Perf/Parallelize Pruning #7541
Merged
Perf/Parallelize Pruning #7541
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
|
Could we go for a higher cache with this? Would it make sense to? |
|
Only 4 time actually. Not much. Trying different key. The write rate is now noticable at rocksdb level, like 75k writes per sec. Thats like 7.5% snap speed. |
|
I mean, that is average, assuming that only 10% of the time is doing pruning, at that pruning time, the write rate is 750k per sec. So, not much headroom left. |
|
Is fast 👀 
|
LukaszRozmej
reviewed
Oct 2, 2024
benaadams
reviewed
Oct 3, 2024
|
4+x is very impressive, but it sounds like it is not saturating 16 core CPU, what is the current bottleneck? |
|
About 40% of it is during persist, where it is disposing batch. The only way to make it faster seems to be to disable wal, but I don't recommend that and its not that much faster. The other part are things such as looping the shards, removing entries and such. |
|
|
The persist (without wal) just dump it into memtable. Already multithreaded. The actual writing to disk is already in background. I don't think other part can be parallelized futher. Maybe it can be micro optimized somehow, but its already paralellized reasonably well. |
|
Am full syncing mainnet, up to block 9,097,554 (16hrs) with a 4GB CacheMB (x4 recommended)
No issues so far; StateDb 105 GB |
|
Also just switched to Parallelize persist cache commit; was on prior |
benaadams
approved these changes
Oct 3, 2024
|
Had an invalid block issue after repeated shutdown and restarts; not sure if related to this or another change I have in-flight |
|
Looking at logs I had a memory prune while shutting down |
|
Wonder why its an invalid block instead of a trie exception. |
|
Do prewarmer or something hide a trie exception? |
First error was "insufficient sender balance"; so assume it was reading a non existing account? |
|
Ok this is different change; not this one |
benaadams
approved these changes
Oct 4, 2024
|
|
Think going too fast now #7549 |
|
Full pruning results in missing nodes on restart. But happens on master also. |
LukaszRozmej
approved these changes
Oct 4, 2024
8 tasks
LukaszRozmej
approved these changes
Oct 4, 2024
rjnrohit
pushed a commit
that referenced
this pull request
Oct 10, 2024
Co-authored-by: Ben Adams <thundercat@illyriad.co.uk> Co-authored-by: Lukasz Rozmej <lukasz.rozmej@gmail.com>
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.
Changes
Types of changes
What types of changes does your code introduce?
Testing
Requires testing
If yes, did you write tests?
Notes on testing