CEPH Health warnings - how to resolve?

[csjjpm]

I have configured two nodes with just one OSD each for the moment, they are 500GB NVME storage. I have some health warnings and wondered what they mean and how I should resolve them?

My nodes are 128GB RAM, 64 x Intel(R) Xeon(R) Gold 6226R CPU @ 2.90GHz (2 Sockets) each. I'm on PVE 7.0-14+

1.) "Reduced data availability: 129 pgs inactive, 129 pgs peering"
2.) "175 slow ops, oldest one blocked for 44533 sec, daemons [osd.1,mon.node1] have slow ops."

On the status I have


It has been like that for about 16 hours. I don't know how long the 'working...' is supposed to go on for. I haven't created any VMs yet.

My osd setup is


thank you
Paul

 

[itNGO]

Hi,
maybe you should first read how Ceph works, and what you need to make it work.
Example... by default Ceph creates 3 copies and needs at least 2 of them available to keep the pool writable.... you have 2 OSDs... the Default CRUSHMAP can not create 3 copies on 2 OSDs.... the Cluster can not stay online if one node fails... and so on.... read the manuals first pls.

Also consider consulting an IT-Service-Partner.... You have such a fat hosts, there should have been budget for consulting....

 

[csjjpm]

Thanks for the "expert" condescending response. I did read the manual and am aware of the min. You could just have said the errors are due to the min requirements not being met yet. I configured for 2 copies so I cannot create 2 copies on 2 OSDs. I need to create 2 more OSDs. Simples!?

Politeness costs nothing.

 

[itNGO]

Thanks for the "expert" condescending response. I did read the manual and am aware of the min. You could just have said the errors are due to the min requirements not being met yet. I configured for 2 copies so I cannot create 2 copies on 2 OSDs. I need to create 2 more OSDs. Simples!?

Politeness costs nothing.

There was nothing condescending at all... just a recommendation... if you don't like recommendations, no problem.... no more answers from my side wanted or needed..... hope you get it running as you want....

 

[csjjpm]

I apologise, I mistook you telling me to read the manuals (twice) and get someone else to do it for me as condescending.

 

[spirit]

what is your pool replication size ?

if you use size=3 (the default), it'll wait for another osd for the replication

 

[tom]

I have configured two nodes with just one OSD each for the moment, they are 500GB NVME storage. I have some health warnings and wondered what they mean and how I should resolve them?

This will not work, check the minimum requirements - 3 nodes.

 

[wigor]

I go with itNGO and tom and don´t understand your needs.
If it is a test cluster only, you can try playing with single host rule. That may fit your needs if you want to learn ceph.

 

[anthony11]

itNGO might learn how to spell Ceph before criticizing others. It still isn't an acronym.​

 

[itNGO]

itNGO might learn how to spell Ceph before criticizing others. It still isn't an acronym.​

Interesting first post.... but for your patience I fixed the letters in the posting to make it the short form for "cephalopod"...