i have checked the forum and it seems clear the /24 rule limits the ingress traffic for nodes in same subnet. Just wanted to know is this rule applicable for egress traffic also
no - 20 letters needed
2 Likes
But this dosen’t mean that you will get double the egress if you run 2 nodes in the same subnet, because the data that should go out as egress is split on 2 nodes, and is the same data that should go on one node, if you only run one node.
The only advantage you have running 2 nodes instead of one, if they are on 2 drives of course, is that you win more races, because the workload is split between them. I’m not sure how big is this advantage in practice though.
But you burn twice the energy.
1 Like
I’d say more tangible benefit of running multiple small nodes on the array is ability to keep deleting the nodes to reclaim space as needed (remember, storj is designed to take advantage of underutilized resources, so its expected you’ll need to actually utilize all or some of those at some point) in smaller increments, because today nodes can’t shrink. And if they could — it probably would not be fast enough.
actually you can shrink it, but slowly. You may set the allocation either to 0b or just less than used.
I thought this will just cause the node to stop accepting the data, and it will only shrink if storj customers delete existing data. Which is non-deterministic, and may in fact never happen. It’s not usable if you need space back soon, because you don’t know when will the space be returned, if ever.
Did this change? I.e. does the node proactively repair the data off to other nodes, (like partial exit), if the space is overused?
It doesn’t proactively remove data, unless your node were offline more than 4 hours. But even then it’s not direct, only if the number of healthy pieces of the segment is below the repair threshold, so not a guarantee either.
1 Like