-
Notifications
You must be signed in to change notification settings - Fork 97
Rename $LATTICE_COORDINATOR_IP to $CONSUL_SERVER_IP #9
Rename $LATTICE_COORDINATOR_IP to $CONSUL_SERVER_IP #9
Conversation
I think technically the $LATTICE_COORDINATOR_IP variable is really to allow diego cells' consul client to connect with the consul server; which might (currently is) running on the coordinator VM/job. Perhaps consider renaming it to CONSUL_SERVER_IP in preparation for allowing reuse of an existing consul server cluster and/or deployment of consul server outside of the coordinator job VM.
I'm not sure that LATTICE_COORDINATOR_IP is only used for consul. There are a number of things apart from consul on the coordinator box (e.g. etcd) and we don't use consul to discover all those things. |
It was the only use case in the PR. Ultimately you would use consul for all other configuration though if there were additional reasons for client-server connections? On Tue, Dec 30, 2014 at 1:42 PM, Onsi Fakhouri notifications@github.com
|
Diego itself doens't use consul to discover things like etcd, etc... and On Tue, Dec 30, 2014 at 1:47 PM, Dr Nic Williams notifications@github.com
|
ETA to stop using static IPs and support consul DNS hostnames? On Tue, Dec 30, 2014 at 1:49 PM, Onsi Fakhouri notifications@github.com
|
Where in lattice is Diego-cell using LAT_COORD_IP than for consul client? (Sorry can't spot it) On Tue, Dec 30, 2014 at 1:49 PM, Onsi Fakhouri notifications@github.com
|
re "ETA to stop using static IPs and support consul DNS hostnames? " -- not On Tue, Dec 30, 2014 at 1:53 PM, Dr Nic Williams notifications@github.com
|
This PR is just for lattice if it helps On Tue, Dec 30, 2014 at 1:55 PM, diego-edge-ci notifications@github.com
|
at the moment, you're right - the only use of the LATTICE_COORDINATOR_IP variable is for the Consul endpoint. let me review the services a little more and see this this makes sense or not. a compromise solution could be to have both variables, the CONSUL_SERVER_IP for the Consul configuration and a LATTICE_COORDINATOR_IP for (everything else) -- assuming its needed. then it would support that decoupled configuration to have the consul server not hosted on the coordinator vm. |
I think the future goal for all our cluster configurations would be to discover configuration from consul; so it's the sole static IP bootstrapping requirement. Here's to 2015 New Years resolutions! :) On Wed, Dec 31, 2014 at 12:24 PM, davidwadden notifications@github.com
|
Rename $LATTICE_COORDINATOR_IP to $CONSUL_SERVER_IP [Finishes #85279452]
+ Bump Version Files Diego Release: 0.1365.0 Diego Release Git SHA: 2be0682c4eeff45057c8e687cfca1a920a07b24d [Delivers #99785964] [Delivers #98882812] [Delivers #98882812] [Delivers #99708666] [Delivers #99457744] +Committed by GOCD- Run #9 of Pipeline: lattice, Stage: promote_and_publish, JOB: promote_and_publish
+ Update terraform example files to point to source v0.2.7-32-g9eaf1a7 + Terraform now downloads Lattice v0.2.7-31-g868be59 +Committed by GOCD- Run #9 of Pipeline: lattice, Stage: promote_and_publish, JOB: promote_and_publish
I think technically the
$LATTICE_COORDINATOR_IP
variable is really to allow diego cells' consul client to connect with the consul server; which might (currently is) running on the coordinator VM/job. Perhaps consider renaming it toCONSUL_SERVER_IP
in preparation for allowing reuse of an existing consul server cluster and/or deployment of consul server outside of the coordinator job VM.