No, I cannot see the future, but I don’t have reasons to believe that this won’t be handled properly. It won’t make sense to leave data wasting space on storage nodes, would it?
I would expect that delete messages will not be sent to storagenodes, at least it was so when the Stefanbenten satellite was decommissioned.
It’s quietly expensive and resource consuming operation. The best case is to send several bloom filters to allow to remove this data by the garbage collector.
But you know - I’m better to remove it myself than using a “lazy” garbage collector (for weeks likely).
Oh absolutely, there is no reason to actively delete chunks from satellites, it’s literally wasted efforts.
But there shall be some ways for storage nodes to prune orphaned data – I thought garbage collection could do/does this for removed satellites.
I’m the opposite. I’d rather let software figure it out than having to mess with internals manually risking breaking things. It’s in the STORJ’s interest to avoid wasting space on nodes with garbage data, so I trust it will be done properly. For me as a storage node operator it makes no difference; it’s node’s internal business.
When you delete millions of small files manually, in my experience (DQed satellite), Windows explorer is frozen. Disk utilization is at 100%. Even calculating folder size takes forever with usual process. There is zero progress shown for deletion unlike usual Windows delete. This creates a fear that disk might not be responding to Audits or there could be Audit failures due to timeouts.
The only way I could delete DQed data was to delete 5 to 10 folders at a time, wait for it to finish then select next batch of folders.
Please this post if you would like node software to handle the deletion.
Do not use a Windows Explorer for that
It’s better to use a CLI, or Wsl2
rm -rf /mnt/x/storagenode2/storage/blobs/pmw6tvzmf2jv6giyybmmvl4o2ahqlaldsaeha4yx74n5aaaaaaaa from wls2 was much faster than
rm -Recurse -Force x:\storagenode2\storage\blobs\pmw6tvzmf2jv6giyybmmvl4o2ahqlaldsaeha4yx74n5aaaaaaaa from PowerShell, I do not know why
Me, too. Can it be deleted already? Is there anything to take into consideration besides just deleting directories? GE the satellites or so?
As already stated in the first post
Sending the heldback amount and removing data from databases like pieceexpiration.db.
For that you need bloom filters sent by the satellite. So you need the satellite to still exist. It’s not too complicated to do clean up afterwards, so at this point I suggest we just wait and see what Storj Labs comes up with. I’m sure they will figure something out or provide instructions. If not, I will, after the satellites have been decommissioned.
I GE both sats, just to end the suspense. I have a question: why the exited sats still show scores in dashboard? Shouldn’t display an OFF or 0, or something?
This is not implemented, satellites and scores will be here until you remove these satellites from the trusted list or add to an exclusion list.
Has the files deletion started?
Huge spike in deletion on some of my nodes
You can check, what satellite is invoked deletion in your logs.
1TB in trash from “12rfG3sh9NCWiX3ivPjq2HtdLmbqCrvHVEzJubnzFzosMuawymB@europe-north-1.tardigrade.io” in the past 48h on a 15TB old node
Perhaps related to this announcement.
Yep, I have 1.3TB trash on a 23TB node. Looks like clean up has started. Saves me from having to expand soon, I guess.
Looks like someone push the trigger and forgot about this part
Personally I’m happy with the deletion of those test data.
It was already announced in the QA section of this post close to the end.
I read this post a few days ago, and i though they will annonce a schedule about deletion also, not just the shutdown. But anyway it as started and it’s as planned