:36:45 PM 
Subject: Re: [Gluster-devel] Glusto failures with dispersed volumes + Samba 



On Thu, Jun 29, 2017 at 6:49 PM, Anoop C S < anoopcs at autistici.org > wrote: 


On Thu, 2017-06-29 at 16:35 +0530, Nigel Babu wrote: 
> Hi Pranith and Xavi, 
> 
> We seem to be running into a problem with glusto tests when we try to run them against dispersed 
> volumes over a CIFS mount[1]. 

Is this a new test case? If not was it running successfully before? 

> You can find the logs attached to the job [2]. 

VFS stat call failures are seen in Samba logs: 

[2017/06/29 11:01:55.959374, 0] ../source3/modules/vfs_glusterfs.c:870(vfs_gluster_stat) 
glfs_stat(.) failed: Invalid argument 

I could also see the following errors(repeatedly..) in glusterfs client logs: 

[2017-06-29 10:33:43.031198] W [MSGID: 122019] [ec-helpers.c:412:ec_loc_gfid_check] 0- 
testvol_distributed-dispersed-disperse-0: Mismatching GFID's in loc 
[2017-06-29 10:33:43.031303] I [MSGID: 109094] [dht-common.c:1016:dht_revalidate_cbk] 0- 
testvol_distributed-dispersed-dht: Revalidate: subvolume testvol_distributed-dispersed-disperse-0 
for /user11 (gfid = 665c515b-3940-480f-af7c-6aaf37731eaa) returned -1 [Invalid argument] 




This log basically says that EC received loc which has different gfids in loc->inode->gfid and loc->gfid. 

<blockquote>

> I've triggered a fresh job[3] to confirm that it only fails in these particular conditions and 
> certainly seems to be the case. The job is currently ongoing, so you may want to take a look when 
> you get some time how this job went. 
> 
> Let me know if you have any questions or need more debugging information. 
> 
> [1]: https://ci.centos.org/job/gluster_glusto/325/testReport/ 
> [2]: https://ci.centos.org/job/gluster_glusto/325/artifact/ 
> [3]: https://ci.centos.org/job/gluster_glusto/326/console 
> 
> 
> _______________________________________________ 
> Gluster-devel mailing list 
> Gluster-devel at gluster.org 
> http://lists.gluster.org/mailman/listinfo/gluster-devel 

</blockquote>




-- 
Pranith 

_______________________________________________ 
Gluster-devel mailing list 
Gluster-devel at gluster.org 
http://lists.gluster.org/mailman/listinfo/gluster-devel 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.gluster.org/pipermail/gluster-devel/attachments/20170705/e67721f9/attachment.html>