Race condition when running ES delete_by_query

[bsofiato]

Description

When running the unit test associated with code search (file modules/indexer/code/indexer_test.go), I ran into the issue in the screenshot below.

As explained here, this is due to a racing condition where ES uses an outdated index while looking for the documents to delete. The safe approach is to ask ES to refresh the index before.

I wrote a PR that changes the behavior when trying to delete the documents of a repo. If the deletion results in a conflict, Gitea will refresh the index and retry.

P.S. ES is running inside a docker container spun by the command below:

docker run --rm --memory="4g" -p 127.0.0.1:9200:9200 -p 127.0.0.1:9300:9300 -e "discovery.type=single-node" -v "$(pwd)/data/elasticsearch:/usr/share/elasticsearch/data" docker.elastic.co/elasticsearch/elasticsearch:7.16.3

Gitea Version

0b1b403

Can you reproduce the bug on the Gitea demo site?

No

Log Gist

No response

Screenshots

No response

Git Version

No response

Operating System

No response

How are you running Gitea?

I ran into this issue while running the tests inside VSCode

Database

None