Playing around with state consolidation #32
Conversation
davissp14
changed the title
| for _, members := range cluster.Members { | ||
| if members.ID == id { | ||
| return nil | ||
| } | ||
| } |
There was a problem hiding this comment.
An interesting state that I haven't fully though about is the case where a member is registered, but for whatever reason the hostname, region, org primary fields might be different. I don't think this should be possible though?
There was a problem hiding this comment.
I think it's probably safe to assume they will be different. If they aren't, something is broken with Machines.
| if errors.Is(err, ErrCAS) { | ||
| c.RegisterMember(id, hostname, region, primary) | ||
| } |
There was a problem hiding this comment.
possible infinite loop in the case of consul misbehaving? feels unlikely though
There was a problem hiding this comment.
Good catch. We don't want to be spamming Consul if it's under load. I'll see about addressing this.
There was a problem hiding this comment.
Actually, this should be fairly safe. We will only retry in the event of a CAS error, which should be super rare on its own.
There was a problem hiding this comment.
We will only retry in the event of a CAS error, which should be super rare on its own.
ah cool yeah. I'm not familiar with consul failure modes so I wasn't sure if a degraded cluster can cause CAS errors or something. Seems like no.
davissp14
force-pushed
the
cluster-state
branch
from
b47d36e to
d51f57c
Compare
Few notable changes:
{ "members":[ {"id":396610339,"hostname":"fdaa:0:2e26:a7b:bf88:6920:c8e0:2","region":"lax","primary":true}, {"id":1949217720,"hostname":"fdaa:0:2e26:a7b:7d16:819f:c83e:2","region":"lax","primary":false}, {"id":2119071075,"hostname":"fdaa:0:2e26:a7b:9adb:c779:c868:2","region":"ord","primary":false} ] }