Resolved rpm db corrupts during any plesk update

[KingSky]

Centos 6
Plesk Onyx 17.8.11

I recently noticed that when I went to Plesk > Tools & Settings > Plesk Components, ALL items were showing "The component was not installed."
clicking "Refresh" did not correct this issue.

I found out via trial and error and Google that my rpm db was corrupted. Whenever I would try to run any plesk update it would stall out or give me the
"rpm db
Thread died in Berkeley DB library
dbenv->failchk: DB_RUNRECOVERY: Fatal error, run database recovery"
error

I found this kb article which partially helps. Plesk autoinstaller fails on yum operations and there are many components marked as not installed

If I follow the first 3 commands, after I run "plesk sbin packagemng -sdf" and check Plesk, all my services are showing up again under Plesk Components.

But when I run the 4th command; "plesk installer update --repatch" it again corrupts the stalls out, rpm db is again corrupted and the components list goes back to none installed.

I've triede this command as well after rebuilding the db: "
plesk installer --select-release-current --reinstall-patch --upgrade-installed-components"
it also fails and corrupts the db

Thanks for any input!
-Kevin

 

[Bitpalast]

Is your database type and version compatible with Plesk Onyx? Or could it be possible that it was upgraded from a formerly approved version to something new in the past?

 

[KingSky]

I would assume so? I've been running Plesk Onyx for over a year now without having this issue.

 

[Bitpalast]

My question was really bad in this case. Of course this does not have anything to do with the "database" server. Sorry for that. I don't know why the package manager of the operating system has an issue there. Maybe when you run the plesk patch it displays errors that could point you to the right cause?

 

[KingSky]

That's one of the issues stymying me, I don't get any errors, it just stalls and never proceeds. I have paid attention to the point it stalls at and found that it is when it is checking the gpg keys of any repository.

I opened a support ticket (Codero hosting) and they said that at some point they got a NOKEY error from Plesk. I asked how they got this error since it stalls out for me but they never specifically answered.

This was their first reply:

I've backed up, removed, and repaired the RPM DB which allowed yum to run, but I still received the NOKEY error on the Plesk packages. Unfortunately I was unable to locate a support article or location to reinstall the Plesk rpm key. I was able to process updated by running "yum update -y --nogpgcheck" however. I have the yum updates running now and I will let you know if I see any issues or errors during the process.

Unfortunately the yum update run with the no gpg check also failed but since they ran it and I didn't, I didn't get to see if there was an error message.

 

[KingSky]

I was considering trying to update a component myself using the --nogpgcheck just to see if I get an error, rather than the full yum update.
For instance I know there is an update to mysql-server

 

[KingSky]

I think (hope) I solved this finally.
Steps I took:
-In "Plesk > Tools and Settings > License Management" I did a "Retrieve Keys"
-I then ran "yum package-cleanup --dupes"
-Next I ran "plesk installer update --repatch --nogpgcheck" (this finally completed without issue)
-I was then able to run "yum update" with no stalls or errors, although at this point it didn't find any updates to install.

I haven't tried running the Onyx to Obsidian update process, that's next.

 

[KingSky]

I ran the Plesk upgrade and it warned me that there were duplicate packages installed.

Back to command line and I ran "package-cleanup --dupes" and let it remove the duplicates.

After that I was able to succesfully run the upgrade via the GUI and it completed successfully! HUZZAH!