Hi all,
we have rh 9.0 (kernel 2.4.20) mysqld 4.0.15 php 4.3.3 apache 1.3.29
and a quite large (10.000 unique visitors/day) server and messageboard (phpbb). every 2 or 3 days the server crashes (mysqld crashes, then all myslqd_safe and httpd processes crashes one after one)
there is no stacktrace in the mysql-data-dir/xyz.err logfile. it simply shows a message that the database wasn't shutdown normally after the crash, nothing more.
however there is a stacktrace in /var/log/messages of the crashed mysqld. I tried the use stack trace method of the mysql documentation (chapter D.1.4) with the stack trace of /var/log/messages but without success. it only shows some numbers and some "?" at every "zero" offset (like 0x000000f ).
- is the /var/log/messages stack trace equivalent with the one which should show up at the err logfile of mysqld, so is it proper to make a bug report?
- how can we find the problem on the server? there's the option to log every sql statement which might show up the error, but because this would slow down the server immensely we don't want to do this right now.
RH 9.0 has done some experimentation with the thread library, and made it the default. I've seen out of memory problems with Resin on it, which I solved by fixing up the symlink on /lib/libpthread.so to point to the normal version instead of the experimental one. I suppose those "experimentations" would affect MySQL as well. If this is indeed the problem in your case, you can either fix up /lib/libpthread.so to point to something sane, or use a statically linked binary from www.mysql.com - the regular one (non-Max) is statically linked (on x86 Linux), but the Max one is linked dynamically to make UDFs work.
-- Sasha Pachev Create online surveys at http://www.surveyz.com/
-- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
