Check if you have these logs:
Hey, Iâm following the discussions around test data, garbage collection, etc. Iâve just used the script from @BrightSilence which I havenât for a while:
REPORTED BY TYPE METRIC PRICE DISK BANDWIDTH PAYOUT
Node Ingress Upload -not paid- 13.41 TB
Node Ingress Upload Repair -not paid- 122.10 GB
Node Egress Download $ 2.00 / TB (avg) 390.62 GB $ 0.78
Node Egress Download Repair $ 2.00 / TB (avg) 455.23 GB $ 0.91
Node Egress Download Audit $ 2.00 / TB (avg) 96.65 MB $ 0.00
Node Storage Disk Current Total -not paid- 24.43 TB
Node Storage â Blobs -not paid- 24.43 TB
Node Storage â Trash â -not paid- 0.00 B
Node+Sat. Calc. Storage Uncollected Garbage ⤠-not paid- 11.45 TB
Node+Sat. Calc. Storage Total Unpaid Data <ââ -not paid- 11.45 TB
Satellite Storage Disk Last Report -not paid- 12.98 TB
Satellite Storage Disk Average So Far -not paid- 11.62 TB
Satellite Storage Disk Usage Month $ 1.49 / TBm (avg) 10.23 TBm $ 15.25
________________________________________________________________________________________________________+
Total 10.23 TBm 14.38 TB $ 16.94
Estimated total by end of month 11.62 TBm 15.76 TB $ 19.17
Do I read this right that my node has 11 TB of uncollected garbage sitting around?
The dashboard shows this:
What to do then?
Maybe to add: It is a Synology DS1019+ with 5x 14TB running in SHR with two drive fault tolereance. Also with 1TB SSD cache.
The node updated to v1.105.4 a bit less than two days ago.
Unfortunately, yes. Though keep in mind this is an estimate. Itâs calculated by comparing the last reported storage usage by the satellite against your local node usage. Sometimes the last reported usage from the satellite is not entirely reliable, though in those cases it has always been lower than real usage, which would only lower the amount of uncollected garbage. This also relies on the file walkers on your node. If you have disabled the filewalker, it may use unreliable local data as well. That said, I have seen big amounts of uncollected garbage on my end as well and this seems to be an issue that has lasted for a while now. I guess Iâm glad I added this metric to surface that information.
Thanks. So since the last updated my Synology is also very very busy⌠whenever the filewalker is active / going through it I have 100% drive busy. This is one month:
Iâve read about how much it should delete etc. / but 11TB is a bit less than 50% of what I have allocated and it is just sitting there doing nothing. I have like 5 TB more space I could allocate but I donât see a reason - it might slow down even more (everything is working fine and responding) but what I donât understand: Why is not âfasterâ to delete the garbage?
Nothing. Until the team would backfill gaps, your Avg Used Space would be way off.
However, it doesnât mean that you have such amount of garbage, because your node doesnât have a full picture right now.
Please also note, it will be paid for the used space and the bandwidth, which your node submitted as a signed orders to the satellites. So, everything is cryptographically confirmed and it will be paid for the all used space and bandwidth.
Of course, if you have errors related to gc-filewalker or retain, you have a problem with the garbage. If you do not have any errors, then you do not have any problem with GC.
Thanks Alexey,
Iâve been part of storj since the beginning, so I am reading along but also donât understand it all. Couple of questions:
- So youâre saying I donât have 50% garbage on my disks. How is it possible to find out how much is for real?
- Or how could I check that nothing is really broken somehow?
- I have more space available that I could allocate but because the load / time the filewalker and what not go through I wasnât sure if I should just allocate more space. Also as it looks it juts gets stuffed and not deleted in time
And yeah I read about TTL etc.
- In my logfiles I see primarily updater stuff, which command to check gc-filewalker?
Thanks!
It is not possible. Only the satellites would know but those donât/canât tell your node timely. It can take weeks until they do. I suspect this might be an architectural scalability issue. As the amount of data stored in the network grows the satellites struggle more and more to cope with it.
You can only check that nothing is broken on your nodeâs side. I would start with Filewalker status - #5 by Alexey
If all your nodes are full and ingress has stopped I would allocate more. Otherwise I doubt it will make any difference.
Yes, this is the biggest issue that SNOs keep complaining about in several threads.
And likely would happen again until this feature would be implemented:
However, SLC still might send a report later.
having more used space for the whole month, the âaverage disk spaced used this monthâ is lower⌠is there any reason? Both v1.108.3
Please check, does all satellites sent a report about usage to your node?
the SN was not sending orders
I added the storage2.orders.sender-interval
variable to the config file, restarted and it sent them all (>5000 orders per satellite).
Thanksss
What was the previous value in your config file?
there was no value in config file same config file since years ago
so i guess it used the default value.
Then the restart itself fixed an issue I guess.
Average usage dropping since Saturday like crazy again.
Are we having a satellite issue again?
You may select each satellite to see which one is missing.
I am seeing my overall average dropping almost every minute or so. This doesnât seem normal.
US⌠on many nodes.
Oh⌠againâŚ