SUMMARY - memory fault (core dump)

From: <ador_sunga_at_tdb.gov.sg>
Date: Mon, 19 Jul 1999 18:11:26 +0800

I received only one response and its from Mohan. However, this doen't fix my
problem. Attached is a reply from him:

 Look at your code very carefully, there could be a bug in the
 compiler/linker itself.
 If you have just upgraded to v4.0D, then check if you have ANY
 variable defined "uname", since the linker does not catch it as
 a reserverd word. This has been fixed in the latest patch kit.
 The other thing is sometimes look at your string allocations
 /deallocations routies carefully.
 cheers

Compilation of web server is simply running the executable files (is a binary
files that comes I believe during the installation of web server 4.0.7). So,
basically there is no code to modify...

I also did some test by increasing the IPC (shm-max) and virtual memory
(max-vas) parameter (from 1GB to 2GB) but this also does'nt fix the problem..
It still producing the error message "memory fault (core dump) during
compilation.

Right now, Our local Compaq support suggested to install a new fresh 4.0D with
the new patch release for the threadsafe_lib_v40d. I forgot to mentioned in my
original question that prior to this problem, I having problem with the
threadsafe libraries with the following error during the compilation "Internal
AIO consistency error: Cannot use libaio_raw with threads...." which I found
out that its a bug from 4.0D. Putting this patch manually fixed the bug,
however, it genned another error which is this memory fault (coredump). I
also tried to re-create the libraries, install the patches and rebuild the
kernel but still does'nt solve the problem......

For now, we are trying to evaluate what would be the impact if we will go ahead
with the suggestion of Compaq which is to install a new 4.0D DIRECTLY.
Basically what we're trying to achieve here is to simulate the following
upgrade:

Original server To New Server
Web server ver 2.01 Web server ver 4.0.7
ORACLE 7.3.3 Oracle ver 8.0.4
OS from 4.0B DUNIX 4.0D
and to simulate as well all the application that seats on this server if its
have some implication on this upgrade. This includes ofcourse of restoring the
file system from original server to new server.

This upgrades includes swapping of server so that we could have a transparent
upgrade with minumun downtime. Another reason is that one requirement of web
server upgrade is to install the new version and then run a utility that will
convert the data from ver 2.0.1(from 4.0B and was restored to new server, then
from 4.0B it was upgraded to 4.0D) to version 4.0.7. Basically we dont the
implication if we will jump directly to 4.0D (new server)....

By the way, my apologies for this a bit off topic.....Im just thinking that
someone out there experienced the same problem, or if there is something
missing about my settings and configuration.

I still appreciate if you have some thoughts about it....

I'll take this as my initial summary, I'll make an addendum whether this has
been fixed or not.

One again Thanks to Mohan and Lisa for theirs inputs,

Ador
**********************************************************************************************************************
Original Questions:

On Thu, 15 Jul 1999 ador_sunga_at_tdb.gov.sg wrote:

> Hi Managers,
>
> Is there anyone who experienced this error msg "memory fault (core dump)" when
> relinking/compiling web server 4.0.7. I tried several test and changes on my
> sysconfigtab, particularly to
>
> ipc:
> shm-max=4194304
> shm-seg=256
> sem-mni=256
>
> but to no avail. I also tried to start the web server but got the same
error > message (w/ coredump also). Below is my sysconfig parameters using the
> /sbin/sysconfig -q ipc command:
>
> msg-max = 8192
> msg-mnb = 16384
> msg-mni = 64
> msg-tql = 40
> shm-max = 4194304
> shm-min = 1
> shm-mni = 128
> shm-seg = 32
> sem-mni = 16
> sem-msl = 25
> sem-opm = 10
> sem-ume = 10
> sem-vmx = 32767
> sem-aem = 16384
> num-of-sems = 60
> max-kernel-ports = 22487
> port-hash-max-num = 1124350
> port-reserved-max-num = 22487
> set-max-num = 1029
> ssm-threshold = 0
> ssm-enable-core-dump = 1
>
> I already passed the core dump to the oracle peope but they say the cannot do
> anything about it, they did'nt even give any recommedation.....I also passed
> this to our
> local Compaq support but they have'nt give any feedback yet.
>
> So, I'm just wondering if someone already experienced this. The server that
> Im using
> was recently upgraded from 4.B to 4.0D. Any thoughts is hightly appreciated.
>
> Thanks,
>
> Ador
>
>
>
>
>
Received on Mon Jul 19 1999 - 10:15:32 NZST

This archive was generated by hypermail 2.4.0 : Wed Nov 08 2023 - 11:53:39 NZDT