error 503, service not available

[Drakies]

Same Issue : plesk 11 503 - Service Not Available

Hey Guys,

Has anyone found a solution to this problem. We are experiencing the exact same problem mentioned in this thread.

Server Details
Ubuntu 12.04 Precise 64bit
8GB RAM
Intel(R) Xeon(R) CPU E5620 @ 2.40GH

Plesk Version
Plesk 11.0.9 Update #27

Logs
DEBUGGER DETECTED... Bye!
2013-01-06 19:37:54: (mod_fastcgi.c.2613) unexpected end-of-file (perhaps the fastcgi process died): pid: 11173 socket: unix:/opt/psa/tmp/sw-engine.sock-0
2013-01-06 19:37:54: (mod_fastcgi.c.3357) child signaled: 9
2013-01-06 19:37:54: (mod_fastcgi.c.3400) response not received, request sent: 497 on socket: unix:/opt/psa/tmp/sw-engine.sock-0 for /index.php?, closing connection
2013-01-06 19:37:54: (server.c.1543) server stopped by UID = 0 PID = 11241
2013-01-06 19:37:55: (log.c.166) server started

I have already checked disk space and available memory and that seems fine, is there anything else I can try to resolve the problem.

 

[ManuG2k]

Hi guys,
I have same problem, every 8-12 hours.

Has anyone found a solution to this problem ?

Manuel

 

[sao sang mo]

I have contact my local supporter. He only says that he found the logs:

2013-04-07 19:50:00: (server.c.1543) server stopped by UID = 0 PID = 18257
2013-04-07 19:50:17: (log.c.166) server started
2013-04-08 00:02:29: (server.c.1543) server stopped by UID = 0 PID = 17713
2013-04-08 00:02:52: (log.c.166) server started
2013-04-08 01:08:56: (server.c.1543) server stopped by UID = 0 PID = 24171
2013-04-08 01:09:03: (log.c.166) server started
2013-04-08 01:49:39: (server.c.1543) server stopped by UID = 0 PID = 28500
2013-04-08 01:49:46: (log.c.166) server started
2013-04-08 02:30:11: (server.c.1543) server stopped by UID = 0 PID = 3238
2013-04-08 02:30:17: (log.c.166) server started
2013-04-08 02:45:24: (server.c.1543) server stopped by UID = 0 PID = 5410
2013-04-08 02:45:37: (log.c.166) server started
2013-04-08 07:46:26: (connections.c.309) SSL: 1 error:1407609C:SSL routines:SSL23_GET_CLIENT_HELLO:http request

Plesk's supporters please help me.

And the problems come at midnight when my server has more free resource.
thanks

 

[esfahanhost]

Hi Dear

i have this error in plesk 11 and windows server 2012 , and Plesk Support Team resolve this problem by this method :

IIS => Application Pools => Right Click on PleskControlPanel => Advanced Setting => Change Identity to LocalSystem => OK => Restart and recycle this pool

Have a good time
Masoud Saeidi
EsfahanHost.com

 

[freaky@]

OMG no solutions?

We have a clean install (new server / new plesk install) in which a backup from a 9.5 was restored (pleskbackup/pleskrestore cli). I'm probably overlooking some things but here we go:
* Special configurations of all subdomains were not imported (vhost.conf)
* DNS notifications no longer worked because it decided to list one of the slaves as primary in the SOA (and bind doesn't notify the server in the SOA record)
* Had to do a lot of stuff to get the restore to work in the first place
* SFTP with chrooted bash doesn't work - have to copy libs manually all over the place not to mention what it'll do with security updates (the roots won't get updated)
* Couldn't change DNS records on several domains

And this is just from the top of my head. What does the 9 stand for, really? Still 9 years away from any expected stability? You'll have 9 years of misery? 11.0.9 should be called alpha 01....

Please tell me - where can I bill my hours? This product isn't stable, isn't well tested, not (really) well documented (heck - the installation manual doesn't even mention you need to disable apparmor on ubuntu).

[edit]Oh did I mention - CAN'T BE DEBUGGED (debugger detected - exiting <- really nice on such a stable piece of software...)[/edit]

 

[ghazlewood]

I've experienced the same problem as this thread and it was because the /tmp filesystem was full.

 

[TysonG]

I have the same issue, well close to the same.
Server
CentOS 5.8
Intel Xeon, 4 cores
8GB RAM

Plesk
11.0.9 Update #41

Plesk Web Server crashes every time I update or add a billing account for a client so I am able to re-create the issue. I tailed all log files mysqld.log, apache, psa, and the only error I have to work with came from /var/log/sw-cp-server/error_log
2013-06-04 10:12:52: (server.c.1543) server stopped by UID = 500 PID = 1955

Then watch dog restarts psa

I have enabled debuging in plesk billing and re-created the issue and no errors show up.

This is very bad for business. My clients sign up for hosting in my store and psa crashes, or if we need to update billing information for a client psa crashes. I have nothing to work with to figure this out other than the error listed above.

I have plenty of RAM and the hard drive isn't full.

Can somebody please help?

 

[freaky@]

I haven't seen the error a little more than a week.

