Reduce gateway failover detection interval #348
Conversation
|
🤖 Created branch: z_pr348/sridhargaddam/leader-interval |
In an HA setup where multiple nodes are labelled as Gateway nodes, when the active SubmarinerEngine/Node goes down, it takes about 15+ seconds for one of the backup Submariner Gateway PODs to become active. This patch makes the leader election values configurable with some defaults that reduces this interval. Signed-off-by: Sridhar Gaddam <sgaddam@redhat.com>
Done |
sridhargaddam
force-pushed
the
leader-interval
branch
from
ff4bc8c to
9334c3f
Compare
|
🤖 Updated branch: z_pr348/sridhargaddam/leader-interval |
| @@ -43,6 +43,19 @@ func init() { | |||
| flag.StringVar(&localMasterURL, "master", "", "The address of the Kubernetes API server. Overrides any value in kubeconfig. Only required if out-of-cluster.") | |||
| } | |||
|
|
|||
| type leaderConfig struct { | |||
| LeaseDuration int64 | |||
There was a problem hiding this comment.
you can also annotate with default:"5"
and the module will handle it, but I'm fin with it now, we need to release.
There was a problem hiding this comment.
Actually, I tried it and it did not seem to work when I tested it. So, I modified it this way.
There was a problem hiding this comment.
oh, ok, makes sense then. I guess the lib is buggy
| err := envconfig.Process(leadershipConfigEnvPrefix, &gwLeadershipConfig) | ||
| if err != nil { | ||
| klog.Fatalf("error processing environment config for %s: %v", leadershipConfigEnvPrefix, err) | ||
| } |
There was a problem hiding this comment.
I wonder if we should start using a ConfigMap for this type of config ...
There was a problem hiding this comment.
There are many params that can be moved to a ConfigMap. Is it ok if we take it up in a separate patch?
There was a problem hiding this comment.
I think it’s something we should do, and also the secrets. There’s an open issue for that. May be something we can do after this release
|
🤖 Closed branches: [z_pr348/sridhargaddam/leader-interval] |
In an HA setup where multiple nodes are labelled as Gateway
nodes, when the active SubmarinerEngine/Node goes down, it takes
about 15+ seconds for one of the backup Submariner Gateway PODs
to become active. This patch makes the leader election values
configurable along with some defaults that reduce this interval.
Signed-off-by: Sridhar Gaddam sgaddam@redhat.com