[report] psa-pc-remote make 400% (and more) cpu usage

[Mickael Salfati]

[bug report] psa-pc-remote make 400% (and more) cpu usage

---------------------------------------------------------------
PRODUCT, VERSION, OPERATING SYSTEM, ARCHITECTURE
Parallels Plesk Panel 10.4.4 Debian Lenny 64 bits

PROBLEM DESCRIPTION AND STEPS TO REPRODUCE
run top command:
13019 postfix 20 0 64244 1564 1172 S 400% 0.0 1128:32 psa-pc-remote

ACTUAL RESULT
psa-pc-remote service don't stop to take %Cpu after a while until psa-pc-remote crash and SMTP become down.

EXPECTED RESULT
psa-pc-remote work normally.

ANY ADDITIONAL INFORMATION
My installation is an upgrade from 10.3.1 if it could help. On 10.3.1 all worked good.
I have 7 plesk servers in the same configuration, and I have only one server with this issue.

I tried to run the command below but same result even after a reboot.

/usr/local/psa/admin/sbin/mchk --with-spam

Here is an interesting (perhaps) extract from /usr/local/psa/var/log/maillog

Nov 22 14:52:52 ns3 /usr/lib/plesk-9.0/psa-pc-remote[13019]: handlers_stderr: SKIP
Nov 22 14:52:52 ns3 /usr/lib/plesk-9.0/psa-pc-remote[13019]: SKIP during call 'check-quota' handler

For information quotas are well enabled on the system.
--------------------------------------------------------------

Thank you per advance

 

[IgorG]

Thank you for report. I have forwarded it to development team. Will keep this thread updated with results.

 

[IgorG]

Developers have informed me that they cannot reproduce this problem. Therefore they have asked about possibility to login to your server for the further investigation of this problem. Please send me login credential in PM if possible and I will forward it to developers. Thank you.

 

[DocD]

Having the same issue. Would be interested in what you find.

 

[Mickael Salfati]

Developers have informed me that they cannot reproduce this problem. Therefore they have asked about possibility to login to your server for the further investigation of this problem. Please send me login credential in PM if possible and I will forward it to developers. Thank you.

IgorG, I sent you login credential by PM. Can you confirm that you well received ?
Actually, we can see with "top" command. (726% cpu busy with that process)

20306 postfix 20 0 230m 1924 1172 S 726% 0.0 4243:05 psa-pc-remote

I don't restart the service for now to allow parallels dev to observe the issue.

 

[Mickael Salfati]

IgorG, I'm sorry but I can't wait more, I restarted my mail service, because my server was crashing:

top - 12:01:45 up 43 days, 19:16, 1 user, load average: 26.46, 24.42, 22.09

 

[DocD]

I tried restarting postfix but the process stayed.

Did some poking around and noticed a strange entry on the Plesk Mail Queue page that had an incoming email listed with no addresses for From: and To: fields. Could not remove it from that page.

Located the email in: /var/spool/postfix/incoming and removed it.

Don't know if it was the cause, but it lined up with the time psa-pc-remote hung up, and it was spammy, so I figured it couldn't hurt.

Ran:
service postfix stop
kill -SIGTERM <pid of psa-pc-remote>
Checked it was gone from top, it was.
service postfix start

Process is now gone from top, load is back to normal and postfix is running fine.

Would like to know what the issue was though.

 

[centron]

Same issue with psa-pc-remote since the update to 10.4.4. I opened a ticket to Parallels support. They will reply in the next 4 hours. I will update this thread if I have some useful information.

 

[DocD]

Much appreciated centron. It's back and ramping up the load. No emails stuck in the queue this time either.

 

[IgorG]

Same issue with psa-pc-remote since the update to 10.4.4. I opened a ticket to Parallels support. They will reply in the next 4 hours. I will update this thread if I have some useful information.

Could you please let me ticket ID? I will check it.

 

[centron]

Hello IgorG,

here the tickets:

[Parallels #1268619] High server load caused by psa-pc-remote

and

[Parallels #1268679] SMTP authentification via DIGEST-MD5 fails

Thanks in advance for your help.

Kind regards.

 

[IgorG]

Thank you. Developers are working on it. I will inform you with results.

 

[centron]

Here the statement from parallels regarding to the psa-pc-remote issue:

Request ID: 90368

This issue will be fixed in the future releases of plesk. In order to get the most actual information regarding specified product maintenance team request (90368), I suggest that you contact your Parallels Sales Representative as Support Department does not possess the information about specific time frames and dates of release.

So I did the following workaround which may help you too:

/etc/crontab

*/15 * * * * root /etc/init.d/pc-remote restart

 

[Mickael Salfati]

Personally, I remark that a dev member was connected on my server this morning, and now issue seems to be fixed

 

[kaesar]

I have the same problems. In a debian 6 with plesk 10.4.4

 

[DocD]

Any idea what they did Mickael? Did they just add in the cronjob to restart the service or actually fix the issue?

 

[jmorby]

Hello IgorG,

here the tickets:

[Parallels #1268619] High server load caused by psa-pc-remote

and

[Parallels #1268679] SMTP authentification via DIGEST-MD5 fails

Thanks in advance for your help.

Kind regards.

Have you had a response or resolution to either of these issues yet? We have both currently and would rather not raise yet more support tickets if there is already a known / public fix ...

Jon

 

[Frater]

Rather than restarting psa-pc-remote each 15 minutes I want to do this only if the system has really gone haywire.

For this purpose I added this line to my /etc/crontab

*/3 * * * * root expr `top -b -d5 -n3 -u postfix | grep '[p]sa-pc-remote' | awk '{print $9}' | sort -n | head -n1 | awk -F. '{print $1}'` \> 90 >/dev/null && /etc/init.d/pc-remote restart

It has the advantage that it will kick in faster (worst case 3 minutes) and only restarts if the process is really doing something nasty.

 

[Mickael Salfati]

Any idea what they did Mickael? Did they just add in the cronjob to restart the service or actually fix the issue?

DocD, I made a mistake. issue is not fixed at all on my server

20038 postfix 20 0 415m 2472 1172 S 740% 0.0 10120:06 psa-pc-remote

Last connection by a Plesk Team member was: (UTC+1)

Nov 25 04:10:36 ns3 sshd[21639]: Accepted password for root from 91.204.24.253 port 60352 ssh2

Rather than restarting psa-pc-remote each 15 minutes I want to do this only if the system has really gone haywire.

For this purpose I added this line to my /etc/crontab

*/3 * * * * root expr `top -b -d5 -n3 -u postfix | grep '[p]sa-pc-remote' | awk '{print $9}' | sort -n | head -n1 | awk -F. '{print $1}'` \> 90 >/dev/null && /etc/init.d/pc-remote restart

It has the advantage that it will kick in faster (worst case 3 minutes) and only restarts if the process is really doing something nasty.

Ty

 

[KurinichinathanT]

Argh, I just upgraded to the latest version and having the same error! any solution?