Fix handling shards dbs purge checkpoints in mem3_rep #5832
+103
−26
Conversation
This file contains hidden or 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
Previous PR [1] failed to account for shards db itself. Shards db (`_dbs`) is
managed differently than regular shard copies. Its `mem3:shards(Dbs)` result is
a single element shard list with a `#shard{}` having `node = node()` and `range
= [0, ff..]`. They are replicated in a ring across all nodes, we expect to find
a purge checkpoint pushing changes to the "next" node in a ring only.
[1] #5827
nickva
added a commit
that referenced
this pull request
Dec 24, 2025
In the previous PR [1] we added special handling for shards dbs since it has a
custom ring replication topology. In PR [1] we considered only the checkpoint
for pushes from the current node to the "next" none in the ring. However, we
should also consider the checkpoints created by the "previous" node when it
pull purges from current node, so that what we fix in this PR.
As a reminder, a replication job from node A to node B will:
1) pull purges from B to A (checkpoint on B with a B->A purge checkpoint doc)
2) push purges from A to B (checkpoint on A with a A->B purge checkpoint doc)
3) push do updates from A to B (checkpoint on A with A->B sync checkpoint doc, and
on B also with an A->B sync checkpoint doc)
[1] #5832
Contributor
Author
|
There is continuation PR after this which adds handling for "previous" purge checkpoints for when the previous nodes in the ring replicates to us. When it pulls purges from "us", it also create its own checkpoint going from current node to the previous node. I added the pervious node consideration in a new PR #5833 |
nickva
added a commit
that referenced
this pull request
Dec 25, 2025
In the previous PR [1] we added special handling for shards dbs since it has a
custom ring replication topology. In PR [1] we considered only the checkpoint
for pushes from the current node to the "next" none in the ring. However, we
should also consider the checkpoints created by the "previous" node when it
pull purges from current node, so that what we fix in this PR.
As a reminder, a replication job from node A to node B will:
1) pull purges from B to A (checkpoint on B with a B->A purge checkpoint doc)
2) push purges from A to B (checkpoint on A with a A->B purge checkpoint doc)
3) push do updates from A to B (checkpoint on A with A->B sync checkpoint doc, and
on B also with an A->B sync checkpoint doc)
[1] #5832
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
Previous PR [1] failed to account for shards db itself. Shards db (
_dbs) is managed differently than regular shard copies. Itsmem3:shards(Dbs)result is a single element shard list with a#shard{}havingnode = node()andrange = [0, ff..]. They are replicated in a ring across all nodes, we expect to find a purge checkpoint pushing changes to the "next" node in a ring only.[1] #5827