Issue with WC server - Hangs and needs reboot.

[RajuPulavarthi]

All - We currently are on Windchill 10.1 M010 running on Redhat Linux 6.3 in a clustered environment. Occasionally we have been seeing issues with our Master server running Background Method Servers where it stops processing Queues completely. Individual Method server restart using JConsole also does not help get the Queues moving. During this time, the system load remains much higher. Only way to get around the problem has been to physically reboot the server.

Our System Administrators have found the messages like this in System logs (dmesg) "task java.exe:<process id="> blocked for more than 120 seconds".

We have opened a call with PTC's Technical support who has been trying to get to the root cause. I was just wondering if anyone has experienced the issue. If so, what did you do to fix this?

Thanks,
Raju

Raju Pulavarthi
Sr. Systems Analyst - PLM Applications
Global Information Technology

[patrickchin2000]

There are a few things you don't mention in your email:

• how much total RAM


• howmany foreground method servers


• how many CPU cores per server (each core should only be alocated 1 method server and leave 1 core for system)


• where is your database (is that tuned)


• how big are your assemblies


• are you performing additional actions (move, save as)


• how many backgrounds (publishing, replication)

you should:

• allocate min and max from xms=768 to xmx=6144 for each method server and background method server


• always have a separate background method servers for publishing with multiple workers (1 worker/10 users)


• allocate 1 foreground method server for 20 concurrent user


• check the speed of clustering versus monolithic (see attached)

that's some hints and hopefully it helpls you out.