Description of problem:
Chain-on-update logs managed-entries-plugin errors on the consumer, which are benign (see Nathan's comments below). Note: the operation itself is successfully finished.
[09/Sep/2010:17:47:26 -0700] NSMMReplicationPlugin -
multimaster_be_state_change: replica o=airius.com is coming online;
[09/Sep/2010:17:49:45 -0700] managed-entries-plugin - mep_mod_post_op:
Unable to fetch postop entry.
Comments from Nathan:
I suppose that this does not need to be a fatal message. The managed
entry plug-in is trying to fetch the postop entry to determine if it is
an origin entry. It seems that SLAPI_ENTRY_POST_OP is not set on a
consumer that chains the write operation to a master. This is fine in
this case since we have no work to do. When the master sends the chained
write over to the consumer, it will also send the changes that were
performed by the managed entries plug-in on the master. Replicated
operations are ignored by the managed entry plug-in, so the consumer
will end up being consistent with the master.
Created attachment 468082 [details]
Pushed to master. Thanks to Noriko for her review!
Counting objects: 13, done.
Delta compression using up to 2 threads.
Compressing objects: 100% (7/7), done.
Writing objects: 100% (7/7), 961 bytes, done.
Total 7 (delta 4), reused 0 (delta 0)
d9cac18..2d80c40 master -> master
Request you to please add the verification steps.
Executed mmrepl/chainonupdate testcase in the acceptance tests.
Checked the logs, no such message found.
Hence marking the bug as VERIFIED.