node: Stop flushing big object when termination signal received #364

New issue

Closed

opened 2023-05-18 06:38:24 +00:00 by acid-ant · 1 comment

acid-ant commented

2023-05-18 06:38:24 +00:00

Member

Flushing big objects from the writecache may take a lot of time.
By default, we have only 90 seconds to stop node gracefully. To avoid killing by system, need to listen for closeCh and interrupt flushing as earlier as it possible.

Flushing big objects from the writecache may take a lot of time. By default, we have only 90 seconds to stop node gracefully. To avoid killing by system, need to listen for closeCh and interrupt flushing as earlier as it possible.

acid-ant added the

bug

triage

labels 2023-05-18 06:38:24 +00:00

acid-ant commented

2023-05-18 06:39:40 +00:00

Author

Member

Routines list before node killing by system.

Node version:

service@az:~$ frostfs-node --version
FrostFS Storage node
Version: v0.0.1-414-gc60029d3
GoVersion: go1.18.4

Routines list before node killing by system. Node version: ``` service@az:~$ frostfs-node --version FrostFS Storage node Version: v0.0.1-414-gc60029d3 GoVersion: go1.18.4 ```

goroutine.zip

3.9 KiB

acid-ant self-assigned this 2023-05-18 06:40:05 +00:00