Skip to content

Comments

Don't error during replay when deleting a table with no in-memory instance#3892

Merged
gefjon merged 2 commits intophoebe/ignore-st-meta-insertsfrom
phoebe/replay-handle-delete-of-nonexistent-table
Dec 17, 2025
Merged

Don't error during replay when deleting a table with no in-memory instance#3892
gefjon merged 2 commits intophoebe/ignore-st-meta-insertsfrom
phoebe/replay-handle-delete-of-nonexistent-table

Conversation

@gefjon
Copy link
Contributor

@gefjon gefjon commented Dec 16, 2025

Description of Changes

Based on #3891 . Begin reviewing from commit 4370b69.

Prior to this commit, we treated it as an error (in fact a panic, prior to an earlier commit of mine today) when deleting a table during replay which didn't have an in-memory instace. This should not have been an error, as it's not problematic, and it can occur in valid commitlogs when a table is initially empty and is never used prior to its deletion.

With this commit, don't error.

API and ABI breaking changes

N/a

Expected complexity level and risk

1

Testing

  • Manually replayed a database with such a deleted table. Got an error prior to this commit, succeeded replay with this commit.

…tance

Prior to this commit, we treated it as an error
(in fact a panic, prior to an earlier commit of mine today)
when deleting a table during replay which didn't have an in-memory instace.
This should not have been an error, as it's not problematic,
and it can occur in valid commitlogs when a table is initially empty
and is never used prior to its deletion.

With this commit, don't error.
@gefjon gefjon force-pushed the phoebe/replay-handle-delete-of-nonexistent-table branch from 8513a43 to 9edc066 Compare December 17, 2025 16:53
@gefjon
Copy link
Contributor Author

gefjon commented Dec 17, 2025

Rebased onto updated #3891 .

# Description of Changes

Based on #3892 ; begin reviewing at 21b3d1d.

I'll expand on this description later.

# API and ABI breaking changes

N/a.

# Expected complexity level and risk

3 - we already rewrote this path once or twice before and thought we'd
fixed it.

# Testing

- [x] Manually replayed commitlog which contained an automigration that
changed multiple column types from the same table.
@gefjon gefjon merged commit c294f7d into phoebe/ignore-st-meta-inserts Dec 17, 2025
17 of 21 checks passed
aasoni pushed a commit that referenced this pull request Feb 5, 2026
Based on #3892 ; begin reviewing at 21b3d1d.

I'll expand on this description later.

N/a.

3 - we already rewrote this path once or twice before and thought we'd
fixed it.

- [x] Manually replayed commitlog which contained an automigration that
changed multiple column types from the same table.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants