New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
feat: rework record sync for improved reliability #1478
Merged
Merged
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
The latest updates on your projects. Learn more about Vercel for Git ↗︎
|
So, to tell a story 1. We introduced the record sync, intended to be the new algorithm to sync history. 2. On top of this, I added the KV store. This was intended as a simple test of the record sync, and to see if people wanted that sort of functionality 3. History remained syncing via the old means, as while it had issues it worked more-or-less OK. And we are aware of its flaws 4. If KV syncing worked ok, history would be moved across KV syncing ran ok for 6mo or so, so I started to move across history. For several weeks, I ran a local fork of Atuin + the server that synced via records instead. The record store maintained ordering via a linked list, which was a mistake. It performed well in testing, but was really difficult to debug and reason about. So when a few small sync issues occured, they took an extremely long time to debug. This PR is huge, which I regret. It involves replacing the "parent" relationship that records once had (pointing to the previous record) with a simple index (generally referred to as idx). This also means we had to change the recordindex, which referenced "tails". Tails were the last item in the chain. Now that we use an "array" vs linked list, that logic was also replaced. And is much simpler :D Same for the queries that act on this data. ---- This isn't final - we still need to add 1. Proper server/client error handling, which has been lacking for a while 2. The actual history implementation on top This exists in a branch, just without deletions. Won't be much to add that, I just don't want to make this any larger than it already is The _only_ caveat here is that we basically lose data synced via the old record store. This is the KV data from before. It hasn't been deleted or anything, just no longer hooked up. So it's totally possible to write a migration script. I just need to do that.
ellie
force-pushed
the
ellie/array-not-list
branch
from
January 3, 2024 18:52
64a9e77
to
5dbaeab
Compare
ellie
force-pushed
the
ellie/array-not-list
branch
from
January 5, 2024 17:46
cb327eb
to
0e8bd74
Compare
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.
So, to tell a story
KV syncing ran ok for 6mo or so, so I started to move across history. For several weeks, I ran a local fork of Atuin + the server that synced via records instead.
The record store maintained ordering via a linked list, which was a mistake. It performed well in testing, but was really difficult to debug and reason about. So when a few small sync issues occured, they took an extremely long time to debug.
This PR is huge, which I regret. It involves replacing the "parent" relationship that records once had (pointing to the previous record) with a simple index (generally referred to as idx). This also means we had to change the recordindex, which referenced "tails". Tails were the last item in the chain.
Now that we use an "array" vs linked list, that logic was also replaced. And is much simpler :D
Same for the queries that act on this data.
This isn't final - we still need to add
The only caveat here is that we basically lose data synced via the old record store. This is the KV data from before.
It hasn't been deleted or anything, just no longer hooked up. So it's totally possible to write a migration script. I just need to do that.