I worked hard to reduce my setup to one node per HDD (I originally started with two nodes on a single HDD). However, after 30–35 months, the nodes are only utilizing 8–10 TB each, having reached a balance between incoming data and deleted data and I’m starting to reconsider my approach. With Badger enabled, I’m thinking it might be time to go back to running two nodes per 20 TB HDD and shutting down half of the disks that are wearing out.
Since I’m using ZFS, it will be quick and easy to revert to a single-node-per-HDD configuration if Storj traffic patterns change in the near future.
Pls stop me if you have good and reasonable insights or alternatives
I do not see any reason to run two nodes on the one HDD, they will inevitable affect each other because they would compete for the same resource, it’s also against ToS.
Two nodes receives the same amount of data as only one node, so it doesn’t make sense to have two nodes and also on the same HDD.
Why? I think it’s useful to mention not only ToS, but also that it would reduce efficiency of each node, because the resource is not infinite, the HDD can do only 200 IOPS, running two IO consuming services will reduce this small amount even more, so each of the node will be slower than if they were run each on own HDD.
My question was about two nodes. Different ip. Same HDD. Against ToS.
Today I think is possible to run 2 “full speed” nodes on 1 hdd without loosing performance (with enterprise 20tb hdd). Filewalker can slow down a bit but with badger ON will be just 2 days maximum of slowdown.
yep, in this case even twice - likely bypassing /24 rule and using the same HDD for two nodes.
Highly not recommended setup.
You may be sure only if they would be run separately and would have a comparable success rate and the same long run for filewalker. I personally not so sure about not loosing performance when you use two nodes on the same HDD. It could be possible on SSD, where the random access time is almost the same (or the difference is negligible).
