Can't Kill HTTPD Process

romeo_tango · 11-29-2010, 08:04 PM

Hi,

I have a problem with httpd. This already occured twice actually.

When I tried to restart the service, it seems that the parent got killed in the stop process but it won't restarted again and left tons of child processes.

I tried to grep the PIDs and killed all in once using :

Code:

ps -fe | grep http | awk '{print $2}' | xargs kill -9

but failed.

In other words, why couldn't I kill -9 these PIDs? These are not even zombies. I checked using 'top' command and it showed

Code:

682 processes: 680 sleeping[

no zombies at all.

Please help me how to get rid of these unkilled process.

Thanks.

kbp · 11-30-2010, 12:01 AM

'pkill -9 httpd' is a little shorter

.. usually processes that don't respond to signals are in uninterruptible sleep ('D' in top output), they may be waiting for IO

hth

romeo_tango · 11-30-2010, 12:03 AM

Quote:

Originally Posted by kbp

'pkill -9 httpd' is a little shorter

.. usually processes that don't respond to signals are in uninterruptible sleep ('D' in top output), they may be waiting for IO

hth

But the case here is that the httpd processes don't have any parent ID and can't be killed.

What should I do instead of rebooting? Have anyone encountered this issue before?

Thanks

kbp · 11-30-2010, 12:18 AM

Child processes that are orphaned ( their parent is deceased ) usually have their parent set to init, please post a sample of 'ps -ef'

romeo_tango · 11-30-2010, 12:43 AM

Sorry since this is very long I'll just put it here

Code:

http://pastebin.com/cjhE3DAr

Thanks.

kbp · 11-30-2010, 04:33 PM

Thanks but I can't see the whole command line, could you please use 'ps -wwef | grep http' ?

If I was to guess, it looks like these are hanging around from a 'httpd -k start' or 'httpd -k restart' - how are you controlling the daemon, init script ?

romeo_tango · 11-30-2010, 08:59 PM

Quote:

Originally Posted by kbp

Thanks but I can't see the whole command line, could you please use 'ps -wwef | grep http' ?

The server was already rebooted and back to normal so I don't think if I run the command now would give some hint to you

Quote:

If I was to guess, it looks like these are hanging around from a 'httpd -k start' or 'httpd -k restart' - how are you controlling the daemon, init script ?

That's right. I did 'httpd -k restart' yesterday perhaps twice or thrice because the http wasn't accessible. I thought it was dead, but it wasn't.

The thing is.. I always do the same thing, same treatment and same command in other Sun servers and never encountered this issue before in other words, never met a process that can't be killed using 'kill -9'

jlliagre · 12-01-2010, 02:20 AM

A process that can't be killed by "-9" has a system call pending that doesn't return for some odd reason.

romeo_tango · 12-01-2010, 02:25 AM

Quote:

Originally Posted by jlliagre

A process that can't be killed by "-9" has a system call pending that doesn't return for some odd reason.

So what should I do with those kind of processes other than rebooting?

Thanks.

jlliagre · 12-01-2010, 03:46 AM

Take measures for the system call to end which might be possible or not depending on what system call it is.

ADxD_7 · 12-16-2010, 07:06 PM

Another thing it may be is that that SMF is controlling apache - I had this issue awhile back, I tried to use the apachectl command to stop httpd and it would not die, I also tried using kill -9, it was because the service was being controlled by SMF so you may do a quick svcs -a | grep -i apache - and see if it is in there cause it will restart it over and over and over if you kill -9 it or whatever you do to it...

romeo_tango · 01-03-2011, 03:16 AM

Hi.. thank you for the reply.

I did svcs -a and found these lines :

Code:

legacy_run     Dec_13   lrc:/etc/rc3_d/S50apache
maintenance    Dec_13   svc:/network/http:apache2

Right now the Apache status is working just fine. But I don't understand what's the difference between those two lines.

I usually started the apache manually by running httpd -k start.

Is it bad?

jlliagre · 01-03-2011, 04:14 AM

The first line apply to apache 1.2 and the second one to apache 2.x. Both are installed on your machine. 1.2 is launched through rc scripts, 2.x is managed by smf, and is reported in maintenance mode, i.e. not functioning properly.

romeo_tango · 01-03-2011, 04:29 AM

Quote:

Originally Posted by jlliagre

The first line apply to apache 1.2 and the second one to apache 2.x. Both are installed on your machine. 1.2 is launched through rc scripts, 2.x is managed by smf, and is reported in maintenance mode, i.e. not functioning properly.

okay.. that is strange because right now the httpd is running well and I am using Apache 2.
How to find out what went wrong? I don't find any errors in /var/adm/messages nor in apache logs.

jlliagre · 01-03-2011, 06:14 AM

Code:

svcs -xv http:apache2

should point you to relevant log files.