Reindex all jobs take too long to start #3075
Labels
Features
Issues related to building and enhancing features
Priority - Medium
Issues that should be prioritized ahead of low but not immediately critical - bulk of work cycles
Ready for Development
The issue has passed review from teams and is ready to be worked on
Descriptive summary
Currently a reindex is starting and it has taken an unacceptably long time to spool up. It claims
5675709.1ms
to query the head of Fedora (though it took closer to several days) and has now been running for21272:51min
, or just under 15 days, without finishing partitioning the URIs. Before any jobs can be submitted to SideKiq, this partitioning has to be completed, so no real work has begin and we're 15 days behind content that would need to be reindexed.We need to find a way to reindex as we find URIs to reindexable assets.
Expected behavior
Reindexing jobs are queued as URIs are found, not after all URIs are found and partitioned.
Alternatively, reindexing is just generally sped up.
Related work
#2760
#2651
The text was updated successfully, but these errors were encountered: