Apache coredump on Stop

[Chris J. Solarczyk]

I just installed the latest version of Apache on an L-Class running HPUX 11. Apache starts up OK but when I stop the web server it produces a core dump. The stop seems to be successful but I'm left with an ugly core file on disk. Any ideas???

hpuxwsAPACHE A.2.0.55.02 HP-UX Apache-based Web Server

# ./apachectl stop
./apachectl[188]: 13683 Memory fault(coredump)

[Alex Glennie]

post the output of the command "file core" would give us the name of what process caused the core.

patch levels upto date as per apache recommendations ? 11.00 or 11.11 ?

if reproduceable anything eating resources ? maybe use glance to check ?

[Chris J. Solarczyk]

# file core
core: core file from 'httpd' - received SIGSEGV

# what core
core:
HP-UX libisamstub.sl 19991217 (135120) B3907DB/B3909DB B.11.01.11
fs_amod.s $Revision: 1.9.1.1 $
libcl.sl version B.11.01.04 - Jul 10, 2000
HP-UX libm shared PA1.1 C math library 980423 (163456) UX 11.0
STREAMS: Revision: B.11.00 $Date: 99/07/07 14:00:00 $
Pthread Interfaces
libpthread HP-UX (Release B.11.00): Tue Aug 31 15:35:32 PDT 1999
$ PATCH/11.00:PHCO_22076 Aug 16 2000 16:53:50 $
SMART_BIND
92453-07 dld dld dld.sl B.11.13 990903

[Alex Glennie]

Chris best I can offer is check kernel tuneables http://www.hp.com/products1/unix/webservers/apache/techtips/hpapache_20_perf.html

that said it maybe we are seeing a known feature that may well be patcheable ... depends on how recent patch levels are ?

If I remember I'll check when I'm back in the office tomorrow.

[Mark Greene_1]

Does it do the same thing when you specify the fully qualifed path instead of doing the dot-slash reference?

mark

[Chris J. Solarczyk]

Yes, same thing happens without the ./

I installed the following patch but it didn't seem to fix the problem. Didn't see any other patch requirements.

PHSS_33032 1.0 ld(1) and linker tools cumulative patch

[Alex Glennie]

Chris,

yes I looked at the known issues that fixed and none matched , the libisam.sl might be worth a look at as may if gdb is installed producing a back trace .... (bt) advise use the gui.

I'll see what I can find tomorrow ... glance show anything wrt memory etc ?

[Chris J. Solarczyk]

glance looks OK. The server has 4GB of RAM with about 2.5GB free so not much memory pressure.

[A. Clay Stephenson]

It is barely possible that the memory violation is occurring because maxssiz is set too small especially if maxssiz is still at the default of 8MB. If so, increase it to 32MB but no more than 64MB. The stack would only need to be larger than about 32MB on very poorly written code.

Plan B. Download and install Apache 2.2.0 from www.apache.org. It's a very easy build.

[Chris J. Solarczyk]

maxssiz is set to 100MB. This server also runs Oracle which may explain why it's set so high.

I'm just surprised that HP's Apache would have these problems right out of the box. I installed the latest version on another server and the same thing happens with the 'stop' command. It coud be a patch problem since these servers are behind in patches. I may try to download the source and build it myself.