BPM works very slow

Hello Experts,

We have documentum on RHEL 6.1. The method server always had been throwing error with message,

Exception invoking com.documentum.bpm.rtutil.JMSHealthCheckMethod.

ClientAbortException: java.net.SocketException: Broken pipe

https://community.emc.com/message/679295#679295

After making some research I come to this point that this error occurs when server sends some data to client but the client closes communication before all information has been sent. So I decided to increase method time out interval for JMS health checker.

Now the result is that the error disappear but the auto tasks are executed very slowly by the BPM engine. Each auto activity takes at least 30 seconds.

I have decreased the values but again no luck. Please share your knowldge on this issue.

Find more posts tagged with

Comments

PanfilovAB

create a thread dump of JMS JVM and post it here

PanfilovAB

A week a go I got the same issue, the root cause was:

the JMSHeathChecker method has a reasonable short timeout settings, so time required to connnect to repository makes sense, to prevent slow session construction to the following things:

1. check dns setting and dns response time (actually dns is always not your friend and I would like to suggest to disable it and use /etc/hosts file only)

2. read esg114371 (it's about lifecycle but not pay attention - the root cause is the same), emc suggests to set DEVRANDOM=/dev/urandom environment variable before start CS, also you can run /sbin/rngd -f /dev/urandom -o /dev/random -f -t 1 (the better way is to place it in /etc/inittab or upstart for RHEL6)