[PPP-26478] After update to onyx backup does not work

[nebuzzer]

hello,

i updated last week to onyx, but now the automatic backup function does not work properly. the backup is created, but the upload to the ftp does not work.

Warning:
Unable to upload backup files to FTP storage: pmm-ras command completed with non-zero exit code at /opt/psa/PMM/agents/shared/Storage/ArchiveStorage.pm line 126. STDOUT:
Warning: hosting "xxx.tld"
Unable to upload backup files to FTP storage: ack Datenübergabe unterbrochen (broken pipe): write( 13, 'domains/xxx.tld/backup_apache-files_1610160244.tgz ' ) at /opt/psa/PMM/agents/shared/IPC/Run/IO.pm line 558 STDOUT:
Warning: hosting "xxx.tld"
Unable to upload archive domains/xxx.tld/backup_apache-files_1610160244.tgz to FTP storage: ack Datenübergabe unterbrochen (broken pipe): write( 13, 'domains/xxx.tld/backup_apache-files_1610160244.tgz ' ) at /opt/psa/PMM/agents/shared/IPC/Run/IO.pm line 558

i also tried it without ftp-ssl and without compression. does not work. also the function "upload local backups to ftp" is missing.

 

[Sergey Prohorov]

Hello,
Could provide more information about pmm-ras fail from /var/log/plesk/PMM/backup-2016-10-16-02-44-xx-xxx/backup.log? Or you may send this log me in PM.

 

[Flavio]

Hi,
I've the same problem on two server after upgrade to plesk onyx, one with centos 6.1 and one with win server 2008 R2.
I've test the connection with filezilla and with cmd in windows server, in both case the ftp connection work fine.

 

[nebuzzer]

i send you two logfiles, one before update (12.0.18) and one after update (17.0.17). i edited the first entry, maybe it was a bit confusing. the backup function itself works fine, there is a problem with the ftp rotation.

the ftp connection works fine, but only 10-20gb are uploaded to the ftp, after that the upload crashes. the original backup remains on the local disk if i make smaller backups(of a single domain), everything works fine. my nightly full backup size is about 140gb.

 

[Sergey Prohorov]

Thank you for the reply. It seems we have a bug in keeping FTP connection alive. I will try to reproduce it in our test environment. Unfortunately, our default log does not include CURL output. Could you repeat the backup with additional login? To turn it on set '1' to parameter 'FORCE_DEBUG_LOG' in '/usr/local/psa/admin/share/pmmcli/pmmcli-rc' before starting backup. I would suggest you turn debug off as soon as a backup process will fail because it may generate a lot of log in this mode.

 

[Sergey Prohorov]

Could you try to add following section at end of panel.ini (/usr/local/psa/admin/conf/panel.ini on Unix or "%plesk_dir%\admin\conf\panel.ini" on Windows) and restart backup.

[pmm]
ftpFileTransferTimeout = 36000

If you already has 'pmm' section in this file - just add 'ftpFileTransferTimeout' to the section.
This workaround increases file transfer timeout to 10 hours.

 

[nebuzzer]

i will send you the log file with "FORCE_DEBUG_LOG 1" after it has finished and run another backup with the increased timeout time. the first backup should be done in 30-40min, if it works with the increased timeout time, i will leave you a message.

 

[nebuzzer]

[pmm]
ftpFileTransferTimeout = 36000

in panel.ini did work.

i have looked in the backups, but i never had a panel.ini. i had to create one. there must be something changed in the backup modul of plesk, which decreased the timeout time

 

[UFHH01]

Hi nebuzzer,

but i never had a panel.ini

Pls. be informed, that the file "panel.ini" is an OPTIONAL file on your server. Plesk will only create a default file with the name "panel.ini.sample", which can be manually renamed to "panel.ini".

You have as well the choice to use the FREE Plesk extension => "Panel.ini Editor" <= over the extension catalog, which could help you to fine tune your Plesk Control Panel.

 

[peraburek]

hello, since I have updated Plesk 12.5 to Plesk Onyx 17.0.17 update #5 my backup is not working (

Unable to upload backup files to FTP storage: ack Broken pipe: write( 21, '--cut--' ) at /opt/psa/PMM/agents/shared/IPC/Run/IO.pm line 558. STDOUT:

I have added "[pmm]
ftpFileTransferTimeout = 36000"

to Plesk.ini but it didn't helped (I have restarted server afterwards)

do you have idea how to solve this error, and what is "/opt/psa/PMM/agents/shared/IPC/Run/IO.pm" ??

