Hi!
i hope someone can help us out here :)
We need to move from 3 datacenters to 2 datacenters (+ 1 small serverroom
reachable via layer 3 VPN)
NOW we have a ceph-mon in each datacenter, which is fine. But we have to move
and will only have 2 datacenters in the future (that are connected, so d
Hi! 😊
It would be very kind of you to help us with that!
We have pools in our ceph cluster that are set to replicated size 2 min_size 1.
Obviously we want to go to size 3 / min_size 2 but we experience problems with
that.
USED goes to 100% instantly and MAX AVAIL goes to 0. Write operations see
hi, thank you Eugen for being interested in solving this ;)
certainly, here are some more infos:
ceph osd tree
https://privatebin.net/?db7b93a623095879#AKJNy6pKNxa5XssjUpxxjMnggc3d4PirTH1pwHQFF3Qk
ceph osd df
https://privatebin.net/?0f7c3b091b683d65#8K4KQW5a2G2mFgcnTdUjQXJvcZCpAJGxcPRc1nUYiLXz
sure!
ceph osd pool ls detail
https://privatebin.net/?85105578dd50f65f#4oNunvNfLoNbnqJwuXoWXrB1idt4zMGnBXdQ8Lkwor8p
i guess this needs some cleaning up regarding snapshots - could this be a
problem?
ceph osd crush rule dump
https://privatebin.net/?bd589bc9d7800dd3#3PFS3659qXqbxfaXSUcKot3ynmwRG2m
Hi!
Thank you 😊
Your message was very helpful!
The main reason why “ceph df“ went to “100% USAGE” was because of the crush
rule said this:
"min_size": 2
"max_size": 2
And the new “size” was 3, so the rule did not want to work with the pool.
After creating a new rule and setting the pools to th
Hi!
it is unclear for us what min_size means besides what it does. i hope someone
can clear this up :)
scenario:
size is 3 and min_size is 2
2 rooms with 100 OSDs each and this crush rule
"op": "take",
"item": -10,
"item_name": "default"
great, thank you Anthony! :)
so size 4 / min_size 2 would be a lot better (of course)
we have to stay at 3/2 for now though, because our OSDs are filled 60% in sum
maybe someone can answer additional questions:
- what is the best practice to avoid a full OSD scenario, where ceph tries to
recre
thank you Robert!
this sounds like split brain again... but we have a quorum system by using 3
monitor nodes. so only the room with the majority of the ceph-mons is available
for I/O.
if the room with the majority of the ceph-mons is the one that is cut-off, I
suppose we'd need to do this:
go
hi!
thanks to all of you, I appreciate this very much! I will have to go through
all of your messages a few more times and do some research.
so our rule from the intial post does make sure, that, when 1 room goes down it
does NOT try to restore 3 replicas in the remaining room but it will only