geo-replication - OSError: [Errno 1] Operation not permitted - failing with socket files?

[Davide Obbi <davide.obbi@xxxxxxxxxxx>]

Hi,

i am trying to understand why georeplciation during "History Crawl" starts failing on each of the three bricks, one after the other. I have enabled DEBUG for all the logs configurable by the geo-replication command.

Running glusterfs v4.16 the behaviour is as follow:

- The "History Crawl" worked fine for about one hr, it actually replicated some files and folders albeit most of them looks empty

- at some point it starts becoming faulty, try to start on another brick, faulty and so on

- in the logs, Python exception above mentioned is raised:

[2019-03-17 18:52:49.565040] E [syncdutils(worker /var/lib/heketi/mounts/vg_b088aec908c959c75674e01fb8598c21/brick_f90f425ecb89c3eec6ef2ef4a2f0a973/brick):332:log_raise_exception] <top>: FAIL:                                                                              
Traceback (most recent call last):
  File "/usr/libexec/glusterfs/python/syncdaemon/gsyncd.py", line 311, in main
    func(args)
  File "/usr/libexec/glusterfs/python/syncdaemon/subcmds.py", line 72, in subcmd_worker
    local.service_loop(remote)
  File "/usr/libexec/glusterfs/python/syncdaemon/resource.py", line 1291, in service_loop
    g3.crawlwrap(_oneshot_=True)
  File "/usr/libexec/glusterfs/python/syncdaemon/master.py", line 615, in crawlwrap
    self.crawl()
  File "/usr/libexec/glusterfs/python/syncdaemon/master.py", line 1569, in crawl
    self.changelogs_batch_process(changes)
  File "/usr/libexec/glusterfs/python/syncdaemon/master.py", line 1469, in changelogs_batch_process
    self.process(batch)
  File "/usr/libexec/glusterfs/python/syncdaemon/master.py", line 1304, in process
    self.process_change(change, done, retry)
  File "/usr/libexec/glusterfs/python/syncdaemon/master.py", line 1203, in process_change
    failures = self.slave.server.entry_ops(entries)
  File "/usr/libexec/glusterfs/python/syncdaemon/repce.py", line 216, in __call__
    return self.ins(self.meth, *a)
  File "/usr/libexec/glusterfs/python/syncdaemon/repce.py", line 198, in __call__
    raise res
OSError: [Errno 1] Operation not permitted

- The operation before the exception:

[2019-03-17 18:52:49.545103] D [master(worker /var/lib/heketi/mounts/vg_b088aec908c959c75674e01fb8598c21/brick_f90f425ecb89c3eec6ef2ef4a2f0a973/brick):1186:process_change] _GMaster: entries: [{'uid': 7575, 'gfid': 'e1ad7c98-f32a-4e48-9902-cc75840de7c3', 'gid': 100, 'mode'
: 49536, 'entry': '.gfid/5219e4b8-a1f3-4a4e-b9c7-c9b129abe671/.control_f7c33270dc9db9234d005406a13deb4375459715.6lvofzOuVnfAwOwY', 'op': 'MKNOD'}, {'gfid': 'e1ad7c98-f32a-4e48-9902-cc75840de7c3', 'entry': '.gfid/5219e4b8-a1f3-4a4e-b9c7-c9b129abe671/.control_f7c33270dc9db9
234d005406a13deb4375459715', 'stat': {'atime': 1552661403.3846507, 'gid': 100, 'mtime': 1552661403.3846507, 'uid': 7575, 'mode': 49536}, 'link': None, 'op': 'LINK'}, {'gfid': 'e1ad7c98-f32a-4e48-9902-cc75840de7c3', 'entry': '.gfid/5219e4b8-a1f3-4a4e-b9c7-c9b129abe671/.con
trol_f7c33270dc9db9234d005406a13deb4375459715.6lvofzOuVnfAwOwY', 'op': 'UNLINK'}]
[2019-03-17 18:52:49.548614] D [repce(worker /var/lib/heketi/mounts/vg_b088aec908c959c75674e01fb8598c21/brick_f90f425ecb89c3eec6ef2ef4a2f0a973/brick):179:push] RepceClient: call 56917:140179359156032:1552848769.55 entry_ops([{'uid': 7575, 'gfid': 'e1ad7c98-f32a-4e48-9902-
cc75840de7c3', 'gid': 100, 'mode': 49536, 'entry': '.gfid/5219e4b8-a1f3-4a4e-b9c7-c9b129abe671/.control_f7c33270dc9db9234d005406a13deb4375459715.6lvofzOuVnfAwOwY', 'op': 'MKNOD'}, {'gfid': 'e1ad7c98-f32a-4e48-9902-cc75840de7c3', 'entry': '.gfid/5219e4b8-a1f3-4a4e-b9c7-c9b
129abe671/.control_f7c33270dc9db9234d005406a13deb4375459715', 'stat': {'atime': 1552661403.3846507, 'gid': 100, 'mtime': 1552661403.3846507, 'uid': 7575, 'mode': 49536}, 'link': None, 'op': 'LINK'}, {'gfid': 'e1ad7c98-f32a-4e48-9902-cc75840de7c3', 'entry': '.gfid/5219e4b8
-a1f3-4a4e-b9c7-c9b129abe671/.control_f7c33270dc9db9234d005406a13deb4375459715.6lvofzOuVnfAwOwY', 'op': 'UNLINK'}],) ...

- The gfid highlighted, is pointing to these control files which are "unix sockets" as per below:

rw-------  2 pippo users     0 Mar 14 16:32 .control_31c3a99664c1f956f949311e58434037e6a52d22
srw-------  2 pippo users     0 Mar 14 16:33 .control_a9b82937042529bca677b9f43eba9eb02ca7c5ee
srw-------  2 pippo users     0 Mar 14 16:32 .control_f429221460d52570066d9f25521011fe7e081cf5
srw-------  2 pippo users     0 Mar 15 15:50 .control_f7c33270dc9db9234d005406a13deb4375459715

So it seems geo-replicaiton should be at least skipping such file rather than raising an exception? Am i the first experiencing this behaviour?

thanks in advance

Davide

_______________________________________________
Gluster-users mailing list
Gluster-users@xxxxxxxxxxx
https://lists.gluster.org/mailman/listinfo/gluster-users