Thanks for the heads up, I appreciate that! We’ll see how much more that cleans up.
that sounds interesting, ill keep an eye on my metrics… sure did create a bit of a workload for a short time, it’s always interesting to see how well it copes.
Looks like my node dumped another 60 GB from europe-west at around 01:30 UTC Oct 2 (this morning), bringing the total to 200 GB.
Over 500 GiB of trash on that satellite. I always get the answer that I was offline too long - looks like I was offline for a month or something. It happened several times that there was a whole bunch of trash generated in a matter of hours.
got another 20gb trash this morning putting me at 69gb trash, didn’t seem to cause much of a spike, i suspect it was deleted over two rounds maybe…
or atleast i got two spikes in my graphs that seem out of place iowait going to 8-10% very similar spikes and unaccounted for…
but had just finished a scrub last night, so maybe some sort of after effects of that…
635GB trash total now, looks like my node got the big clean up part 2 as well.
Noice
It was not that you were offline, rather than a very big amount of zombie segments that have been cleaned up.
@littleskunk above explained why. The difference in amount of data is simply explained by the node age.
All of this trash data is from files older than 2019-11-01. Nodes that joined the first minute, got most of the data.
Did that happened several times before too? Like several months ago.
how did the workload look while deleting all that?
It happened a while back on the stefanbenten satellite as well. And these two recent times last weekend and this weekend.
Took about 2 hours, but it barely impacts my system. The SSD cache helps out a lot with these processes. I see an IO spike on the underlying array, but node performance doesn’t suffer at all.
will usually be some kind of effect with additional load… even if unable to be observed…
but yeah i know what you mean…
how does your successrates look on an avg day btw?
That’s weird because I started my node in may 2020 and still got an extra 10GB of trash this morning.
Must’ve been when they implemented it or something, it was a while ago. We were told you get significant garbage collection if you are offline a lot. So the zombie segment reaper actually deletes through garbage collection, not through deletes. Explains why there were several instances of hundreds of GB of trash out of the blue before.
That can still be zombie segments from normal customers.
mine is from march and still have a bit… and the satellite ran until very recently … so not really that odd…
@ general topic
in this rainbow happy wonderland we now live in where anything can be anybody and vice versa…
i take offense of all this trash talk…
well this is new… can’t remember having seen 0 trash since when i started my node…
and wasn’t trash suppose to stay longer than 2 days… because my last added trash was the 20gb only 2 days ago…
or is something just displaying it wrong… or did the storj team purposely delete it… (i suppose the latter would make sense)
still got 108GB in trash. but it was 220GB a few hours ago.