Yes in my case I received two bloom filters for US1, causing a few nodes to hit 100% utilization on a single core (not iowait) like in this post: High CPU usage by nodes caused by receiving a new bloom filter while still processing the old one. I didn’t want to interfere so I just let it run and eventually the first bloom finished I believe on day 10, whereas the trash filewalker kicked in on day 8. At this point I just restarted the node to stop the second GC from finishing. I had also deleted the bloom filters from the retain folder, as I already collected 2+TB of trash.
Now I see the following:
The node had restarted.
GC has appeard to resume, meaning that it created a new trash date folder with current name.
But when I look now there is only a single subfolder in it (the one that should be resumed with) and nothing else. And all stored bloomfilter files from the retain folder have vanished. This means that the bloomfilter for the US-1 satellite has vanished which was under processing before node restart as well as the 2 bloomfilters from EU-1 satellite are now gone.
Not a problem. It is a bit of a race condition (whether a given piece was moved to trash quickly enough to be caught by trash collector), but worst case is that a piece would be collected one day later.
2 Likes