In our environment we need to keep job accounting history indefinitely. However, the job environment and script need only be kept accessible for the duration of time typical for support personnel. This is typically 2-4 weeks, or for some projects up to 8 weeks. However, we would always want to purge the job scripts and environment after the project activity diminishes. Please consider adding PurgeJobEnvAfter and PurgeJobScriptAfter functionality to the slurmdbd logic. We have done direct MariaDB SQL hacks (as suggested by SchedMD personnel) to address table growth, but would strongly prefer a supported slurmdbd-level fix. Thank you.
Thanks for filing this, I'm moving it into the enhancement queue. I'll discuss internally and get back to you with a timeframe, but this should be relatively straightforward of an NRE project. - Tim
Hello, There is renewed interest in this feature as it is manifesting on more of our clusters. Would you please generate an SOW proposal to implement this feature? Thank you.
SoW should be on it's way. (Updating some of the metadata on the ticket while here.)
*** Ticket 16746 has been marked as a duplicate of this ticket. ***