From https://www.drupal.org/project/lingotek/issues/3055788#comment-13389986:
lingotek_post_update_lingotek_content_metadata_job_id() crashed our updb because LingotekContentMetadata::loadMultiple() tries to load like 100,000 entities. However, we also only have NULL job_id, so it doesn't do anything in our case. Instead of loading all lingotek content metadata, we should filter it down ahead of time, and probably use some batch processing.
Also worried about whether I have too much lingotek content metadata, and whether I should have job_ids.
| Comment | File | Size | Author |
|---|---|---|---|
| #6 | 3100900-postupdate-batches-6.patch | 5.17 KB | penyaskito |
| #5 | 3100900-5-entity-query.patch | 911 bytes | charginghawk |
Comments
Comment #3
penyaskitoComment #4
penyaskitoComment #5
charginghawk commentedHere's an initial stab at a solution. My database doesn't actually have any of the bad job IDs, so I don't know if it solves that problem, but it doesn't try to load all entities any more.
Comment #6
penyaskito@charginghawk Thanks for the patch! Looks like the regex query is not working.
I've created a patch that process them in batches, and the same for the other related post-update function for config.
Comment #7
penyaskitoTests passed as expected.
Comment #9
penyaskitoCommitted b34fc84 and pushed to 8.x-2.x. Thanks!