Bryan Boreham

If the program doesn’t become ready until registered, and if you have `minReadySeconds` set, then the rollout will slow down until the additions do succeed. I’m still interested to know...

Similar: ``` --- FAIL: TestRing_ShuffleShardWithLookback_CorrectnessWithFuzzy (0.00s) --- FAIL: TestRing_ShuffleShardWithLookback_CorrectnessWithFuzzy/num_instances_=_9,_num_zones_=_3,_update_oldest_registered_timestamp_=_false (3.23s) ring_test.go:1849: random generator seed: 1697620659695190631 ring_test.go:1967: subring generated after event 70 is expected to include instance instance-62 from ring state...

More detail required on why you think `promauto` is involved. Most of the errors you point to don't even include dskit in the stack.

I've done a series of further tests scaling up, down and creating new clusters, and could not repeat this result.

From a [Slack conversation](https://weave-community.slack.com/archives/C2ND76PAA/p1509631300000212): > If the time at when a peer was last seen was tracked as part of the peers list, then dead peers could be identified with...

We did this for the Kubernetes integration; it was rather a lot of work to coordinate across nodes. See https://github.com/weaveworks/weave/issues/2797 and subsequent fixes.

@2opremio notes there is a way to do that through a packer parameter https://github.com/hashicorp/packer/issues/1762#issuecomment-68086756

@leth would you care to explain your regex? ISTM that a regex approach should work, but I had `git status --porcelain` like this: ``` M run-integration-tests.sh M src/kubernetes.js M src/server.js...

what is happening with this?

I’ve not looked at the Pusher package, but “foo-job” is invalid per the [spec](https://prometheus.io/docs/concepts/data_model/#metric-names-and-labels) - minus sign is not allowed.