Hung/Orphan Process in SDE

[JoeBorgione]

From time to time I find hung processes attached to my sde instance(s).  With a recent migration to v10, this has become more a rule than an exception.

For example, I found 3 connections dated yesterday just this morning.  Usually I can use sde-o kill <pid> blah blah blah.  Lately however, this approach won't allow me to delete the process so I stop and restart the service.

In the 9 years or so of administering SDE, I have never had my user login hang a process either locally or remotely.    I've set tcpkeepalive = true and it dosen't seem to do the trick.  Can anyone shed some light as to what I should be doing or what the remote users should be doing to correct this issue?

Also, when I stop the service I get some error that windows can't stop it, although it actually does; I start and all the hung/orphan process are gone.  Any ideas whats up with that?

Oh, ArcGis v 10 SDE SP3, Windows 2008r 64bit os (on server)

[DanMcCoy]

Hey Joe,

Did you ever figure out what was going on?

Thanks,

Dan

[JoeBorgione]

Nope.  But I cycle my services off and on more often now; that drops any that might be there.

[SachinKanaujia]

I have seen similar issues in some versions like 9.3. Generally it is advised to follow the sequence during restarts. Shutdown SDE, then Shutdown database and restart in the reverse way.

Also you can analyze these hung connections. Are these from some specific machines? What type of connections are being used?