Hi David, On Tue, Jan 09, 2001 at 01:02:34PM +0000, David Lane wrote: > I am having major problems with MySQL 3.22.32 running on a > Linux-Mandrake 7.1 box with RAID 1 drives... Although mysqld is running > (through safe_mysqld, initiated by /etc/rc.d/init.d/mysql.server start), > any command line client interactions (e.g. mysql, mysqladmin, etc.) > hang. So do PHP queries (Apache-AdvancedExtranetServer/1.3.12 > (NetRevolution/Linux-Mandrake) mod_ssl/2.6.4 OpenSSL/0.9.5a PHP/4.0.2 > configured). Despite all log files reporting business as usual, even > "mysqladmin status" hangs. Calling a php page with MySQL generated > content results in an eventual time out on the browser but no error > messages. Shutting down MySQL, which requires kill -9 (the normal > "gentle" kill, as executed by /etc/rc.d/init.d/mysql.server stop, > doesn't have any effect) results in "could not connect" errors from php > pages, as would be expected. Restarting MySQL (after it's been killed) > works fine, no complaints in any of the log files: activity.log, > $hostname.err, or update.log..., but still no response of any kind from > the server. Wow! That's nasty! :-( > I've done isamchks on all the database tables (MySQL wasn't running at > the time), with no errors reported. I don't know what the problem could be, but if I may suggest some straws to clutch at, here are some: o Are you definitely looking at the right logs? An `lsof' on the main mysqld thread might be useful; o When you `strace' the main mysqld thread, does it produce any useful information? o Has mysqld produced many threads and is it using much cpu time, as reported by `ps' and `top'? o When you do issue a kill -9 and isamchk all the tables, are there definitely no mysqld processes running? o Are you hitting the maximum number of connections to MySQL? Perhaps when you start it up, /something/ connects and takes up all the available slots, even the one reserved for administrative tasks. o Do you have many tables? Perhaps with corrupted .frm files or something. Can you try to connect with `mysql -A' (IIRC) to avoid it reading in all the tables so it can autocomplete? o Could this be a glibc or general Mandrake bug? Have you applied all the Mandrake updates for 7.1? o Is something really weird going on? Perhaps /dev/null isn't correctly formed or /tmp isn't writeable (a+rwxt) by the mysqld user or mysqld's log and data directories have inappropriate permissions? Perhaps /tmp/mysql.sock is a real file? o When you start up MySQL, does /anything/ get written to the logs? Are you running `tail -f' on them to make sure...? :-) o Are the disks `thrashing' while all this is going on? o Did you install Mandrake in `paranoid' mode? (It's a while since I played with Mandrake, but ISTR this made all sorts of programs break in interesting ways... :-) ATB, Wesley. --------------------------------------------------------------------- Before posting, please check: http://www.mysql.com/manual.php (the manual) http://lists.mysql.com/ (the list archive) To request this thread, e-mail <[EMAIL PROTECTED]> To unsubscribe, e-mail <[EMAIL PROTECTED]> Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
