[RESOLVED] LSWS 4.0.13 - Signal 9's gone... Signal 15's new

[mistwang]

We are still working on it.

In the mean time, you can try our wswatch.sh script. Update /usr/local/lsws/bin/wswatch.sh with bellow shell script.

#!/bin/sh

BASE_DIR=`dirname "$0"`
cd "$BASE_DIR"
BASE_DIR=`pwd`


while [ "1" -eq "1" ]; do

ERR=1
if [ -f /tmp/lshttpd/lshttpd.pid ]; then
    kill -0 `cat /tmp/lshttpd/lshttpd.pid` 2>/dev/null
    ERR=$?
fi

if [ $ERR -ne 0 ]; then
    sleep 2
    if [ -f /tmp/lshttpd/lshttpd.pid ]; then
        kill -0 `cat /tmp/lshttpd/lshttpd.pid` 2>/dev/null
        ERR=$?
    fi
fi

if [ $ERR -ne 0 ]; then
    ./lshttpd
    D=`date`
    echo "$D: LSWS stopped, start LSWS." >> $BASE_DIR/../logs/error.log
fi
sleep 2

done

and edit lsws/bin/lswsctrl script, uncomment

WSWATCH=1

at the beginning of the file. do

./lswsctrl stop
./lswsctrl start

It will start lshttpd in 2 seconds if cPanel kills off LSWS.

 

[aww]

okay, thanks, just did the changes and will watch this weekend to see if it helps

my lsws location is under /opt/lsws btw, I guess you changed the default over time

 

[anewday]

okay, thanks, just did the changes and will watch this weekend to see if it helps

my lsws location is under /opt/lsws btw, I guess you changed the default over time

How did it go?

 

[aww]

How did it go?

Oh sorry for the lack of updates, the above workaround works perfect, no more emails about signal 15s

 

[aww]

Actually, you know what - with the 4.0.15 it's back

Every few hours or so

At [12/Jul/2010:01:11:02 -0400], web server with pid=18064 received unexpected signal=15, no core file is created.

At [12/Jul/2010:03:09:57 -0400], web server with pid=15548 received unexpected signal=15, no core file is created.

At [12/Jul/2010:05:08:50 -0400], web server with pid=11271 received unexpected signal=15, no core file is created.

At [12/Jul/2010:09:06:40 -0400], web server with pid=24521 received unexpected signal=15, no core file is created.

Would the upgrade have clobbered the old wswatch.sh or lsws/bin/lswsctrl ?

 

[MikeDVB]

Actually, you know what - with the 4.0.15 it's back

Every few hours or so


Would the upgrade have clobbered the old wswatch.sh or lsws/bin/lswsctrl ?

The newest .15 disabled wswatch by default as it was causing issues on cPanel servers when editing the httpd.conf (or at least, that's what it said in my support ticket).

 

[aww]

I really hope they find a workaround for this soon, the sheer number of times it's happening is worrying me.

Why is cpanel so anti-non-apache?

 

[mistwang]

Please download and try 4.0.16 release package, it should take care of the signal 9 issue. Will announce it tomorrow.

 

[aww]

Ah, I noticed the new date on 4.0.16 yesterday and gave it a try and sure enough it did indeed stop the problem.

Thank you!

 

[aww]

I spoke too soon.

Getting Signal 15's every 2 hours now on 4.0.16

 

[tachra]

Me too
web server with pid=1985 received unexpected signal=15, no core file is created. A new instance of web server will be started automatically!

 

[aww]

Not positive yet, but I think they are gone again with the unpublished 4.0.17

Nope, still there

At [28/Aug/2010:15:03:31 -0400], web server with pid=26374 received unexpected signal=15, no core file is created. A new instance of web server will be started automatically!

 

[Tony]

Not positive yet, but I think they are gone again with the unpublished 4.0.17

Nope, still there

At [28/Aug/2010:15:03:31 -0400], web server with pid=26374 received unexpected signal=15, no core file is created. A new instance of web server will be started automatically!

I think the best solution to this problem is from someone from cPanel to actually provide documentation of cPanel's Apache logic. They are sending kills among other things at the Litespeed processes. Why I don't know but a lot of their Apache logic is hidden in compiled binaries now so without them providing a helping hand no one will know. It could be something as simple as them waiting 10 seconds then issuing kills at all processes still existing.

This problem showed up when cPanel released 11.25 and unfortunately still is not fixed. Soon after 11.25 was released I tried contacting them about it. Unfortunately you talk to someone active on their forums they say make a ticket. The ticket people are unable to help provide such information. So I don't know where you'd go on this I've tried and failed as I don't have the connections to cPanel team it seems.

 

[aww]

I think the best solution to this problem is from someone from cPanel to actually provide documentation of cPanel's Apache logic. They are sending kills among other things at the Litespeed processes.

If cpanel wants to survive they really need to start working with vendors that use things other than Apache. There are now a dozen Apache replacements and a few of them like litespeed are too good to pass up.

Fortunately my linux and CLI knowledge/experience has grown to the point where I might be able to give up cpanel entirely in a year or two, especially if they don't start playing nice with others.

 

[aww]

Now I am periodically finding the webserver completely unresponsive.

I have to do a service lsws restart to get it going.

Whatever cpanel is doing is making it worse.

Discovered WSWATCH was automatically commented out somehow again.

Does the upgrade process replace wswatch.sh ? So I have to edit it after each update? (That's fine, I just need to know that.)

 

[Tony]

Now I am periodically finding the webserver completely unresponsive.

I have to do a service lsws restart to get it going.

Whatever cpanel is doing is making it worse.

Discovered WSWATCH was automatically commented out somehow again.

Does the upgrade process replace wswatch.sh ? So I have to edit it after each update? (That's fine, I just need to know that.)

We reported a problem with it when we tried 4.0.15 and eventually it was completely disabled. It was causing the cPanel restart scripts to lock up.

 

[aww]

Hmm.

Well I am using 4.0.17 (pre-release) and I think I re-enabled wswatch - and 18 hours later no lockups (or more importantly server stalls).

However I do not see any notices in the error.log about it restarting anything, so maybe it's not even running.

This is starting to become a serious problem as far as unreliability.

 

[mistwang]

wswatch may cause trouble for shared hosting environment which need to apply configuration changes frequently.
For signal 15 issue, please send us the error log and lsws/logs/ap_cmd_param.log