[liphp] Big MySQL problem

[Frank Imburgio]

Can you tell if the date/time stamp on the MYSQL data files are touched? 
If they are current, this may indicate there is database access where 
two proceses are waiting on each other -- a 'deadly embrace'. Even then, 
you should be able to 'kill' one of the MYSQL queries through the 
mysqladmin program. And, rebooting may not have cleared the deadly 
embrace, if they are rolledback from a transaction log. There may be a 
way to examine transactions in the rollback log -- this is a new MySQL 
feature.

When you telnet from the local machine to the port that MySql normally 
runs on (telnet localhost 3306) , does it answer? This would tell you if 
the communications portion is up or not. If not, the executable may be 
misconfigured or corrupt.

Does the daemon bring itself up and down gracefully ('service mysql 
restart' on redhat, 'mysqlstart restart' on some others)?

I have had instances where there is a 'hung' Mysql process that required 
me to perform the 'cleanup' on the tables, but never one where even 
MySQLadmin didn't answer.

Please let us know what you find out; this is a nightmare scenario.

--Frank

Christopher R. Merlo wrote:

>Folks:
>I sent a similar message to NYPHP, but I thought one of you MySQL guys
>on the list could help me out.
>
>MySQL on my server is using 99% of the CPU, according to root.
>Anything that tries to access it, like my web page, mysqldump, or
>mysqladmin, waits forever.  I'm getting concerned here.  I rebooted
>the box last night, and it didn't change MySQL.  Worse, MySQL appears
>to not be creating any log files at all.
>
>Anyone have any hints how to proceed?
>-c
>
>  
>