Does anyone know of a KB that explains what to do to shut down a source database (for unrelated reasons) despite the presence of any ongoing replication issues?
It sounds like this has been an issue for other customers as well (eg here: https://community.progress.com/community_groups/openedge_rdbms/f/18/t/15113 )
We are trying to shut down a database, but 30 mins after requesting an emergency shutdown, the database is still waiting on fathom for whatever strange reason. In the database logs we see the following, even ten and twenty minutes after doing a "-F" shutdown:
[2017/02/17@09:30:00.766-0700] P-5643 T-1 I REPL : (10717) The Fathom Replication Utility cannot connect to database /xxx/mydatabase.
Any pointers would be appreciated. The need to shut down a database is not ideal, Fathom replication issues are bad enough, and trying to shut down a (source) database while Fathom is misbehaving seems virtually impossible. Good thing that power button still worked.
kill the rpserver process. It may crash the DB but you're shutting it down anyways.
Please next time post opsys and OE version info.
Paul
Thanks for the tip. We run HP-UX version 11.3.
Is this something that every OpenEdge DBA should know? IE. Is it well-known to a DBA that this "rpserver" could potentially prevent a database shutdown? When I had looked for them, I didn't easily find any KB articles that indicated that Fathom replication trouble can also interfere with database shutdown (I only found the link to the forums posting). But it sounds like you are speaking from experience.
I'm not actually a OpenEdge DBA, just a kibitzer (a very confused one).
Also, is it possible that killing rpserver cause more problems than it may solve (ie. our DBA was nervous about having to re-seed replication.)
Thanks again, David
Killing replication server does not mean re-seed replication. When you stop RPLS, you can restart it later without needing to re-seed by dsrutil db -C restart server
If you want to stop replication server, I'd suggest following sequence
1) dsrutil db -C terminate server
if that does not do, then (would wait minute or 2)
2) proshut db -C disconnect usernumofRPLS (it is usually 5, but you can get that off proshut db -C list | grep RPLS)
if that does not do, then (would wait minute or 2)
3) kill -8 PID(of_RPLS)