Re: Problems with system locking
In summary we have: MySQL 4.1, Linux 2.4, at least one INSERT in state 'update', most other threads waiting (probably for this one), one CPU 100% busy (probably the 'update' thread).
Regarding the suggested analyze, You didn't read the other thread, did you?
I suggested there about this:
It would be great if you could attach a debugger to the running server and look at the stack traces for all the threads to see where that insert waits. I'd appreciate to see that stack backtrace. Or you kill the server with SIGSEGV and do the same analyze on the core file. Rename the core file. Then you can do the analyze while the server is running again. This might give some insight, but I can't guarantee anything. :)
Ah. Hm. To get a core file, the server must have been started with --core. Sorry that I forgot to mention this.
Regards
Ingo Strüwing, Senior Software Developer - Storage Engines
MySQL AB, www.mysql.com