Recent days, I noticed a very high repair, ingress and egress wise. In my case 505GB egress/repair, 15GB egress/usage and 310 ingress/repair, 471GB ingress/usage. There were times with high repairs in the past, but this ratio is “new” also for ingress (in my case).
Did you noticed similar behavior?
What is out repair and in repair?
I guess “in repair” are the pieces “moved” to your node from other ones. But what is “out”?
The repair egress from their nodes.
We adjusted RS numbers several months ago, so now it’s noticeable.
The repair workers downloads missed pieces.
Date | GET_AUDIT | GET_REPAIR | PUT_REPAIR | GET | PUT
===========+============+============+============+============+=============
2026-09-03 | 42952 | 157030 | 454423 | 1962259 | 837478
2026-09-04 | 52356 | 137041 | 367339 | 2511068 | 718060
2026-09-05 | 56140 | 333556 | 770160 | 2351905 | 883165
2026-09-06 | 66803 | 551956 | 1266504 | 2836181 | 1231079
2026-09-07 | 75056 | 490340 | 1084581 | 3191220 | 960957
2026-09-08 | 77789 | 463571 | 1129989 | 3805527 | 952881
2026-09-09 | 80006 | 763994 | 1458308 | 3959563 | 980490
2026-09-10 | 85794 | 1007237 | 1851661 | 4414135 | 1235652
2026-09-11 | 90576 | 1319653 | 2195952 | 4587262 | 1146359
about 300 nodes was DQ some days ago
There is 23800 nodes offering free space in global network atm. So most of “right sizing” still ahead.
This is the count of repair get and put operations, not storage volume so it may be skewed, but the count seems to have increased a bit on my end. Going back even a few months I don’t see a jump quite this high.
I’m not sure if this is actual node departure or the RS changes mentioned, my nodes all seem to be both taking in and sending more repair traffic.
12 were mine. Unfortunately I turned off about 200 TB of data. 
Same here, 16 nodes 186TB currently in ‘graceful exit’ with 19 days left.
The Storj developers have created a fantastic product but with the total disasters on the management side, both the old and the new with Inveniam in the background, I have lost all confidence in Storj.
It could be a usual wave of a repair traffic. Nodes can lost data due to many reasons. It usually happens over time. So today you are likely seeing a repair traffic for the data uploaded a few years ago.
It does feel like ingress+egress are both up: the repair system seems to be doing its job. Thank you to everyone who used GE!
Though I’m not sure bloom filters are working? I kinda expected to see more trash too…
Bloom filters are not related to repair. You don’t delete the original (your) piece just to upload a new piece to a new node (mine).
Edit: because someone will explain that repair isn’t moving pieces around, I know, just using a simple example and not typing out a thesis each time I reply, explaining everything in excruciating detail, being historically and factually accurate etc etc.
but we see that clients also delete data
Yeah, Eleks reports show paid-data on a gentle decline. So I expected nodes to track and constantly be deleting: but maybe my nodes are superheros who win every race! 
I see a nice growth of both repair and normal traffic for my nodes. I guess the latter is because the fastest nodes are also most expensive and therefore shutting down now and making my potatoes more competative.
Yeah I also saw a moderate uptick in repair data the last few days.
We enabled an automatic chore to delete the pending deletion accounts, we also reduced time periods a lot, especially for a billing freeze; this allow us to cut costs a little more.
