Hi,
I have a cmf site that has begun restarting frequently and at random intervals for no apparent reason. This first occurred about 2 weeks ago for about 2-3 days with the instance being restarted every 15 mins at worst. This problem disappeared as quickly as it had appeared.
2-3 days ago this same problem reappeared but with the site sometimes being restarted 10 times in a 15 minute period. I can find no sequence of url calls or any common reason for this to be occurring. The only thing I could find was the following entry appearing in the event log matching the times I have witnessed the instance restarting via top.
ERROR(200) zdaemon Process 25606 terminated by signal SIGSEGV(11)
The only changes that have occurred around the time of this first problem turning up was (other than normal site operation with members/items being add/deleted): -the cpu and memory being upgraded. This was returned to the original memory and cpu state to make sure this wasn't the cause. These have been swapped in/out with no change so it looks like it is just coincidence.
There is no obvious signs of anything malicious going on, but this is always a possibility I guess? What to check for?
As this has basically brought the site to its knees with proxy errors displayed more often than not, has anyone got suggestions on what to look for. possibilities (however remote) as I am pretty desperate to get this fixed. If there is a simple explanation, what could cause it to start out of the blue?
The only possible answer I have come up with while searching is about the thread stack size being to small. And having to recompile python with this set to a larger figure. Is this related to the problem I am having, and is there an easy way to check if this is the case for the version I have running? If this is what I need to do, is there a good link for details on what needs to be done and how to do it?
CMF: 1.4.2 Zope: (unreleased version, python 2.1.3, linux2) Python: 2.1.3 (#1, Sep 7 2002, 15:29:56) [GCC 2.95.4 20011002 (Debian prerelease)]
Thank you for all your help on this as I am lost at what steps to take next. Thanks, Richard.
Hi,
Just some quick suggestions:
- First, can you reproduce the problem in a test environment?
- In general, I would say one of the first steps you should take is try it out with the latest stable versions of Zope (2.7.0) and Python (2.3.4). You shouldn't be surprised to find bugs running old versions of software and prerelease versions of your OS.
For example, there are some BTree-related bugs in older Zope versions that could cause this, IIRC
- I doubt the thread stack size is the issue, although others may be better equipped to say so
- I guess it's possible that when you replaced the CPU you didn't upgrade the ventilation/cooling adequately? Hot CPUs can cause all kinds of strange things.
seb
Richard Ettema wrote:
Hi,
I have a cmf site that has begun restarting frequently and at random intervals for no apparent reason.
<snip>
ERROR(200) zdaemon Process 25606 terminated by signal SIGSEGV(11)
-the cpu and memory being upgraded. This was returned to the original memory and cpu state to make sure this wasn't the cause. These have been swapped in/out with no change so it looks like it is just coincidence.
<snip>
The only possible answer I have come up with while searching is about the thread stack size being to small.
<snip>
CMF: 1.4.2 Zope: (unreleased version, python 2.1.3, linux2) Python: 2.1.3 (#1, Sep 7 2002, 15:29:56) [GCC 2.95.4 20011002 (Debian prerelease)]
Richard Ettema wrote at 2004-5-28 11:28 +0100:
I have a cmf site that has begun restarting frequently and at random intervals for no apparent reason. This first occurred about 2 weeks ago for about 2-3 days with the instance being restarted every 15 mins at worst. This problem disappeared as quickly as it had appeared.
2-3 days ago this same problem reappeared but with the site sometimes being restarted 10 times in a 15 minute period. I can find no sequence of url calls or any common reason for this to be occurring. The only thing I could find was the following entry appearing in the event log matching the times I have witnessed the instance restarting via top.
ERROR(200) zdaemon Process 25606 terminated by signal SIGSEGV(11)
Let your operating system write core files (usually disabled, read the manual page for "ulimit"). Analyse the resulting core file with "gdb".
-
Dieter Maurer -
Richard Ettema -
Seb Bacon