You guys should come here
Avg disk space used dropped with 60-70% - Node Operators / troubleshooting - Storj Community Forum (official)
Do you know which one?
The one that sats donāt send the correct daily report to nodes which makes a difference between the real avg disk usage and the one we see on nodes dashboard. I see many topics about this same confusing bug.
This is not a bug, but resources congestion on the satelliteās sideā¦ However, yes, it can looks like a bug on the storagenodeās side.
I believe it requires to implement a new feature to retry the calculations on such cases. But not sure that it should be run constantly. Perhaps running once before the payout calculations would be sufficient, but this is also mean that on storagenodes we would have a wrong estimation for a whole monthā¦
At the moment the team is aware of that and some data is recalculated and filled some gaps.
1 Like
No, I totally disagree. As we have seen that used space calculated on the node side can be totally wrong, at least the satellite side should be displaying correct values.
I mean how are SNOs supposed to run nodes without correct data on usage and payout and the chance to check if everything is running well?
There really should be no question about the correctness of the data and currently almost nothing is displayed correctly to be able to rely on.
5 Likes
I do not think thatās a good solution too. However, running it all the time is just increasing the load.
I hope our engineers will find a solution.
I just imagined if banks were like that and your balance constantly randomly changed 
Billing is billing and it must be bullet-proof accurate. Until the system achieves that, hopefully customers wonāt find out or I guess it can cause a crisis.
3 Likes
Imagine the bank balance being correct all the time, but your accounting software on your PC has a display problem.
4 Likes
Definitely one of the most annoying glitches that keeps dragging on month after month. You can say its cosmetic but it puts people on edge and weāre already there with getting half the revenue as before and all the trash data that we lost many terabytes. Test data is great but how long will that last before Iām making peanuts again per month. /rant
Thatās probably true however the issue here is not just financial but more importantly reputational. If youāre one of the engineers who developed the system and know how exactly the system works itās one thing. But if youāre a SNO or even a customer what are your options? Just trust to the engineers who developed the system? You can trust them but itās much easier to trust when the numbers help that trust rather than damaging it.
There are definitely things needing improvement. And SNOs concerns are understandable. Imagine for ex., a customer who takes 2TB DB backups every night and stores it in Storj. It eats 2TB ingress of SNOs with 0 payment. Then it sits on the disk for 24 hours and gets deleted by customer who uploads the fresh backup. So it was never downloaded thus no egress payment to SNOs and only 24 hours storage is paid. But thatās not the end of the story. It takes a few days or even longer until multiple file walkers run and complete and when finally it gets to trash it sits there at least another week. So out of about 2 weeks storage only 24 hours is paid and no traffic is paid. Not to mention that all those file walkers put extra load on the disks and donāt seem to be optimal either. For ex., instead of running mv command and just rewriting file location metadata it seems physically copying all the trash.
And when on top of that the SNO sees that statistics from satellites bite out his earnings from the charts this all can be quite demotivating.
3 Likes
100% agree. This nails it.
2 Likes
You can always use your own monitoring, it would be pretty accurate.
1 Like
It will be accounted anyway. Your node sends signed orders (three sides signed actually: the customer, the node and the satellite), and your node will be paid.
But yes, the graph on the dashboard is wrong right now, because the satellite was busy serving customers to update it in time. However, the node will be paid for whatās it provided, i.e. used storage and used egress (cryptographically confirmed) anyway.
Does the temporarily wrong graph is still worrying? Perhaps. Does it affect business and relations or maybe payouts? no
2 Likes
hm. But I should confirm - it definitely affects estimations. Is it a big problem, if it will be paid accordingly cryptographically confirmed orders anyway?
Yes it is a big problem.
why? Your node will be paid accordingly factual (you know - pay as you go) usage.
For me it is telling me if it is worth it to keep the nodes running or not.
Ahh. I see. Then itās a wrong metric. I would use my payouts instead, but well.
2 Likes
But you see the payments only after a month. Not during the month.
Therfore the estimate is very helpful.
Well, itās kind of arguable. My example above shows a case of when only 1/14 of the used storage is paid due to file walkers latency and trash retaining. In my case Iām sitting with overused disks for several days and waiting for the trash to clean up before I can get back to business.
And by the way, what triggers gc-filewalker? It doesnāt run for us1 on my system at least for a week now.