Let me know what you think!
Nothing. What is your dashboard estimations?
Maybe this also did not help:
If the problem was that bloom filters weren’t effective enough on large nodes before… maybe we’re just seeing the first effects of the improved filters? (they’re deleting data now that should have been deleted long ago from those large SNOs?)
Yes, maybe. And if so it could be the solution why satellite has different used space than node has occupied.
I have a 4yo node, with about 25Tb of data, and I still have not received any bloom
10% of the entire network went in the trash in one day?
I hope we are considering Storj Select cSNOs too.
I believe since the Free Tier is being discontinued on March 31, a lot of people migrated their storage out and we’re just seeing the effects of that recently, although in the discontinuation announcement it seems that the deletion of free-tier data from SNOs is being done on a controlled basis.
The last garbage collection wave was on Mar 17 so a lot of trash to accumulate within that period as well.
1 Like
Please note that there may be a significant number of Free Tier users that opted to delete their data themselves off the platform in the last few days before the April 1 deadline. We can only control the timing of when we will delete data left on the platform from abandoned not empty accounts after they have been suspended due to not having added a payment method (after April 1.) After account freezing, we will give those account holders extra time to decide if they want to add a payment method to either continue using our service or to recover any data they wanted to keep, before we would mark the data they left behind for final deletion.
4 Likes
I believe the network statistics are showing the data stored on the network from the satellite perspective, not what is actually stored on the nodes.
Bloom filters are being distributed just to clear the data that was deleted from the network already, and this would not influence the network statistics in any way I think.
There apparently are some issues with generating bloom filters, as distribution of these was quite erratic in the last couple of months. In my case I’m storing significant amounts of data that as per the satellites should be trashed already, and this is going on for couple of months now.
This sudden stats decline must really be someone moving significant amounts of data away from the network. If someone was storing such amounts of data here, it would be paid already and thus no reason to move them away before the Apr/1st, unless someone was combining multiple free tier accounts to store it all here for free. I also don’t think all the people that won’t upgrade to paid tier would leave the migration for the last couple of days, so this really must be an individual or some group of people doing this. But of course this might be totally unrelated to free tier change and just a coincidence.
1 Like
That widget is fixed to 30d no matter what, and changing the time window on the main dashboard only changes the sampling rate.
To see the ingress over time, click the title of that widget or the three dots and then select Explore. You can then change the time window inside that.
1 Like
I this this is just new reality, about half of the data was wiped out, I think it is not the end.
If that’s real, then they did not listen:
But it is hard to believe.
That’s gotta be a bug. In the Mar 27 Town Hall it was reported ~12PB of paid data. With erasure coding this needs to be inflated by 80/29 = 2.75 so 33PB checks out before this large drop.
If this is true, it implies:
- Currently, we are storing petabytes of data on our nodes without any compensation due to the malfunctioning Bloom filter.
- Nodes that have been expanding over the years are now facing profitability issues. If the Storj team does not implement an urgent payout, we might witness a widespread, GE (or not GE, just shut down) from the network.
1 Like
We are f****ed. 
Still there is hope it is just a bug. (Like so many others surfacing currently).
4 Likes
So many bugs for last month.
1 Like