 

[TomBoB]

hi, have upgraded a server 2 days ago and ran into exactly same issue as peraburek above.

past backups to FTP (always full backup) were just above 10GB, split into 2GB volumes.
After the upgrade, the last volume of 200MB and one volume of 2GB is uploaded, the remaining 4 2GB volumes are missing

Am still investigating ...
[on CentOS 7.2, Onyx#6]

 

[peraburek]

I forgot to mention, my Plesk Onyx installation is upgrade from Plesk 12.5 as well

 

[Raymond_Davelaar]

https://talk.plesk.com/threads/ppp-26569-ppp-26843-problems-after-backup.339879/

Problem is resolved for me by Plesk support team, they have replaced /opt/psa/admin/sbin/pmm-ras with file from attachment. Not sure if this is the solution for you, but it certainly works for me.

 

[Sergey Prohorov]

Problem is resolved for me by Plesk support team, they have replaced /opt/psa/admin/sbin/pmm-ras with file from attachment. Not sure if this is the solution for you, but it certainly works for me.

This fix is built for Ubuntu 14.04 x64 only and will not work with other OSes. pmm-ras for other OSes is available in the following KB https://kb.plesk.com/en/130278

 

[Pandur2000]

I copied the pmm-ras file over manually, that did the trick for a few days. Suddenly the errors are back and I realized: My copy was overwritten. I repeated the manual fix. Let's see how that works...

 

[Sergey Prohorov]

Hello. Could you please post output of the following 'plesk version'. I guess pmm-ras was overwritten by MU#9.

 

[Pandur2000]

Hello. Could you please post output of the following 'plesk version'. I guess pmm-ras was overwritten by MU#9.

Hi, Im on MU9.

Product version: Plesk Onyx 17.0.17 Update #9
Update date: 2016/12/06 22:30
Build date: 2016/11/17 16:00
OS version: Ubuntu 14.04
Revision: ab6766191d3ba26e7b21255ab007fc7fc56d84c6
Architecture: 64-bit
Wrapper version: 1.2

 

[TomBoB]

further to Pandur2000 and my own post above, have multiple servers, some

Product version: Plesk Onyx 17.0.17 Update #9
Update date: 2016/12/07 03:34
Build date: 2016/11/17 16:00
OS version: CentOS 7.2.1511
Revision: ab6766191d3ba26e7b21255ab007fc7fc56d84c6
Architecture: 64-bit
Wrapper version: 1.2

others

Product version: Plesk Onyx 17.0.17 Update #8
Update date: 2016/11/29 03:36
Build date: 2016/11/17 16:00
OS version: CentOS 7.2.1511
Revision: ab6766191d3ba26e7b21255ab007fc7fc56d84c6
Architecture: 64-bit

Problem

<?xml version="1.0" encoding="UTF-8"?>
<execution-result status="error" log-location="/usr/local/psa/PMM/sessions/2016-12-09-020003.695/migration.result">
<message id="e465a87b-4ac8-48c5-8760-f40d227ce520" severity="error" code="fatal">
<description>Runtime error: ack Broken pipe: write( 14, &#39;clients/clientsloginname/domains/clientdomain.org/backup_user-data_1612090200.tgz
&#39; ) at /usr/local/psa/PMM/agents/shared/IPC/Run/IO.pm line 558.

still persists.

When posting original post above, servers were on #7. I did NOT install the fix as provided by Sergey but was waiting for the next MU. Unfortunately it seems Sergeys fix slipped through the net and hasn't made it even into MU#9 as I still get the same error.
If I can help with detailed error logs, please let me know.

In addition to the PPP mentioned in the title, Sergey mentioned PPPM-5234 in a PM regarding this error.

Cheers,
Tom

 

[UFHH01]

Hi TomBoB,

actually, we are now at "Plesk Onyx - Version 17.0.17 Update #10". Pls. consider to use the command

plesk installer --select-product-id plesk --select-release-current --reinstall-patch --install-component base

... over the SSH - command line ( logged in as user "root" ), to make sure, that you receive the latest patches and updates immidiatly instead of waiting for the scheduled task.

 

[TomBoB]

Hi UFHH01,
was deliberately lagging behind on some servers for purpose of testing out the backup failure scenario.
The servers that are on MU#10 still throw the

<?xml version="1.0" encoding="UTF-8"?>
<execution-result status="error" log-location="/usr/local/psa/PMM/sessions/2016-12-11-001402.481/migration.result">
<message id="93bcede7-0d42-4a4b-8b53-4353aaa21d81" severity="error" code="fatal">
<description>Runtime error: ack Broken pipe: write( 14, &#39;clients/clientlogin/domains/clientdoman.com/backup_user-data_1612110014.tgz
&#39; ) at /usr/local/psa/PMM/agents/shared/IPC/Run/IO.pm line 558.

error and don't back up.
CentOS 7.2, Full backup to FTP space. ~11GB backup size

Will have to apply Sergeys fix manually and re-check after each MU update as the problem seems to resurface after every MU update (see above)

Cheers,
Tom