storage/disk: bundles issue 4868 #4877
Closed
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
srenatus
commented
Jul 11, 2022
storage/disk/disk.go
Outdated
| @@ -351,17 +347,20 @@ func (db *Store) Truncate(ctx context.Context, txn storage.Transaction, params s | |||
|
|
|||
| // update symlink to point to the active db | |||
| symlink := filepath.Join(path.Dir(newDB.Opts().Dir), symlinkKey) | |||
There was a problem hiding this comment.
I've included #4870 to unbreak my workflow, please disregard.
srenatus
commented
Jul 11, 2022
| return ret | ||
| } | ||
|
|
||
| return []file{f} // try it anyways |
There was a problem hiding this comment.
That's the 馃-it-might-still-work-if-we-try code path. We might still fail, but maybe we don't, because the smallSize cutoff value isn't based on any calculation, but a guess.
Before, we'd failed to activate a bundle with a large data.json which would end up being attempted to load in a single write. Now, we'll recognize if the value is larger than a threshold, and split it into multiple blobs, which hopefully go in with a single txn. Activating the bundle in question works, but re-activating it afterwards still fails in a different place: eraseBundles accumulates too many writes in a single txn, attempting to reset the data in store.) Signed-off-by: Stephan Renatus <stephan.renatus@gmail.com>
srenatus
force-pushed
the
sr/storage/disk-bundles-issue-4868
branch
2 times, most recently
from
917f5c5
to
d6150b4
Compare
Before, we've accumulated too-large transactions by retrieving data keys according to their partitioning, and deleting them one-by-one. Now, we'll do the same thing, but in an iterator passed to Truncate, which deals with restarting transactions. This also means that we've got an extra call to Truncate, so one Truncate operation more than before. However, we've been having one Truncate per bundle already. To fix this, we could combine multiple storage.Iterator into a chaining iterator, and only do one Truncate operation, but that's a future optimization. Signed-off-by: Stephan Renatus <stephan.renatus@gmail.com>
srenatus
force-pushed
the
sr/storage/disk-bundles-issue-4868
branch
from
d6150b4
to
4577fad
Compare
|
馃挕 tests fail because the logic is wrong: calling Truncate in where it's now called will loose the transaction previously written to -- i.e. the delta bundles' patches are applied, but never written to newDB. They're an open txn for oldDB, but that doesn't matter when truncate switches DBs. I'll redo those bits. I think I'm at the point where I'm this close to not understanding anymore how the current setup works 馃槄 |
|
TBC |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
馃毀 馃懛 PR for discussion purposes (#4868)