Kingsley
2016-Jul-15 13:19 UTC
[Gluster-users] lingering <gfid:*> entries in volume heal, gluster 3.6.3
On Fri, 2016-07-15 at 18:38 +0530, Ravishankar N wrote:> On 07/15/2016 06:05 PM, Kingsley wrote: > > chomp (my @output=`getfattr $path`); > > > Could you try with `getfattr -d -m. -e hex $path` ?Sure. I'm not really sure what I should be seeing, so I've uploaded the full output here as it's rather large for a group mailing list IMO: http://gluster.dogwind.com/files/gfid-getfattr.txt The output is the same for each of the stayed-up bricks. FWIW, many of the entries have the same values: [root at gluster2b-1 ~]# sort gfids | perl gfid-to-fattr.pl 2> /dev/null | grep trusted.afr | sort | uniq -c 73 trusted.afr.callrec-client-0=0x000000000000000000000000 73 trusted.afr.callrec-client-1=0x000000000000000000000000 25 trusted.afr.callrec-client-2=0x000000020000000300000000 5 trusted.afr.callrec-client-2=0x000000040000000300000000 33 trusted.afr.callrec-client-2=0x000000050000000300000000 1 trusted.afr.callrec-client-2=0x00000005000000ea00000000 1 trusted.afr.callrec-client-2=0x00000005000001f900000000 1 trusted.afr.callrec-client-2=0x000000050000021400000000 3 trusted.afr.callrec-client-2=0x000000060000000300000000 2 trusted.afr.callrec-client-2=0x000000080000000300000000 1 trusted.afr.callrec-client-2=0x000000090000000300000000 1 trusted.afr.callrec-client-2=0x0000000e0000000300000000 73 trusted.afr.callrec-client-3=0x000000000000000000000000 73 trusted.afr.dirty=0x000000000000000000000000 Cheers, Kingsley.
Ravishankar N
2016-Jul-15 15:01 UTC
[Gluster-users] lingering <gfid:*> entries in volume heal, gluster 3.6.3
On 07/15/2016 06:49 PM, Kingsley wrote:> On Fri, 2016-07-15 at 18:38 +0530, Ravishankar N wrote: >> On 07/15/2016 06:05 PM, Kingsley wrote: >>> chomp (my @output=`getfattr $path`); >> >> Could you try with `getfattr -d -m. -e hex $path` ? > Sure. I'm not really sure what I should be seeing, so I've uploaded the > full output here as it's rather large for a group mailing list IMO: > > http://gluster.dogwind.com/files/gfid-getfattr.txt > > The output is the same for each of the stayed-up bricks.If it is the same in all 3 bricks, the the xattrs of the file indicate heal is pending on the 3rd brick (trusted.afr.callrec-client-2) which I guess was the one that was down. Selfheal daemon should heal them ideally. But you're saying there are no messages in the shd logs corresponding to these gfids. Could you check the hardlink count (star -c%h) of these gfids (the ones corresponding to files, not directories)? It must be > 1. -Ravi> > FWIW, many of the entries have the same values: > > [root at gluster2b-1 ~]# sort gfids | perl gfid-to-fattr.pl 2> /dev/null | grep trusted.afr | sort | uniq -c > 73 trusted.afr.callrec-client-0=0x000000000000000000000000 > 73 trusted.afr.callrec-client-1=0x000000000000000000000000 > 25 trusted.afr.callrec-client-2=0x000000020000000300000000 > 5 trusted.afr.callrec-client-2=0x000000040000000300000000 > 33 trusted.afr.callrec-client-2=0x000000050000000300000000 > 1 trusted.afr.callrec-client-2=0x00000005000000ea00000000 > 1 trusted.afr.callrec-client-2=0x00000005000001f900000000 > 1 trusted.afr.callrec-client-2=0x000000050000021400000000 > 3 trusted.afr.callrec-client-2=0x000000060000000300000000 > 2 trusted.afr.callrec-client-2=0x000000080000000300000000 > 1 trusted.afr.callrec-client-2=0x000000090000000300000000 > 1 trusted.afr.callrec-client-2=0x0000000e0000000300000000 > 73 trusted.afr.callrec-client-3=0x000000000000000000000000 > 73 trusted.afr.dirty=0x000000000000000000000000 > > Cheers, > Kingsley. >