api/server: Allow noop objstorage to be used in development mode
AbandonedPublic
Actions

Authored by ardumont on Dec 7 2021, 11:13 AM.

Details

Reviewers

None

Group Reviewers

Summary

Providing the SWH_DEV_ENVIRONMENT is set, allow to actually use the noop objstorage.
This can be useful for example in docker environment.

Test Plan

tox

Diff Detail

Repository

rDOBJS Object storage

Branch

master

Lint

No Linters Available

Unit

No Unit Test Coverage

Build Status

Buildable 25414
Build 39728: Phabricator diff pipeline on jenkins	Jenkins console · Jenkins
Build 39727: arc lint + arc unit

Event Timeline

ardumont created this revision.Dec 7 2021, 11:13 AM

Herald added a reviewer: Reviewers. · View Herald TranscriptDec 7 2021, 11:13 AM

Build is green

Patch application report for D6759 (id=24540)

Rebasing onto b7adf3823c...

Current branch diff-target is up to date.

Changes applied before test

commit 63330f48f765c587c05d06b0fcfd55eac6c4e579
Author: Antoine R. Dumont (@ardumont) <ardumont@softwareheritage.org>
Date:   Tue Dec 7 11:11:04 2021 +0100

    api/server: Allow noop objstorage to be used in development mode
    
    Providing the SWH_DEV_ENVIRONMENT is set, allow to actually use the noop objstorage.
    This can be useful for example in docker environment.

See https://jenkins.softwareheritage.org/job/DOBJS/job/tests-on-diff/71/ for more details.

Harbormaster completed remote builds in B25414: Diff 24540.Dec 7 2021, 11:16 AM

ardumont requested review of this revision.Dec 7 2021, 11:16 AM

ardumont added inline comments.Dec 7 2021, 11:21 AM

swh/objstorage/api/server.py
240	feel free to propose me some better name, i usually fall short... maybe SWH_BYPASS_CHECK is fine enough?

I don't think you should add these checks. We also don't check the in-mem objstorage backend, and it loses data just as much as this one.

I don't think you should add these checks. We also don't check the in-mem objstorage backend, and it loses data just as much as this one.

We are not consistent indeed. Like i said in the description, some are tested some are not [1]
I'm fine either way but it'd be best to decide and conclude then stay consistent afterwards.

[1] https://forge.softwareheritage.org/source/swh-scheduler/browse/master/swh/scheduler/api/server.py$0-115

It's more trouble than it's worse IMO. "noop" needs to be explicitly written in config files, so I think it's unlikely to happen by accident.

ardumont mentioned this in D6780: api/server: Do not prevent noop objstorage to be used as rpc server.Dec 8 2021, 9:01 AM