forum.vmspython.org

[tsgdavid]

I am running MYSQL 5.1.22-rc on an Integrity RX2620 system which is running OpenVMS 8.3. The system has 2 GB of RAM.

I have noticed that the overall memory usage on the system grows over time ranging from about 40% after a reboot to at least 78% in under two months.

After further investigation, I have seen that the number of pages in the working set has increased for several processes. One of these is the MYSQL051_SERVER process. The number of pages in the working set has increased from 89616 (after a reboot) to 316144. I am wondering if this is something that I should be concerned about and if this is going to result in the need for periodic reboots (or restarting MYSQL).

I am trying to investigate further to see what happens when I restart MYSQL and exactly how this grows over time. Maybe it will taper off at a certain level?

Does anyone have any experience with this or have any suggestions as to what I should be looking at?
_________________
David Williams

[bisonmalembouche]

Hello

Are you up to date with your VMS patches ?

For the process mysql, can you post the pgflquota initial and used ?

Can you post, if your Mysql process has the pid 2B4

wr sys$output f$getjpi("2b4","pgflquota")
wr sys$output f$getjpi("2B4","pagfilcnt")

It seems there is a memory leak, until this is fixed, a workaround can be to purge the working set regularly. If you do so, add some pagefile before.

$ help system $purgws

System_Services

$PURGWS

Removes a specified range of pages from the current working set
of the calling process to make room for pages required by a new
program segment.

Format

SYS$PURGWS inadr

C Prototype

int sys$purgws (struct _va_range *inadr);




Additional information available:

Argument

System_Services $PURGWS Subtopic?
System_Services Subtopic?
Topic?
$

[jfp]

MySQL has many internal caches which grow over time, so it is probably difficult to decide if it is a memory leak or a normal behavior.

You can, restart it, time to time if it is a problem, or lower the various caches size.


JF

[JeremyBegg]

Hi David,

No doubt JFP is correct and what you're seeing is likely to be normal behaviour.

At a more general level I would suggest you increase the RAM in your Integrity server. 2GB is very low for Integrity; I'd be putting another 2GB-4GB in it if you can.

Regards,
Jeremy Begg