Hi all,
I have three Windows nodes on 1.62.3 with storage2.piece-scan-on-startup: false but when I restarted the machine, the filewalker process kicked off regardless - any ideas?
How did you determine that the process was kicked off? Knowing that will help me track down the code paths that were run and figure out what’s going on.
I was looking at my disk utilisation through the task manager under performance. Before restarting they were around ~20%, after restarting they shot up to 100% for a while, and then went back to normal
I think there are a bunch of reasons why the node may use more disk i/o than just the piece scan during startup. Can you check for the log line "Startup piece scan omitted by configuration" at the Info level from the piecestore:cache service? It should log that if it’s skipped.
4 Likes
The initial piece scan is the biggest process that goes over all pieces. There are still 2 other processes that will do something similar.
- On startup and every 24 hours a cleanup job will run over all pieces in the trash folder and delete pieces that are older than 7 days.
- Executing garbage collection (triggered by a message from the satellite) will run the filewalker over all pieces for one satellite.
If you want to get some more details about runtime and stuff you can check the debug endpoint: Guide to debug my storage node, uplink, s3 gateway, satellite
4 Likes
That’s a good one, I did see that line there, so it’s safe to assume that as @littleskunk had mentioned it was other factors causing the load upon restart.
Sorry for the false alarm everyone!
4 Likes
Your feedback and questions are always welcome!
5 Likes