You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Restoring from state-sync one mainnet (with a patch for #8325 applied), is very slow (more than 30 minutes), and uses more disk than expected (double). It turns out the culprit is the WAL mode which doesn't like an almost 40GB transaction.
To Reproduce
import-kernel-db with a mainnet state export.
Expected behavior
Data directory not using more than the target size and faster than current 30 minutes.
Screenshots
2023-11-09T17:16:08.076460335Z Restoring SwingSet state from snapshot at block height 12416000 with options {"exportDir":"/tmp/agd-swing-store-restore-12416000-702237452","artifactMode":"replay","exportDataMode":"all"}
2023-11-09T17:49:58.197655501Z 5:49PM INF restored swing-store export exportDir=/tmp/agd-swing-store-restore-12416000-702237452 height=12416000 module=x/swingset submodule=SwingStoreExportsHandler
$ ls -la
total 77078156
drwxr-xr-x 2 benchmark benchmark 89 Nov 9 17:16 .
drwx------ 9 benchmark benchmark 167 Nov 9 17:16 ..
-rw-r--r-- 1 benchmark benchmark 37754957824 Nov 9 17:49 swingstore.sqlite
-rw-r--r-- 1 benchmark benchmark 76513280 Nov 9 17:43 swingstore.sqlite-shm
-rw-r--r-- 1 benchmark benchmark 39391472472 Nov 9 17:42 swingstore.sqlite-wal
The text was updated successfully, but these errors were encountered:
Describe the bug
Restoring from state-sync one mainnet (with a patch for #8325 applied), is very slow (more than 30 minutes), and uses more disk than expected (double). It turns out the culprit is the WAL mode which doesn't like an almost 40GB transaction.
To Reproduce
import-kernel-db with a mainnet state export.
Expected behavior
Data directory not using more than the target size and faster than current 30 minutes.
Screenshots
The text was updated successfully, but these errors were encountered: