Red Hat Bugzilla – Bug 1025471
CI tests are failing with "0-testvol-dht: (null): failed to get the 'linkto' xattr No data available" on clients.
Last modified: 2013-11-27 10:45:26 EST
Description of problem:
Our BVT and FS Sanity tests are currently failing on the latest 2.1 u1 build with:
[2013-10-31 14:52:53.727404] W [client-rpc-fops.c:1170:client3_3_fgetxattr_cbk] 0-testvol-client-11: remote operation failed: No data available
[2013-10-31 14:52:53.727807] W [client-rpc-fops.c:1170:client3_3_fgetxattr_cbk] 0-testvol-client-10: remote operation failed: No data available
[2013-10-31 14:52:53.727874] E [dht-helper.c:761:dht_migration_complete_check_task] 0-testvol-dht: (null): failed to get the 'linkto' xattr No data available
[2013-10-31 14:52:53.727914] W [page.c:991:__ioc_page_error] 0-testvol-io-cache: page error for page = 0x7f383805c4a0 & waitq = 0x7f38380ee500
[2013-10-31 14:52:53.727946] W [fuse-bridge.c:2617:fuse_readv_cbk] 0-glusterfs-fuse: 15163: READ => -1 (No such file or directory)
All the failed tests that I saw had similar logs on the clients. On the server side I see:
[2013-10-31 15:35:49.675693] W [posix-helpers.c:788:posix_handle_pair] 0-testvol-posix: Extended attributes not supported (try remounting brick with 'user_xattr' flag)
[2013-10-31 15:35:49.675740] E [posix.c:915:posix_mknod] 0-testvol-posix: setting xattrs on /bricks/testvol_brick0/run6006/linux-220.127.116.11/.tmpconfig.7289 failed (Operation not supported)
[2013-10-31 15:35:57.423014] E [posix.c:915:posix_mknod] 0-testvol-posix: setting xattrs on /bricks/testvol_brick0/run6006/linux-18.104.22.168/.tmpconfig.h failed (Operation not supported)
[2013-10-31 15:36:00.109750] E [posix.c:915:posix_mknod] 0-testvol-posix: setting xattrs on /bricks/testvol_brick0/run6006/linux-22.214.171.124/include/linux/utsrelease.h.tmp failed (Operation not supported)
Has anything changed with how I need to mount bricks or is that warning a red herring?
Version-Release number of selected component (if applicable):
Every time I have ran automated tests on this build.
Steps to Reproduce:
1. Compile the linux kernel on a glusterfs filesystem.
Compile is successful.
This is indeed a glusterfs regression. Bad patch is a quota related commit which has broken rename() and link() calls:
Author: Raghavendra G <email@example.com>
Date: Wed Oct 30 14:15:23 2013 +0530
cluster/dht: instruct marker whenever it shouldn't do accounting
This is needed for two reasons:
* since dht-linkfiles are internal, they shouldn't be accounted.
* hardlink handling in marker is broken. link/unlink of hardlinks
present in same directory can break marker accounting. Hence, if src
and dst are in same directory in case of rename, dht - if it breaks
rename into link/unlink operations - should instruct marker to not to
Signed-off-by: Raghavendra G <firstname.lastname@example.org>
Reviewed-by: Krishnan Parthasarathi <email@example.com>
Tested-by: Krishnan Parthasarathi <firstname.lastname@example.org>
Verified on glusterfs-126.96.36.199rhs. To note, I did see BZ 1025604 which still needs to be addressed, but the issues this BZ were opened for are resolved.
*** Bug 1026272 has been marked as a duplicate of this bug. ***
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory, and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.