Admin Console incorrect instance status and deployment timeout issues

Question

Dear All
I am using weblogic 10.3.6 on two AIX, 3 instances on each and 2 instance are clustered. I found that when I start all instances, admin console will become unstable. Such as when starting 1 instance, will affect another started instance to be 'STARTING'.
https://www.dropbox.com/s/o225w9w74mzn33d/weblogic_admin.PNG
Also, when all instances are started, I could not complete deployment for EAR. Timeout will occur will activating change.
https://www.dropbox.com/s/rjhae9haw06926d/weblogic_timeout.PNG
Any idea ?
Many thanks
Chris

Answer 1

I assume the admin interface is running on the same machine as the server instance.
By default the Web Server instances use a directory in /tmp as a scratch/temp dir (something like /tmp/instance_name.hash/, e.g. /tmp/www.qweqwe.com-7f54fa82). One of the files stored there is the pid file. The pid file contains the pid of the running instance and is used by the admin interface (and the stop script, and the Admin agent I think) to keep track of which webservd "owns" which instance, etc. Unfortunately the /tmp directory is kept clean by the OS, and sometimes the OS daemon will remove the Web Server's directory and its associated files. When this happens the Web Server will (usually) continue working as though nothing were wrong, but the admin interface (and stop script, etc) will show that the instances are "Not Running."
For this reason one of the first things I do with a new Web Server instance is move its directory into the Web Server install location (something like /export/WS7/instance_name.hash/) and update the server.xml temp-path element to reflect the new location. Note that Web Server is very specific about how it wants the permissions set on this directory. It will report in its errors log if you get the permissions wrong.