Serious httpd problems.

[begore]

Hi,

I'm running FreeBSD 6.1 and Plesk 8.0.1 (though this problem also seems to exist on FreeBSD 6.0 and Plesk 8.0.0).

Whenever a httpd processes is launched it never closes down cleanly, and remains in memory in 'lockf' state - this ultimately leads to the memory on the server filling up, followed by the swap partition, and then the box falls over...

I have a 'top' full of:

86432 apache 1 20 0 49820K 24172K lockf 0:13 0.00% httpd
85065 apache 1 20 0 56048K 4568K lockf 0:11 0.00% httpd
86452 apache 1 20 0 58692K 33004K lockf 0:10 0.00% httpd
86447 apache 1 20 0 30920K 7688K lockf 0:10 0.00% httpd
86441 apache 1 20 0 54900K 4408K lockf 0:09 0.00% httpd
86438 apache 1 20 0 60696K 37508K lockf 0:09 0.00% httpd
86434 apache 1 20 0 32988K 9740K lockf 0:09 0.00% httpd

I've tried everything - and to give you some idea, there's only about 3 websites currently on this box, and they are all fairly low traffic.

1Gb of RAM in the box and a 2Gb swap partition.

The box is currently falling over about twice a day.

If I get in quick enough and killall -9 httpd and then restart it, all is well again (the RAM returns to normal and the swap space is instantly cleared).

Help!

Cheers,
Steve.

 

[DCNet_James]

I'm looking into this as well. Though I have yet to have a server failure on a 6.0 machine running 8.0.1 with the latest build. I would say setup a cronjob to run every 4 or 5 hours for now until SW-Soft, myself, or someone else comes up with something, executing the command, /usr/local/psa/rc.d/httpd restart

Seems to be some sort of file locking issue.

Thx
James

 

[begore]

Thanks James, fingers crossed that someone comes up with the answer sooner rather than later!

Cheers,
Steve.

 

[DCNet_James]

I've been tooling around with it, and from my tests, everything seems ok right now. One of my servers is very high traffic and is holding pretty stable right now.

I guess only time will tell if it decides to flip out. I'll keep a close eye on my box and see if I can find the issue if there is one.

J

 

[DCNet_James]

You might want to look at the following: Apache resource limiting

 

[begore]

Thanks, I'll take a look!

Cheers.

 

[begore]

Still having the same problem - I really need to find an answer to this one - I've got a server sitting here that I just can't use.

 

[begore]

I've found the answer to the problem (though I've not managed to implement it yet)

http://tinyurl.com/j528j
http://tinyurl.com/l4nod

Any ideas what config changes in Plesk would make this happen?

 

[Vide]

Hi

I'm experiencing a very similiar problem with a mixed FreeBSD 6.0/6.1 webfarm...when there is a peak of petition, Apache stalls on "lockf" with zillion processes. I've just tried to change to a *:80 configuration as explained in the links you posted and something happens...now the "lockf" status has disappeared, but anyway performances are still the same (with load avg pushing up till it "kills" the machine).
We are serving PHP pages with Smarty templates...any idea?

 

[begore]

Hi,

What did you change in your config to make plesk allow the *:80 ? I tried a few things but kept getting config errors.

Cheers.

 

[Vide]

I'm not using Plesk but a custom PHP webapp, anyway the Apache sympton were very similiar (tons of httpd in lockf status). The change suggested by the link you posted is effective, changing the status to "accept" but I still have performance problem with my app