Found some entries in a log file that some cgi couldn't ptrace some other cgi. Ubuntu by default has some additional security policies, amongst which apparmor (which needs to be disabled - but the installation manual fails to mention) and ptrace protection (a process is only allowed to ptrace itself or it's children - normally all processes can be ptrace'd). I have disabled the ptrace security and the error hasn't returned since. Don't know if it's really the solution - only that the issue hasn't reoccurred so far...

To disable the ptrace protection on ubuntu edit /etc/sysctl.d/10-ptrace.conf and set the kernel.yama.ptrace_scope variable to 0. Either reboot then or do 'echo 0 > /proc/sys/kernel/yama/ptrace_scope'.

Again - I'm not sure if this is truly the solution to begin with - but after 4 days of the management interface going down on a daily basis it hasn't reoccurred so far thus.

Have a hard time find the error I found in the logs back. There's more people mentioning issues with ptrace protection tho'.

See for example:
http://forum.parallels.com/pda/index.php/t-112880.html
https://atomicorp.com/forums/viewtopic.php?f=3&t=4636
https://www.atomicorp.com/forums/viewtopic.php?f=18&t=5681

 

[MarcoS71]

So not funny

Hi,

Just to clarify, do you login then get the error, or does the error occur when you navigate to the Plesk panel? (The former I think used to be caused by lack of memory for the panel APC instance but that seems somewhat unlikely in your case since we used to see it on 256MB VPSs not 8GB dedicated servers).

Is it the same rough time each day (22:08) that you have those errors in the log (and I assume that isn't your password)?

I'm only running Plesk 11 on various Centos 5 and 6 VPSs, but I'm not having a problem with the Control Panel crashing.

Paul.

Just wait for it.

I'am so bloody tired of this Parallels ****. How does a company manage to make such crappy software and still be able to sell this.
Ridiculously limited logging.

I see a lot of questions by "experts" asking for logging. Well yes that would be so nice....
However the only msg in when starting the PSA service is :

"Failed to start the "psa" service.
Cannot start/stop/restart service: service "psa" status is not "running" after command "start" "

This is some of my environment:

Parallels Panel v11.5.30 core files

"OS Template centos-5-x86
Applications Installed jre,
mailman,
mod_perl,
mod_ssl,
plesk8-fcgid,
plesk9-tomcat,
pp-php5,
pp10-mailman,
pp10-migration,
pp10-qmail,
pp10-spamassassin,
pp10-webmail,
spamassassin,
tomcat,
webalizer"

I see that this issue seems to be reappear every so many updates. Honestly I am thinking of getting rid of all of the Plesk ****. Unless this forum has the greatest support ever.
Any time I can't get my service to run properly I need to hire support to reinstall cause that then seems to be the only way to repair anything.

 

[MarcoS71]

Oh by the way... This would be the first time this forum would be any use to me.

 

[MarcoS71]

I agree this is very BAD for business.
Have you seen any actual employees of Parallels here ???

 

[MarcoS71]

OMG no solutions?

We have a clean install (new server / new plesk install) in which a backup from a 9.5 was restored (pleskbackup/pleskrestore cli). I'm probably overlooking some things but here we go:
* Special configurations of all subdomains were not imported (vhost.conf)
* DNS notifications no longer worked because it decided to list one of the slaves as primary in the SOA (and bind doesn't notify the server in the SOA record)
* Had to do a lot of stuff to get the restore to work in the first place
* SFTP with chrooted bash doesn't work - have to copy libs manually all over the place not to mention what it'll do with security updates (the roots won't get updated)
* Couldn't change DNS records on several domains

And this is just from the top of my head. What does the 9 stand for, really? Still 9 years away from any expected stability? You'll have 9 years of misery? 11.0.9 should be called alpha 01....

Please tell me - where can I bill my hours? This product isn't stable, isn't well tested, not (really) well documented (heck - the installation manual doesn't even mention you need to disable apparmor on ubuntu).

[edit]Oh did I mention - CAN'T BE DEBUGGED (debugger detected - exiting <- really nice on such a stable piece of software...)[/edit]

Did you get any help...?

 

[freaky@]

No, I mostly don't.

I fixed it myself. See the ptrace misery a bit lower.

To answer your other question, yes, there's people from Parallels here. Unfortunately they seem to be cleaning up my language from frustrated posts most of the time. See who edited my post . There's at least one besides IgorG I've seen respond on occasion.

From what I got it's supposed to be more stable on CentOS. Last time I checked the installation manual for Ubuntu *still* failed to mention you need to disable AppArmor before installing. Luckily it's not like that's been the case since the first release that ran on Ubuntu (v9 iirc). Obviously everyone reads the entire knowledge base before installing anything I suppose...

 

[MarcoS71]

Ah thanks for that. I was beginning to think they just don't like me.
Problem just seems to all these loose packages updating it self without proper compatibility checks. Also I think simple users like me are sort of expected to give our hosting provider a call and ask them to fix it. However that ll cost me 60 euro an hour.
Annoying as hell ! I seem to remember one day I didn´t need plesk! Just FTP access to my map and a DNS mapping. I am a Java developer... Wauw if I delivered shitty software like this I would be out of a job quick. Thanks for your reply I will have to face the music I guess.