Implement container db folder cleanup on startup

[tmpolaczyk]

Container database folders are normally removed when the collator gets unassigned. However, if the collator stops the node right before being unassigned, and then starts it later, the container db will not be removed. The db will be removed the next time the collator is assigned and unassigned from that container chain, which may never happen.

To fix this edge case, this PR introduces a new db folder cleanup functionality. When the node starts, it will check the container folders and delete all of them except the ones it is assigned to as a collator. This check is only performed once, the first time a finalized block is imported.

[github-actions]

WASM runtime size check:

Compared to target branch

dancebox runtime: 1412 KB (no changes) ✅

flashbox runtime: 832 KB (no changes) ✅

dancelight runtime: 2008 KB (no changes) ✅

container chain template simple runtime: 1088 KB (no changes) ✅

container chain template frontier runtime: 1388 KB (no changes) ✅

[github-actions]

Coverage Report

(master)

@@                     Coverage Diff                      @@
##           master   tomasz-db-folder-cleanup      +/-   ##
============================================================
- Coverage   65.47%                     65.34%   -0.13%     
  Files         310                        310              
+ Lines       54041                      54174     +133     
============================================================
+ Hits        35379                      35395      +16     
+ Misses      18662                      18779     +117     

Files Changed	Coverage
/client/service-container-chain/src/spawner.rs	40.94% (-3.50%)	❗
/container-chains/nodes/frontier/src/command.rs	18.02% (-0.11%)	❗
/container-chains/nodes/simple/src/command.rs	17.79% (-0.11%)	❗
/node/src/service.rs	15.54% (-0.02%)	❗

Coverage generated Fri Nov 15 10:35:18 UTC 2024

[girazoki]

When the node starts, it will check the container folders and delete all of them except the ones it is assigned to as a collator.

What happens if I stop it while it is syncing a chain (2000) but is assigned to another one (2001), then I boot it up again and we are still in the same situation? would I remove 2000?

[girazoki]

looks good from a code perspective!

[tmpolaczyk]

What happens if I stop it while it is syncing a chain (2000) but is assigned to another one (2001), then I boot it up again and we are still in the same situation? would I remove 2000?

"assigned" means "current or next", so obviously it won't be removed

[ParthDesai]

Code looks good.