Tasks DB not cleaned up after `ipcluster stop` about ipyparallel

[minrk]

It is the intended behavior for the task database to persist because it allows result retrieval even after the Hub has died.

[minrk]

You can clear the task database with a Client when you are done, if that's what you want:

client.purge_hub_results('all')

[jakirkham]

So, I tried starting up the cluster and running client.purge_hub_results('all') and also tried running client.purge_everything() to no avail. To be clear, this is in regard to data from a previous cluster session. Is there another way I should be doing this?

[minrk]

which db backend are you using?

[jakirkham]

SQLite

[minrk]

Thanks, I'll investigate why it's not doing the right thing. It may be something that I've forgotten about.

[minrk]

Can you share your tasks.db file, perhaps? I just tested this with master, and verified that running the Hub for a while, stopping it, starting a new one, and calling Client.purge_everything() does delete everything in the task database.

[jakirkham]

Hmm...sharing the database may be tricky. It often gets quite large during a run so I would have to think of the best way to share it. Also, there is the concern of non-published data being in the database. Let me think of what would be best to do this.

[minrk]

That's perfectly understandable. I'm just trying to think of the best way to see what's up, since it seems to be cleaning up over here.

You can perhaps try doing some basic SQLite browsing of the file, to see about tables/rows, etc. before/after purge.

[minrk]

Hi! I’m going through and cleaning up old/stale issues on this repo. Feel free to open a new Issue if you are still having this trouble.