01-18-2017 12:25 PM
Hi, I got stuck in a pretty difficult situation.
I wanted to upgrade my cloudera manager from 5.7.3 to 5.9 and launched stop services with the CM UI.
However I stopped the services but the active commands haven't been finished.
So I receive an error message when I want to start the new version of cloudera manager server. I saw in the logs of cloudera-scm-server that I need to go back to the older version (5.7.3) to abort or finish these commands.
The issue is I can't find anywhere the archive or tarballs of this version of cloudera... https://archive.cloudera.com/cm5/redhat/6/x86_64/cm/5.7.3/ doesn't exist...
For information, the command service cloudera-scm-server force_start doesn't work and don't kill the active commands... How can I do? Is there a place where I can get the archive of CM 5.7.3? Or is there a way to delete the active commands? I'm not sure but I think there are written in the amon database in the embedded cloudera manager server database.
Thanks for your response.
01-18-2017 04:46 PM
Sorry you hit this issue. I am not certain what happened to Cloudera Manager 5.7.3, but it was likely pulled due to a critical issue.
In your situation, then, there is a way to work around this as long as you are OK with the commands that are running being killed without completing.
If so, you can run the following with the CM 5.9:
# server cloudera-scm-server force_start
This tells Cloudera Manager to forcably terminate any running commands and start. If you have any questions or concerns about the commands that are still running, you can post them here as that information should be visible in the error. Note that the error also mentions using force_start (as you should see in your /var/log/cloudera-scm-server/cloudera-scm-server.log.
Most times it is OK to kill the commands as your cluster is down anyway, but proceed with caution. We can't tell if it is safe without seeing the commands that are running.