This query using experimental text search results in docker container stopping #1947
Labels
community
community
Effort - Unknown
Effort - Unknown
Frequency - EveryTime
Frequency - EveryTime
Reach - VeryFew
Reach - VeryFew
Severity - S1
Severity - S1
Memgraph version: docker image memgraph/memgraph-mage:1.16-memgraph-2.16
Environment: Controlled by docker-compose.yml
The bug
I'm attempting to use the experimental procedure text_search.search_all() in a rather complex query. When I run the query in Memgraph Lab, I get a message saying "Query failed: Connection was closed by server", and the docker container stops. The docker container does not log any errors through docker, nor are any errors written to the memgraph log file.
To Reproduce
Setup:
Start up a new docker container (and new volumes) with experimental text search enabled. Connect to it with Memgraph Lab and run the following statements one by one in order:
Full query
This is similar to the query that I want to run, and it crashes the docker container every time.
parameters:
query:
At this point, after attempting the above query, my docker container is no longer running. I can start the same container back up and run simple queries fine, but the above query will always make it crash again.
Simplified query
I tried simplifying it down to figure out what part of the query was causing the problem, but the simpler I went the more sporadic the problem became. Here's an example of a query that (with the database set up the same way with those 4 nodes and 6 indexes) can crash the docker container, but only does so rarely (maybe two percent of the time).
parameters (same as before):
query:
If you're having trouble getting that query to produce the bug, then try running it a lot in quick succession right after the docker container starts up.
Expected behavior
I expect the query to either work or give a meaningful error, and I expect the docker container to stay running regardless of whether the query worked or not.
Side note
I know that text search is very experimental right now, but I was so excited to use it that I wanted to try it out in a realistic environment. It's understandable that it has some issues, and I just hope that my bug report can help you to fix the bugs and to know that you have at least one user that really wants this feature. Thank you for all your work.
The text was updated successfully, but these errors were encountered: