JVM dies/quits with DrWatson error

Question

Hi,
On a busy box, where a dozen of java applications run, JVM abruptly exited several times throught the day. The machine was scanned after hours for hardware errors; memory, disk, bus adapters - everything is in proper order. DrWatson reported an application crash and made a memory dump.
My question:
How do I proceed debbuging memory dump. Has anyone had any experience ? Also, java.exe has no symbol information and there is no (as far I know ) a debugging release of JVM? So will my research of assembly code lead somewhere ? I need to know the cause of this problem.
Application Details: It's an order routing piece, which does a lot of IO & tcp/ip. It has been in production for long time-no complains. Also, the application interacts with native middle-ware product libraries through JNI ( and that product- tibco bus- is very reliable). Rebooting the server "fixed" everything. I should rather say, we have not seen the problem reoccur.
Box: Windows 2000, fully patched, 4 CPUs, 2Gb RAM, plenty of disk space. All components and software are constantly checked and monitored via snmp, and HP open-view. No errors. Server is rebooted nigtly.
JVM:
Java(TM) 2 Runtime Environment, Standard Edition (build 1.3.1-b24)
Java HotSpot(TM) Client VM (build 1.3.1-b24, mixed mode)
thanks,
eugene
DrWatson error, see where it says FAULT -> that's the instruction that crashed the app.
Application exception occurred:
App: (pid=3060)
When: 8/4/2004 @ 14:09:33.089
Exception number: c0000005 (access violation)
State Dump for Thread Id 0xd78
eax=448bff33 ebx=00000000 ecx=6d4b4bc2 edx=00000000 esi=e761a6c4 edi=00000020
eip=6d46643f esp=08bcfd20 ebp=08bcfd38 iopl=0 nv up ei pl zr na po nc
cs=001b ss=0023 ds=0023 es=0023 fs=0038 gs=0000 efl=00000246
function: <nosymbols>
6d466422 7705 ja JVM_FindSignal+0x1c784 (6d46ed29)
6d466424 89480c mov [eax+0xc],ecx ds:450d9e19=????????
6d466427 eb02 jmp JVM_FindSignal+0x17186 (6d46972b)
6d466429 33db xor ebx,ebx
6d46642b 85db test ebx,ebx
6d46642d 0f858c000000 jne JVM_FindSignal+0x13f1a (6d4664bf)
6d466433 8b45fc mov eax,[ebp+0xfc] ss:093e9c1e=????????
6d466436 8b4004 mov eax,[eax+0x4] ds:450d9e19=????????
6d466439 8d4808 lea ecx,[eax+0x8] ds:450d9e19=????????
6d46643c 8b4008 mov eax,[eax+0x8] ds:450d9e19=????????
FAULT ->6d46643f ff9088000000 call dword ptr [eax+0x88] ds:448bffbb=????????
6d466445 85c0 test eax,eax
6d466447 7416 jz JVM_FindSignal+0x179ba (6d469f5f)
6d466449 3b35fc0f4e6d cmp esi,[6d4e0ffc] ds:6d4e0ffc=00004000
6d46644f 7c0e jl JVM_FindSignal+0x1c9ba (6d46ef5f)
6d466451 8b0d440f4e6d mov ecx,[6d4e0f44] ds:6d4e0f44=007c75a0
6d466457 56 push esi
6d466458 8b01 mov eax,[ecx] ds:6d4b4bc2=448bff33
6d46645a ff5038 call dword ptr [eax+0x38] ds:450d9e19=????????
6d46645d eb57 jmp JVM_FindSignal+0x1c211 (6d46e7b6)
6d46645f 833d24094d6d00 cmp dword ptr [6d4d0924],0x0 ds:6d4d0924=00000000
6d466466 7442 jz JVM_FindSignal+0x1ca05 (6d46efaa)
----> Stack Back Trace <----

Answer 1

Hi there
Debug VM can be obtained from Sun if u have a contract with them but before raising the escalation, do try out the latest 1.3.1 VM. You are 12 versions behind. Almost 3 years worth of fixes ;)
A quick search with "JVM_FindSignal" in bug database brings up some bugs.
http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=5013339
Hope this helps.