-
Notifications
You must be signed in to change notification settings - Fork 5.5k
Enable conservative out-of-bound snapshot cleaning #8811
Enable conservative out-of-bound snapshot cleaning #8811
Conversation
| } | ||
|
|
||
| #[test] | ||
| fn test_accounts_purge_chained_purge_after_snapshot_restore_complex() { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I understood this mystical test finally...
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Moreover, https://github.com/solana-labs/solana/pull/8811/files#diff-2099c5256db4eb5975c8834af38f6456R1497 made this previous form of this test fail.
d372bdd to
4cc6cee
Compare
Codecov Report
@@ Coverage Diff @@
## master #8811 +/- ##
======================================
Coverage 80.0% 80.0%
======================================
Files 265 265
Lines 57477 57481 +4
======================================
+ Hits 45998 46011 +13
+ Misses 11479 11470 -9 |
sakridge
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
lgtm
|
I ran v1.0 validator with this PR patched overnight, and it seems that it still produces valid snapshots. Overall, size reduction is 10x and restore-time reduction is 20x. That's amazing. |
|
Also, snapshot creation-time reduction is 20x (10min -> 30s): grafana-testnet-monitor-edge-ryoqun-without-out-of-bound-snaphost-cleaning.pdf |
|
That is amazing!! |
* Enable conservative out-of-bound snapshot cleaning * Add tests (cherry picked from commit 4bbf09f)
|
I also tested against the SLP cluster. And there was no problem, too. |
Problem
There is no codepath for cleaning unused storages in snapshot when starting a validator
Summary of Changes
Our past efforts has paid off; Now just need small plumbing and test clarification.
And, I'm now very conservative with good reason of newly-introduced bugs with the adventure of #8168. So, this PR doesn't contain aggressive strategy, originally had in #8337.
Let it rip little by little. :)
Not codewise, but this PR requires #8724
Split from #8337
Towards #8168