stopping writes to webspeed server logfile

[danishgambit]

Afternoon 'talkers,

OE10.2A on redhat.

we've currently got a situation were the writes to a webspeed server logfile are continuing despite the fact we've shut the webspeed brokers. Just wondering if anyone has come across a similar situation ? If so, how did you deal with it.

TIA,

DG.

 

[RealHeavyDude]

AFAIK, the server log file is shared and per default when you start the broker, the agents it starts will always append their messages. IIRC, you can specifiy the number of logfiles and a logfile threshold to change this behavior.

But, in your case I suppose, although you've stopped the WebSpeed broker, one or more of the agents is still busy or "hangs". The only remedy that I know is to kill this process the hard way. Just pf -ef and grep for java and you'll find it ...


Regards, RealHeavyDude.

 

[danishgambit]

Afternoon 'Dude,

thanks for the info - killed the process and that did the job. Yep, I'd come across the logfile threshold and number of logfiles while looking into the problem and hope to use these to prevent this happening again.

Regards,

DG.

 

[RealHeavyDude]

From what I know - nothing in IT is perfect - it does sometimes happen that WebSpeed or AppServer agents run wild and are not stopped when you stop the broker. Until now I've found no other solution but to let them die hard (with a vengeance:awink using either kill -9 or the windoze task manager.

Regards, RealHeavyDude.

 

[rstanciu]

If you kill -9 the _progres process the database server will log a a-normal situation. He sais ... ooooooo-la-la
I have a system SIGTERM error and I will loose data. In this case The database server will shutdown a-normaly.
NEVER USE THE: kill -9/-15 !!!!!!

The command to use is: proshut [database] -C disconnect [user num]
to known the user-num ... proshut [database]->>> option 1
or wtmban -trim

 

[RealHeavyDude]

To be honest, I forgot to mention that the "kill -9" medicine should only be used if you have gracefully stopped the concerned databases.

Regards, RealHeavyDude.

 

[StuartT]

kill -15 is usually the safe way to go as this will shut the processes down as cleanly as possible, if you do it on a server process then check the log file you will see that it initiates a PROSHUT

 

[rstanciu]

kill -15 works safe only if the user who initate the wtbman -start is the same user who kill.

From KBase: ID: P50670

kill -1 (SIGHUP) Terminal hang up signal.
kill -2 (SIGINT) Same as doing a Ctrl+C, terminal interrupt.
kill -8 (SIGQUIT) Floating Point Exception. This might leave shared
memory in a corrupted status, like kill -9.
kill -9 (SIGKILL) Truly kills the process. If issued against a client,
the database might shut down abnormally. If issued
against the broker, the effect will be the same as a
power outage; in addition to that, semaphores and shared
memory segments will not be removed from the OS memory,
so they will have to be manually removed. 

Alternatively, just to be on the safe side, after a
kill -9 to the broker, the machine should be rebooted.