You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Debt from the StorageID tierFS issue. As per Ariel's comment: #8549 (review)
On the dev side, I am unable to find anywhere that tests a non-empty storage ID. This might be worrying? (This request is blocking: we should test non-empty storage IDs, or state on the description of this PR why the existing tests are sufficient, or keep the issue open until we do have a PR that tests it.)
Note the comment in the design: the StorageID will currently only be allowed to be empty/nil.
What levels should we test at?
TierFS- this is the lowest level change and interesting stuff happens at this layer. Add tests
I understand that this issue is to test TierFS with nondefault StorageIDs. This is really important! Please note that #8549 is actually broader than its title makes it sound; in addition to TierFS, it adds StorageIDs to at least:
Catalog
Committed data
including import
Commits and merges
Range and metarange managers and writers
All this plumbing has to work, so we will have to test it. By deciding to pass StorageID as a parameter rather than on a context or on objects, we pretty much committed to testing all those parameters.
nadavsteindler
changed the title
Tests for non-empty string storageID tierFS
Tests + Fix for non-empty string storageID tierFS
Feb 10, 2025
I understand that this issue is to test TierFS with nondefault StorageIDs. This is really important! Please note that #8549 is actually broader than its title makes it sound; in addition to TierFS, it adds StorageIDs to at least:
Catalog
Committed data
including import
Commits and merges
Range and metarange managers and writers
All this plumbing has to work, so we will have to test it. By deciding to pass StorageID as a parameter rather than on a context or on objects, we pretty much committed to testing all those parameters.
Sorry I didn't write out all my thoughts. I want to also test CommitedManager and Catalog BUT Catalog didn't significantly change yet in that PR and CommittedData will change in the Ranges & Metaranges PR so I will write those tests there
Debt from the StorageID tierFS issue. As per Ariel's comment:
#8549 (review)
On the dev side, I am unable to find anywhere that tests a non-empty storage ID. This might be worrying? (This request is blocking: we should test non-empty storage IDs, or state on the description of this PR why the existing tests are sufficient, or keep the issue open until we do have a PR that tests it.)
Note the comment in the design: the StorageID will currently only be allowed to be empty/nil.
What levels should we test at?
The text was updated successfully, but these errors were encountered: