Hi all,
I’m wondering if anyone can provide information around dedicated disk / setup,
what counts as a dedicated disk / what doesn’t
changes needed to the docker run?
can be implemented on already running nodes not having dedicated disk enabled ?
can dedicated disk be reversed?
Pros/Cons …
Thanks in advance !
Dedicated disk means the node isn’t tracking the used space/trash in databases but instead is checking how much space is free on the drive configured as the destination for the data directory. You can configure a free space threshold - how much space to leave free on that drive. Once this threshold is met, the node stops receiving customer uploads.
As things get deleted, it will start accepting the uploads again, and until the threshold it met again.
The pros are no need for used-space-filewalkers, no need to restart the node after the partition/LVM is extended, no problems with free space being reported incorrectly by the node and the node stopping receiving uploads, even if there is still free space on the drive.
The cons are just the broken dashboard probably and no stats, such as how much trash the node currently stores.
It can be enabled/disabled at will on any node.
Can’t help with Docker, but on binary installs it is just the configuration file edit.
nice - what in the config do you amend ? - i can translate that to docker
how would you know how much you are getting paid on per drive ?
does payout calcuation still work on the dashboard?
cheers
These lines:
# use whole disk
storage2.monitor.dedicated-disk: true
# how much size to reserve as always free
storage2.monitor.reserved-bytes: 25 GB
Payout calculation still works.
It is still one drive per node as the node is determining the drive it is using from the data directory path. And using that just to determine the free space on it, the rest is the same.
Perfect information- thank you.
I should have used this along time ago in this case 
@zip - do you happen to know if it creates a performance increase ?
I can imagine it would - there is no need to run used-space-filewalker, no need to update databases on read, delete etc.
But have no benchmarks to prove it.
I guess with hashstore there is no benefit from using dedicated disk anymore.
Ooo interesting to know ! I turned it on 1 of my nodes but makes me anxious not knowing what’s going on
It would depend on the configuration.
Database accesses are by default a lot of small reads and a lot of sync writes. Cumulative IO exceeds that of actual useful work done by storagenode.
In naive configuration that will make ginormous difference between overwhelming or not overwhelming disk.
But in a correct configuration, where sync is off and small reads are cached — the amount of IO hitting disks is so miniscule that even 10x increase of overal traffic would not make any difference in what hitting the disks.
So no.
I have DBs , hashstore etc on SSD and single disk literally just for storage
any way to enable it on docker?
pack of nodes on my FC SAN on dedicated volumes.
okay, will answer ty myself/ these parameters can be passed (appended) with command line on docker container creation .
--storage2.monitor.dedicated-disk=true \
--storage2.monitor.reserved-bytes=25GB
You may use both features in the same time 
how is payout calculated without storage knowledge? - reliant , on the satellite information?
I think so. However, it is not a payout, it is an estimation anyway.
The real payout is calculated on the satellites at the beginning of the next month accordingly provided orders by your node for the past period.