MySQL: Lost connection to MySQL server during query problem

kid_c

I'm running MySQL 4.0.13 on a Linux machine. I frequently encounter the error:

"Lost connection to MySQL server during query"

It isn't from one specific script or page, but happens anywhere on the site, from scripts, or even when I'm just connecting to mysql through the mysql client.

If I refresh the page, or simply wait a second and try to connect again, it works fine. This error occurs about 20 times a day.

Any ideas why this is happening and how to fix it ? On identical hardware / linux installations, I'm running much larger databases and this problem doesn't happen.

Thanks

Xoid

I don't really know much about Linux dude, but have you tried running a command;

mysql_ping() before your connect and query calls? perhaps if you ping the database it may 'wake-up' as it were.

$ping = @mysql_ping();
if(!$ping)
{
print "Database is stoned!";
}
else {
print "Database is happy enjoying a good spliff!";
}

Sorry if this was absolutely no help, but, hey, I tried 😛 hehe

tsinka

Hi,

which Linux distribution do you use ? Does the problem appear with any kind of queries or only with specific queries ?

kid_c

I'm running Redhat 7.3, with kernel 2.4.18-3smp

It happens from any type of query (just at random though, not all the time), even if I am connecting through the command line.

tsinka

Hi,

this could be a bug in the glibc. Which version of glibc is installed on your system ?

kid_c

GNU C Library stable release version 2.2.5

tsinka

Hi,

I don't know if this solves the problem but look at the following ERRATA:

http://rhn.redhat.com/errata/RHSA-2002-197.html

It has been outdated by the subsequent errata:

http://rhn.redhat.com/errata/RHSA-2003-089.html

kid_c

I've upgraded to glibc-2.3.2-4.80.6, hopefully that might fix the problem. I'll need ot wait and see now. Thanks for the help!

kid_c

Problem still persists.......any other suggestions ?

tsinka

Did you compile MySQL yourself ? You might need to recompile MySQL...

kid_c

It is a fresh install of mysql 4.0.13, the binary from mysql.com

The problem was there before when I had version 4.0.6

tsinka

Hi,

you might try to recompile MySQL from sources to see if the problem still persist with a custom compiled MySQL version. Maybe the MySQL provided binary has some incompatibilities with RedHat 7.3 ?

davidegs

I have the same problem
os red hat 7.3 with all the last patches avaiable
mysql 4 last stable from mysql

mysql hangs up at 1 least 1 time per hour

(we have more than 60 unique visitor online at the same time for all the day)

any suggestion?

thx in advance

Davide

Sxooter

Firstly, anytime your database acts up, look for bad hardware.

memtest86 is a good tester for bad CPU/memory/cache etc...

also check your drive for bad blocks (fsck -c /dev/hdx1 etc...)

Also, this isn't the first I've heard of MySQL 4.0.x crashing randomly, it could be a bug in there somewhere. Likely you'll get better help from the MySQL development lists if you can get a core dump with debugging symbols and back trace it, then feed the data to the developers to figure out why it's crashing.

davidegs

Hello

i have downgraded to mysql 3.23.56 to see if something change...
(i'm managing a remote machine so i'm not able to make all the tests i need)

in mysqld.log i see this error:

030806 13:36:56 /usr/libexec/mysqld: Forcing close of thread 18 user: ''

number of the thread changes in every line..

we have 2 machine

machine a with php pages and machine b with mysql db

i thought about a too many connection problem and tried forcing a bigger value (1000) but i still receive this error

the strange thing is that in about 10 minutes all returns to work without doing anything (for around 1 hour then a new crash...)

any idea?

thx

Davide

tsinka

Which linux distribution ?

davidegs

redhat 7.3 (with all patches from redhat) on a p4 single processor

the situation is this

we had 1 server...

due to high load we decided to move to 2 server

so i dumped all the db from the old server and installed on the new one...

i thought... all is working on the old server... then it should work on the new one... (old is a redhat 7.2, new is a redhat 7.3)

but ... well my assuption seems wrong 🙂

thx in advance

Davide

Sxooter

I repeat, have you tested the memory in your new server? If a database is crashing for no apparent reason, suspect the hardware first. If you haven't proven it good to yourself, you can't just assume it is and keep looking for the cause elsewhere.

davidegs

i'm in italy
the server is in the USA

i can't check memory or make other tests that requires floppy disk or rebooting the server...

i have asked my isp to make some checks also but i will receive an answer at least tomorrow...

thx for now

Davide

however the mysql server is not "crashing" simply collapses and becomes really slow

i have found that there are some slow queries...

since this is a porting from a mysql db on a P3 to a mysql db on a P$ it is quiet strange that the same queries here takes so long...

thx in advance

Davide

Sxooter

A nice check to see if the server might have memory issues is to compile something large and complex (the linux kernel, MySQL, and PostgreSQL are good candidates) over and over and just log the output and then grep it for sig 11. If you get a sig 11 during compile, there's usually bad memory in the box.