replication_mgr "partial transfer due to error"

[Gr-R]

check the rsync.log for information. It could be the snapshot is filling up.

[rmoulton]

Strange. On the source system 'gverc', the timestamp on rsync.log is 2014-07-29 16:18. I don't see anything logged since then.

On the destination system 'dunkel', the rsync.log is full of entries of this sort:

2014/08/08 00:53:06 [4452] connect from gverc.gcc.biostat.washington.edu (128.208.235.147)
2014/08/08 00:53:06 [4452] module-list request from gverc.gcc.biostat.washington.edu (128.208.235.147)
2014/08/08 00:53:06 [4453] connect from gverc.gcc.biostat.washington.edu (128.208.235.147)
2014/08/08 00:53:06 [4453] module-list request from gverc.gcc.biostat.washington.edu (128.208.235.147)
2014/08/08 00:53:26 [4622] connect from gverc.gcc.biostat.washington.edu (128.208.235.147)
2014/08/08 00:53:26 [4622] rsync to GCC_archive/ from gcc-repl@gverc.gcc.biostat.washington.edu (128.208.235.147)
2014/08/07 22:53:26 [4622] receiving file list
2014/08/07 22:54:54 [4622] IO error encountered -- skipping file deletion
2014/08/08 01:02:15 [13490] connect from gverc.gcc.biostat.washington.edu (128.208.235.147)
2014/08/07 16:02:15 [13490] module-list request from gverc.gcc.biostat.washington.edu (128.208.235.147)
2014/08/08 01:02:15 [13491] connect from gverc.gcc.biostat.washington.edu (128.208.235.147)
2014/08/07 16:02:15 [13491] module-list request from gverc.gcc.biostat.washington.edu (128.208.235.147)
2014/08/08 01:02:31 [13555] connect from gverc.gcc.biostat.washington.edu (128.208.235.147)
2014/08/07 16:02:32 [13555] rsync to GCC_archive/ from gcc-repl@gverc.gcc.biostat.washington.edu (128.208.235.147)
2014/08/07 23:02:32 [13555] receiving file list
2014/08/07 23:03:52 [13555] IO error encountered -- skipping file deletion

[Gr-R]

with IO errors, check connection(latency) as well as raid for issues.

[rmoulton]

Okay, but how/where/what to check exactly? And is the lack of recent entries in the source system rsync.log expected?

[Gr-R]

Check destination RAID for any issues. Also verify connection between units is stable.
If the destination fails to accept the data, nothing would be logged at the source.