Frezzing DB Server on HPUX

[jerrys]

Got a interesting DB server issue here. For months, I have a database that has been running with no problems, then suddenly the server freezes each day and at various times for no apparent reason. Users who try to attach are frozen. Root tries to attach or use promon, freezes right up. Proshut also freezes. The only thing I can do is issue a kill -15 against the _mprosrv process which (most of the time) kills off the other processes except the user _progres sessions. The user sessions have to be killed seperately (of course). The error the first user gets when attaching to the database (occasionally) is that they cannot write to the logfile. Error# 13. Of course, the log file is there, permissions are rwx-r-r and owned by root. The servers was started by root on batch after the nightly backup.

After killing the database and restarting the server, everything seems OK for the rest of the day (most of the time). Once in a while, it will freeze again later in the day.

Signed -

Frustrated!

 

[Norman Biggar]

We are experiencing a similar problem, though perhaps with less frequency.

We are running 8.3B on a sun solaris unix box, users accessing via appserver.

Twice in the last few months, our system has "frozen". No obvious reason why. Log files for databases, appserver all check OK.

I know there is a 2GB for files on UNIX. Is there perhaps a lower maximum for database .lg files?

The only solution Progress have suggested is implementing "socket keepalive" (see knowledgebase for more info).

Our UNIX team seem to be very against this, so if anyone has a solution I'd also be delighted to hear it.