cluster performance degradation

[waltar]

A replication is a data sync without using the new one while a migration is using the new data.
Used data is hot data and has prio in cache which reduces I/O.

 

[UdoB]

I wanted to understand why ZFS, if I remove a disk, the pool is no longer reachable and it gives me suspended. Shouldn't a raid 10 with 6 HDDs also work with 3 less disks?

The problem arose in mirror-0, look at post #130: one device is REMOVED and the other one has 138/244 read/write errors.

And - of course - only one device is allowed to fail in a mirror. The moment the survivor delivered errors the pool stopped working --> suspended for maintenance...

For sure the timing - the sequence of events - is important, but this is my impression.

In the same #130 your wrote:

I tried to remove an HDD and the pool went SUSPEND,

It is not a good idea to remove one twin while the other part of the mirror has reportedly errors. A scrub would have possibly repaired that. To scrub later on will leave damaged data in some files ( with a high probability) - but the pool in itself should be usable again.

 

[toto-ets]

I added the third node, now I will do all the other tests, but now I have this problem, I can't delete this HA, it gives me this error, I don't see it configured in the VM, what command do I have to send to delete it

 

[UdoB]

I can't delete this HA, it gives me this error, I don't see it configured in the VM, what command do I have to send to delete it

I do not know - it seems you are drifting further into problematic areas while pve1 is still missing.

Fortunately "Quorum: OK" is stated --> "/etc/pve" should be writable. So if you want to work below the official surface: take a look into /etc/pve/ha/resources.cfg. If you are brave you may edit that file (make a backup-copy first!) and remove the probably present "vm: 101" stanza completely. This manual and for sure unsupported direct manipulation should be visible in the WebGui after a very few seconds.

Good luck.

 

[waltar]

In that list you see just the available vm/lxc which could be given a resource !!
So it's normal that you cannot remove it as for this vm no ha is defined yet

 

[toto-ets]

look now, pve1 active, and vm on pve 1, same error, and as if the vm does not exist, but in reality it is there..

 

[toto-ets]

In that list you see just the available vm/lxc which could be given a resource !!
So it's normal that you cannot remove it as for this vm no ha is defined yet

if I try to create it it gives me this

 

[toto-ets]

I do not know - it seems you are drifting further into problematic areas while pve1 is still missing.

Fortunately "Quorum: OK" is stated --> "/etc/pve" should be writable. So if you want to work below the official surface: take a look into /etc/pve/ha/resources.cfg. If you are brave you may edit that file (make a backup-copy first!) and remove the probably present "vm: 101" stanza completely. This manual and for sure unsupported direct manipulation should be visible in the WebGui after a very few seconds.

Good luck.

in this file, i can't find anything, it's empty

 

[toto-ets]

I solved it, I added another vm to AH, then I deleted it, and the one from vm 101 also disappeared, it's becoming a proxmox odyssey for me