Customapache kills sshd

dennisc

Verified User
Joined
Mar 14, 2005
Messages
68
sshd is exiting whenever I make use of the Customapache build script. It happens on all our servers [Fedora Core 2].

Has anybody experienced this behaviour. I am glad we've installed Webmin in which I can fire up sshd again so I don't have to make a trip to the datacenter.

Nothing's lost since I do all important tasks in screen [http://www.gnu.org/software/screen/] but it's not desired behaviour. Does anyone know a solution to this?

Kind regards
 
Last edited:
the same happens on freebsd when using the base sshd, when its been compiling for about 30 seconds sshd will be killed. I worked around it by installing openssh-portable in /usr/local.
 
Thanks for sharing your solution. I hope Direct Admin folks have a fix for it, because I don't want openssh-portable. :-)
 
Hello,

I'm not sure how that's even possible. My only guess would be the system is overheating and rebooting the server due to the compiling usage.

If it happens mid compile, then there is pretty much no way it's the build script as 'make' is the active program.. It would likely be an overheating issue.. (or hardware problem with bad ram, or bad sectors etc.. something like that)

Check your /var/log/messages file for clues.

John
 
John,

All our servers do the same. They don't reboot, sshd gets killed. I have to fire it up through Webmin, and, because I use Screen I can attach the screen again so I can continue.
 
Same here on CentOS, I figured we were the only ones experiencing this problem.
It is most definitely not overheating though, temperature has been monitored and only went up 1 Celcius :)
 
When that happens for me, i have to call data center to reboot because i lost SSH :\
 
PeterT said:
After your edit:

It is how I've managed to restart SSH..

Maybe i have this problem more because i used php in CGI and CLI :\ i have a lot of different things to do after compiling so in fact, I compile, sshd killed, i wait 30 minutes (to be sure that compile is finished lolol ..; very sure ....) and I call data center -> please reboot server X ...

One day, (last compile) i remember that ssh wasn't killed. What I didi ? I think that i stoped all services before, except ssh and it was working ...

(sorry for my english ...)
 
Titam: I recommend you using Webmin and Screen. Screen is very handy: You open a screen, you start *something*, you do "ctrl-a d" to detach it. You logon to your box at another place and you retach the screen with "screen -rD". Very handy..

[ontopic]
Well I still don't know what actually happens. Hope we'll find out
[/ontopic]
 
dennisc said:
Titam: I recommend you using Webmin and Screen. Screen is very handy: You open a screen, you start *something*, you do "ctrl-a d" to detach it. You logon to your box at another place and you retach the screen with "screen -rD". Very handy..

Thanks, I sent you a Private Message ;)
 
Well I first noticed this issue when I had been setting up DA on new boxes, shortly after the install script started the apache and php compiling sshd would die (not the server), leaving telnet up as a backdoor I could login and see sshd was killed on every attempt the customapache was running, I couldnt see why. the moment I installed a new sshd the problem was solved.
 
Chrysalis said:
Well I first noticed this issue when I had been setting up DA on new boxes, shortly after the install script started the apache and php compiling sshd would die (not the server), leaving telnet up as a backdoor I could login and see sshd was killed on every attempt the customapache was running, I couldnt see why. the moment I installed a new sshd the problem was solved.

You speak better English than me lol :D ! It's exactly that, except i haven't telnet lol :p

I'll try Screen, but it's right that if the problem with kill could be solved ... :)
 
this hasnt always happened, about 2 or 3 months ago it started so I think they need to look at what changed in the last half a dozen versions.
 
Hello,

The only thing changed in the last few months would be the version number for apache and/or php. You could try editing the build script, and changing the numbers back a few versions to see if the issue stops.

One thing that might be possible (Just guessing here)...is if the load of a process gets too high, the OS might decide enough is enough and kill it. One way to get around that would be to use the "nice" program to run the script with a very low priority, so the load doesn't get too high.

John
 
what does this line do

echo "action=directadmin&value=restart" >> /usr/local/directadmin/data/task.queue

restart all services or just directadmin?

If I had a licence for it I would happily test DA on a non production box trying to find the exact cause of problems like this, so if john you want me to then let me know, unfortenatly I dont have a way of repeating the scenario I posted earlier otherwise.
 
Last edited:
Hello,

That will restart DA using /etc/init.d/direcatdmin or /usr/local/etc/rc.d/directadmin on FreeBSD.

John
 
Back
Top