Link to home
Start Free TrialLog in
Avatar of nswartz
nswartz

asked on

Unexpected signal : 4 in JVM running under Websphere 5.1.1

When an MDB is invoke a few tens of times, the JVM running the WebSphere application server dies with a signal 4. Platform is HPUX, WebSphere 5.1.1, JDK is 1.4.2.09a. Error messages in stdout are:

Unexpected Signal : 4 occurred at PC=0x77E6C6A0
Function=[Unknown.]
Library=/app1/was/java/jre/lib/PA_RISC2.0/server/libjvm.sl
NOTE: We are unable to locate the function name symbol for the error
      just occurred. Please refer to release documentation for possible
      reason and solutions.
Current Java thread:
"MessageListenerThreadPool : 0" daemon prio=5 tid=02827958 nid=54 lwp_id=8278556 runnable [0x00001000..0x2c48f4f0]

Stderr is:
/usr/lib/dld.sl: Unresolved symbol: $ARGV (data)  from /usr/lib/libcl.2
/usr/lib/dld.sl: Unresolved symbol: $RECOVER_START (data)  from /usr/lib/libcl.2
/usr/lib/dld.sl: Unresolved symbol: $RECOVER_END (data)  from /usr/lib/libcl.2
/usr/lib/dld.sl: Unresolved symbol: _SYSTEM_ID (data)  from /usr/lib/libcl.2
/usr/lib/dld.sl: Unresolved symbol: __text_start (data)  from /usr/lib/libcl.2
/usr/lib/dld.sl: Unresolved symbol: __dld_loc (data)  from /usr/lib/libcl.2
/usr/lib/dld.sl: Unresolved symbol: _SYSTEM_ID (data)  from /usr/lib/libcl.2
/usr/lib/dld.sl: Unresolved module for symbol: U_STACK_TRACE (code)  from /app1/was/java/jre/lib/PA_RISC2.0/server/libjvm.sl
Avatar of CEHJ
CEHJ
Flag of United Kingdom of Great Britain and Northern Ireland image

Sounds like a bug. Try at least a minor version upgrade
Avatar of nswartz
nswartz

ASKER

We just upgraded to this version from 1.4.2.03. This is the version recommended by IBM support for this release of Websphere. I think this is the highest 1.4.2 on this platform.
Try reinstalling
Avatar of nswartz

ASKER

It did not work with 1.4.2.03, we already re-installed with 1.4.2.09a to no effect, it would not likely help to re-install the 1.4.2.09a that we just installed. I was hoping someone has already seen this problem.
It could be a shared library problem, a known bug, or something else?
doubt reinstalling will help, would just waste youtr time.
Thats a VM crashing, due to a bug. Whether its known or not I'm not sure.
You see to have missing/incorrect binaries
Are you doing an install or just copying it in place?
This sort of error is what happens when you just copy the files in. We do that with identical systems because it's faster to replicate, but if the library versions are different (such as when we got new machines running Fedora Core 4, while the old ones are running Red Hat EE 3, we had to do an actual install, and then could replicate the installed version to all the identical machines.

Not saying this is what happened in your case, but it's nice to know.

Another possibility: make sure that you are running the version you think you're running. If you are getting an old version of java in the path, or classes from an old install, again this is the kind of thing that happens.
Avatar of nswartz

ASKER

We run the same test over and over. After the 20th time or so, we get the crash. Hard to see how the binaries could be incorrect.  Any suggestions how to look for this? We are using ldd to trace things down, but the libs all seem old.
This was working last week with WebSphere 5.1
> This was working last week with WebSphere 5.1

so you've recently upgrade?
Might be worth contacting WS support and see what they have to say
Avatar of nswartz

ASKER

We finally got hold of IBM Level 3 support (offshore!) and they suggested turning off JIT. That worked, but slowed everything down by 300%! We will run it that way since the app is not time critical. IBM will get a new patch from HP sometime to fix it.
Thanks anyway for the help.
ASKER CERTIFIED SOLUTION
Avatar of CetusMOD
CetusMOD
Flag of Netherlands image

Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial