MU #58 dies on CentOS 6 with Epel

[RobHost]

Hi,

we have dozens of CentOS servers failing while updating to MU #58 since 2 days (all have EPEL enabled - that wasn't a problem so far)

Since you use one or more 3rd-party repos (say, atomic), be careful when installing different package versions from different repos as this may lead to installation failures. For example, you may encounter a problem if you first install PHP from a 3rd-party repo and then upgrade it using the Parallels repo. To avoid such situations, install and upgdare packages from the same repo.

Traceback (most recent call last):
File "/usr/local/psa/bin/yum_install", line 194, in <module>
main()
File "/usr/local/psa/bin/yum_install", line 189, in main
installer.perform(to_install, opts.remove)
File "/usr/local/psa/bin/yum_install", line 149, in perform
raise Exception("Scriptlet or other non-fatal errors occurred during transaction.")
Exception: Scriptlet or other non-fatal errors occurred during transaction.
~emptyFileFetcher: get file (~empty)/PSA_11.0.9/plesk-11.0.9-cos6-x86_64.inf3
FileFetcher: get file (~empty)/SITEBUILDER_11.0.10/sitebuilder-11.0.10-rhall-all.inf3
FileFetcher: get file (~empty)/BILLING_11.0.9/billing-11.0.9-rhall-all.inf3
FileFetcher: get file (~empty)/NGINX_1.3.0/nginx-1.3.0-cos6-x86_64.inf3
Error: Das Dienstprogramm Yum konnte nicht ausgeführt werden.
Das Dienstprogramm Yum konnte nicht die erforderlichen Pakete installieren.
Achtung! Ihre Software ist vielleicht nicht betriebsbereit.
Bitte kontaktieren Sie den technischen Produktsupport.

Any suggestions? Thanks!

PS: If I randomly install a module (i.e. "watchdog") with autoinstaller, the updates runs fine. It seems to be a bug with the automatic update functionality.

 

[IgorG]

You can find many threads on these forum where disabling epel repo did help with such cases.

 

[RobHost]

You can find many threads on these forum where disabling epel repo did help with such cases.

As EPEL is one of our central repo's, this is not a solution. In the last years this also was not a problem, why should it be from now on?
What makes the MU 58 different from the 57 before?

Thanks!

 

[Nikolay.]

Don't you think it could be EPEL that has changed, not Plesk? Try updating your installation via yum update (be careful though), then retry running Parallels Installer.

 

[RobHost]

Don't you think it could be EPEL that has changed, not Plesk? Try updating your installation via yum update (be careful though), then retry running Parallels Installer.

EPEL did not change (why should it?) and there are not updates available for these systems :/
I think Plesk should not die, even in this case. The panel is not available after this autoupdate fails.

 

[Nikolay.]

Really? Which services are not running? sw-cp-server, psa?

What does # /usr/local/psa/admin/sbin/autoinstaller --version say?

All I can suggest is going into Parallels Installer either:
1) via WebUI;
2) via console mode and selecting 11.5 release, going to components screen, then quitting.

This will update Parallels Installer to the latest version (something like 3.15.15), which might solve the problem. To revert the change if it doesn't help: # rm -f /var/cache/parallels_installer/parallels_installer*

 

[RobHost]

Really? Which services are not running? sw-cp-server, psa?

Hi,

psa ist not running anymore after update fails.

What does # /usr/local/psa/admin/sbin/autoinstaller --version say?

Parallels-Installationsprogramm 3.12.1 (built on 2012-07-03 11:48)