When attempting to "stop" the protocol, usually during a server shutdown, the timeout value is set to 20 minutes. If replication is busy it can appear that the server is hung. The default timeout should be not be 20 minutes, and it should be configurable.
Added new config attribute: nsds5ReplicaProtocolTimeout
This can be added the cn=replica entry, sort of like a global setting. And/or you can add it to the repl agmt - basically creating individual timeouts.
Is this related to https://fedorahosted.org/389/ticket/560 ?
This is definitely a duplicate of https://fedorahosted.org/389/ticket/546
As for ticket 560, I don't think it's the exact same issue. They say they get a hang in the midst of doing delete operations, and then the shutdown also hangs. This ticket is only for server shutdown. So it could help with the shutdown part of their issue, but not for the random hang. Of course a full stack trace would be more revealing.
git merge ticket558
ldap/schema/01core389.ldif | 5 +-
ldap/servers/plugins/replication/repl5.h | 6 ++
ldap/servers/plugins/replication/repl5_agmt.c | 20 +++++++
.../servers/plugins/replication/repl5_connection.c | 2 +-
.../plugins/replication/repl5_inc_protocol.c | 25 +-------
.../plugins/replication/repl5_prot_private.h | 1 +
ldap/servers/plugins/replication/repl5_protocol.c | 60 +++++---------------
ldap/servers/plugins/replication/repl5_replica.c | 12 ++++
.../plugins/replication/repl5_tot_protocol.c | 5 +-
ldap/servers/plugins/replication/repl_globals.c | 1 +
10 files changed, 64 insertions(+), 73 deletions(-)
git push origin master
Counting objects: 33, done.
Delta compression using up to 4 threads.
Compressing objects: 100% (17/17), done.
Writing objects: 100% (17/17), 2.79 KiB, done.
Total 17 (delta 15), reused 0 (delta 0)
5285304..d63ff3d master -> master
Ticket has been cloned to Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=918714
Metadata Update from @nkinder:
- Issue assigned to mreynolds
- Issue set to the milestone: 1.3.1
to comment on this ticket.