Sun One 6.1 freeze
I am struggling with a problem on a Sun One 6.1 web server with SP4 running on top of Solaris 8 SPARC.
Without going into details I can reveale that one specific webserver instance suddenly "freezes" for 30 minutes during the day. There are other webserver instances on the machine and all of these function perfectly during the freez period.
When it "freezes", the connection queue is building up rapidly. After 30 minutes have passed the webserver starts working again like nothing has happened. The webserver error log then gives me a lot of "not connected" messages because clients have disconnected during the freeze.
During the freez there are no entries in neither the access log or the error log. The solaris syslog reports no abnormal situations. In addition I see from perfdump that the total number of active sessions are way beyond the configured limit (128).
When clients connect to the webserver, the webapplication will do backend communication using java's URLConnection before responding.
There are no indications of queueing, high load, slow response times etc when the freeze occurs.
I could perfectly understand this if 128 worker threads were stuck on some backend communication or so, but that is not the case.
So my question is:
Are there any other logs, statistices etc I can analyze ?
Other ideas to resolve this problem?
Regs,
Stephen
