This message was deleted.
# harvestera
adamant-kite-43734
05/20/2023, 10:25 AMThis message was deleted.
b
bumpy-portugal-40754
05/20/2023, 9:29 PMYou can't. Only replicas (=copies) are possible on different nodes.
o
orange-lion-57504
05/20/2023, 9:48 PMUhh.. Then what things can I do with a cluster with multiple nodes?
b
bumpy-portugal-40754
05/20/2023, 10:19 PMHarvester is a VM solution and Longhorn its (redundant) storage layer. It's intended to create highly available and redundant volumes which continue to work when one server is down for any reason. This is why the default number of replicas is 3. Compare this with raid level 0 and 1. Longhorn does raid 1 (with 3 copies). If you want to have raid 0, you need to look for something different.
o
orange-lion-57504
05/20/2023, 10:20 PMOkay so clustering it's just for if one machine fails, another can do the job?
orange-lion-57504
05/20/2023, 10:22 PMAnd another question, why when I create a volume with node1 and node2 selected it says ''replica shchedule failed''?
b
bumpy-portugal-40754
05/20/2023, 10:26 PMIn Harvester, yes. Maybe there is some way to achieve what you want with some Longhorn trick. You should ask in the Longhorn channel!
For the other question: You kept the default replica number of 3? Then the volume will never become available... no node to place the 3rd replica.
o
orange-lion-57504
05/20/2023, 10:27 PMyes, number of replicas is 3. If i have only 2 nodes, it should be 2?
b
bumpy-portugal-40754
05/20/2023, 10:30 PMYes, 2 or even one if you don't care about redundancy.
o
orange-lion-57504
05/20/2023, 10:31 PMokay
orange-lion-57504
05/20/2023, 10:31 PMand also, when creating a storage class I should select the 2 nodes and the 2 disks, right?
b
bumpy-portugal-40754
05/20/2023, 10:35 PMIf your cluster has only 2 nodes, choosing 2 for the replicas should be enough.
o
orange-lion-57504
05/20/2023, 10:37 PMokay
orange-lion-57504
05/20/2023, 10:48 PMThis would work?
test1: node1
test2: node2
disk1: disk from node1
disk2: disk from node2
b
bumpy-portugal-40754
05/20/2023, 10:50 PMTry it out. I can't, I don't have 2-node-clusters. Did you try to keep the defaults and just change the number of replicas?
o
orange-lion-57504
05/20/2023, 10:51 PMyes
orange-lion-57504
05/20/2023, 11:19 PMIt doesn't work : (
orange-lion-57504
05/20/2023, 11:19 PMstill says replica schedule failed
orange-lion-57504
05/20/2023, 11:27 PMI removed the tags (disk1,disk2,test2,test1) from storage class and now works. It is doing the replica, right?
b
bumpy-portugal-40754
05/21/2023, 12:00 AMI don't know what you're doing. And since you don't post lots of information, the only way is guessing.
When you set the storageclass correctly, that's the first step. Next is to create a VM which then allocates a volume from Longhorn automatically.
o
orange-lion-57504
05/21/2023, 8:59 AMThey should do a tutorial of how to do all of this
g
great-bear-19718
05/22/2023, 12:01 AMits hard for me to keep up with all that has been done
great-bear-19718
05/22/2023, 12:01 AMa support bundle would be nice for us to figure out why the replica scheduling is failing
o
orange-lion-57504
05/22/2023, 5:26 AMYeah
2 Views