Cannot unlock file "/home/runner/.cache/mongodb-binaries/5.0.7.lock", because it is not locked by this process

[jgcmarins]

We've been experiencing some random test failures with this library and GitHub Actions.
Not sure which is causing the issue, but here's the error:
Cannot unlock file "/home/runner/.cache/mongodb-binaries/5.0.7.lock", because it is not locked by this process

[EmanuelCampos]

@jgcmarins you solved this?

[jgcmarins]

@EmanuelCampos not yet

[guilherme-gm]

I am experiencing the same issue, which I found very weird since I have been using this for months with no errors.

I tried following the mongdb-memory-server library and I think I may have found the culprit.

The only place this error may be throw is at:
https://github.com/nodkz/mongodb-memory-server/blob/master/packages/mongodb-memory-server-core/src/util/lockfile.ts#L273

default here could only mean lockedDifferent. If we look into checkLock function, the only place where this value is returned is at:
https://github.com/nodkz/mongodb-memory-server/blob/master/packages/mongodb-memory-server-core/src/util/lockfile.ts#L126

And isAlive returns true only when the PID in the file (I guess) is not the same of the running process:
https://github.com/nodkz/mongodb-memory-server/blob/master/packages/mongodb-memory-server-core/src/util/utils.ts#L163

Thus, I think whenever there are multiple process running, there is a chance for this to happen. I am not sure why this never happened for 5 months and is now happening a lot though...

I don't know if there is a way to fix without changing mongodb-memory-server, also I am not sure why it works like that.

I will attempt to make jest workers using threads instead of child process, as this hopefully will make all of them have the same PID, and hopefully I don't get errors again. I will try to update here in a few days.

[aboruchovas]

@guilherme-gm any update? i'm also struggling with this

[guilherme-gm]

I've only made a few CI runs since then (~5, I think), but using 1 Child Process seems to work.

Using Worker Threads instead of Child Processes did not work. It spawned several instances of mongo memory server at the same time and it was a mess.

I changed back to Child Process and limited to 1 worker when running in CI. It didn't fail in the few runs I had after that and I didn't notice any major performance impact on my tests (I don't have THAT many tests, so a few extra seconds in CI was not a problem)

I think this makes sense, since having a single process spawning mongo servers, running tests and shutting it down will prevent a case where another process is using the lock. But I still think it is something that should be addressed (maybe on mongodb-memory-server and not on jest-mongodb... but I am not 100% sure).

I don't remember the config from the top of my head, but should be easy to find it in jest docs. I will try to update this message tomorrow once I have access to my code.

Edit:
This is the setting I am using:

maxWorkers: (process.env.CI === 'true' ? 1 : '50%'),

I chose to only limit in CI, since I don't have issues when running locally (so far)