<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<p>The test is failing on master without any patches:</p>
<tt>[root@tuxpad glusterfs]# prove tests/bugs/bug-1371806_1.t</tt><tt><br>
</tt><tt>tests/bugs/bug-1371806_1.t .. 7/9 setfattr: ./tmp1: No such
file or directory</tt><tt><br>
</tt><tt>setfattr: ./tmp2: No such file or directory</tt><tt><br>
</tt><tt>setfattr: ./tmp3: No such file or directory</tt><tt><br>
</tt><tt>setfattr: ./tmp4: No such file or directory</tt><tt><br>
</tt><tt>setfattr: ./tmp5: No such file or directory</tt><tt><br>
</tt><tt>setfattr: ./tmp6: No such file or directory</tt><tt><br>
</tt><tt>setfattr: ./tmp7: No such file or directory</tt><tt><br>
</tt><tt>setfattr: ./tmp8: No such file or directory</tt><tt><br>
</tt><tt>setfattr: ./tmp9: No such file or directory</tt><tt><br>
</tt><tt>setfattr: ./tmp10: No such file or directory</tt><tt><br>
</tt><tt>./tmp1: user.foo: No such attribute</tt><tt><br>
</tt><tt>tests/bugs/bug-1371806_1.t .. Failed 2/9 subtests</tt><tt><br>
</tt><br>
Mount log for one of the directories:<br>
<tt>[2017-10-06 05:32:10.059798] I [MSGID: 109005]
[dht-selfheal.c:2458:dht_selfheal_directory] 0-patchy-dht:
Directory selfheal failed: Unable to form layout for directory
/tmp1</tt><tt><br>
</tt><tt>[2017-10-06 05:32:10.060013] E [MSGID: 109011]
[dht-common.c:5011:dht_dir_common_setxattr] 0-patchy-dht: Failed
to get mds subvol for path /tmp1gfid is
00000000-0000-0000-0000-000000000000</tt><tt><br>
</tt><tt>[2017-10-06 05:32:10.060041] W
[fuse-bridge.c:1377:fuse_err_cbk] 0-glusterfs-fuse: 99: SETXATTR()
/tmp1 => -1 (No such file or directory)</tt><br>
<br>
Request the patch authors to take a look at it.<br>
Thanks<br>
Ravi<br>
<br>
<div class="moz-cite-prefix">On 10/05/2017 06:04 PM, Atin Mukherjee
wrote:<br>
</div>
<blockquote type="cite"
cite="mid:CAGNCGH34XBbNMvu9y-VuxOJ3SZuFVbGs6cMncHGtLmEvf7ms_g@mail.gmail.com">
<div dir="ltr">The following commit has broken the brick
multiplexing regression job. tests/bugs/bug-1371806_1.t has
failed couple of times. One of the latest regression job report
is at <a
href="https://build.gluster.org/job/regression-test-with-multiplex/406/console"
moz-do-not-send="true">https://build.gluster.org/job/regression-test-with-multiplex/406/console</a>
.<br>
<br>
<br>
commit 9b4de61a136b8e5ba7bf0e48690cdb1292d0dee8<br>
Author: Mohit Agrawal <<a href="mailto:moagrawa@redhat.com"
moz-do-not-send="true">moagrawa@redhat.com</a>><br>
Date: Fri May 12 21:12:47 2017 +0530<br>
<br>
cluster/dht : User xattrs are not healed after brick
stop/start<br>
<br>
Problem: In a distributed volume custom extended attribute
value for a directory<br>
does not display correct value after stop/start or
added newly brick.<br>
If any extended(acl) attribute value is set for a
directory after stop/added<br>
the brick the attribute(user|acl|quota) value is
not updated on brick<br>
after start the brick.<br>
<br>
Solution: First store hashed subvol or subvol(has internal
xattr) on inode ctx and<br>
consider it as a MDS subvol.At the time of update
custom xattr<br>
(user,quota,acl, selinux) on directory first check
the mds from<br>
inode ctx, if mds is not present on inode ctx then
throw EINVAL error<br>
to application otherwise set xattr on MDS subvol
with internal xattr<br>
value of -1 and then try to update the attribute
on other non MDS<br>
volumes also.If mds subvol is down in that case
throw an<br>
error "Transport endpoint is not connected". In
dht_dir_lookup_cbk|<br>
dht_revalidate_cbk|dht_discover_complete call
dht_call_dir_xattr_heal<br>
to heal custom extended attribute.<br>
In case of gnfs server if hashed subvol has not
found based on<br>
loc then wind a call on all subvol to update
xattr.<br>
<br>
Fix: 1) Save MDS subvol on inode ctx<br>
2) Check if mds subvol is present on inode ctx<br>
3) If mds subvol is down then call unwind with error
ENOTCONN and if it is up<br>
then set new xattr "GF_DHT_XATTR_MDS" to -1 and
wind a call on other<br>
subvol.<br>
4) If setxattr fop is successful on non-mds subvol
then increment the value of<br>
internal xattr to +1<br>
5) At the time of directory_lookup check the value
of new xattr GF_DHT_XATTR_MDS<br>
6) If value is not 0 in dht_lookup_dir_cbk(other
cbk) functions then call heal<br>
function to heal user xattr<br>
7) syncop_setxattr on hashed_subvol to reset the
value of xattr to 0<br>
if heal is successful on all subvol.<br>
<br>
Test : To reproduce the issue followed below steps<br>
1) Create a distributed volume and create mount point<br>
2) Create some directory from mount point mkdir
tmp{1..5}<br>
3) Kill any one brick from the volume<br>
4) Set extended attribute from mount point on
directory<br>
setfattr -n user.foo -v "abc" ./tmp{1..5}<br>
It will throw error " Transport End point is not
connected "<br>
for those hashed subvol is down<br>
5) Start volume with force option to start brick
process<br>
6) Execute getfattr command on mount point for
directory<br>
7) Check extended attribute on brick<br>
getfattr -n user.foo
<volume-location>/tmp{1..5}<br>
It shows correct value for directories for those<br>
xattr fop were executed successfully.<br>
<br>
Note: The patch will resolve xattr healing problem only for
fuse mount<br>
not for nfs mount.<br>
<br>
BUG: 1371806<br>
Signed-off-by: Mohit Agrawal <<a
href="mailto:moagrawa@redhat.com" moz-do-not-send="true">moagrawa@redhat.com</a>><br>
<br>
Change-Id: I4eb137eace24a8cb796712b742f1d177a65343d5<br>
<br>
</div>
<br>
<fieldset class="mimeAttachmentHeader"></fieldset>
<br>
<pre wrap="">_______________________________________________
Gluster-devel mailing list
<a class="moz-txt-link-abbreviated" href="mailto:Gluster-devel@gluster.org">Gluster-devel@gluster.org</a>
<a class="moz-txt-link-freetext" href="http://lists.gluster.org/mailman/listinfo/gluster-devel">http://lists.gluster.org/mailman/listinfo/gluster-devel</a></pre>
</blockquote>
<br>
</body>
</html>