Do you have free space on your node?
You can use the multinode dashboard to check; it displays the available space being reported to the satellite.
Is your version actual? https://version.storj.io/
Yes I have free space on my disk.
I don’t have multinode dashboard; any other way to do it? command line?
Yes my storagenode version is the latest.
Do you have “overused” on your dashboard?
Nah; here’s my dashboard.
The data keeps being deleted month by month. Like 10% per month.
I have these in my config.yaml. It’s too old that I can’t remember why. Anything wrong with these?
filestore.write-buffer-size: 512KiB
healthcheck.details: false
healthcheck.enabled: true
pieces.enable-lazy-filewalker: false
pieces.file-stat-cache: badger
No, they are not related to absent uploads. Are you sure that you do not have any uploads? Or do you have them, but not so much?
@nyancodex
Check the neighbors to see if you are alone in the subnet.
Stop node, remove node, delete all db-es, start node.
If you didn’t already, activate active and passive hashstore migration.
Wait for it to finish. Stop, remove, delete databases, disable badger cache, start.
Check the system for malware with a good antivirus, not the crappy free ones.
Check the drive’s stats, SMART, bads, etc.
Causes of low performance (some): neighbors, databases with bad info, piecestore, viruses, hardware problems.
Now, many of us moved to hashstore and the nodes win more races than piecestore nodes (in theory).
I have but not much; the ingress still shows data. But it seems that the amount of deletion is way much higher than new data.
May I ask how to activate active and passive hashstore migration? Thank you.
2 Likes
How many Neighbors ?
2 Likes
thank you I’ll try and update here
no neighbor; let me try with hashstore; thank you.
but is the active migration set as default in the latest version @Alexey?
Yes, but you can always disable it and remain on price store for now.
1 Like
I’m not aware about the active migration. Maybe only Saltlake?
The passive migration should be enabled though, i.e. all new pieces going to hashstore, all requested pieces should be moved to hashstore.
So, It could be half on piecestore and half on hashstore at the moment.
But I can enable the active migration for all the old pieces? Like moving everything old and new to hashstore?
Yes you can. The link to wiki was already posted above.
2 Likes
I have enabled the passive and active migrations. How about these options? I can leave them as default or I have to set them?
Leave at defaults.
I would also delete all zero length files from blobs folder. They are invalid, and will cause migration to never complete.
Then after no more pieces are being migrated in every chore invocation — review logs from last chore. You will likely find blobs folder empty. If not - remaining few files will be corrupted, with corresponding message in the log about failure to migrate them.
At that point you can delete all under blobs folder and stop the chore.
2 Likes