/juju

Increase timeout for api connection in worker/pubsub. #8055

Merged
merged 1 commit into from Nov 13, 2017

Conversation

Reviewers

@wallyworld wallyworld

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
3 participants
@howbazaar @jujubot @wallyworld
@howbazaar
Owner

howbazaar commented Nov 13, 2017

From the time the pubsub workers until now, something has caused api.Open to take longer than 50ms. The 50ms number was a somewhat arbitrarily small number. This branch takes it to two seconds. This is the maximum amount of time we expect an api.Open call to take from one controller machine to another in the HA cluster. It should be way shorter than that, but it taking over 50ms just now.

This is causing a pinger to be leaked, which creates another pinger.being. This underlying bug still needs to be fixed, but this change stops irritating that other bug.

QA steps

juju bootstrap lxd test-pinger
juju enable-ha
# wait for the machines to start
juju ssh -m controller 0
juju-pubsub-report

The other controller machines should show as connected.

Bug reference

Partial fix for https://bugs.launchpad.net/juju/+bug/1731745
@wallyworld

wallyworld approved these changes Nov 13, 2017
View changes

I kinda wish all of our dial opts were common instead of snowflakes in each place that uses yhem.

@howbazaar
Owner

howbazaar commented Nov 13, 2017

$$merge$$
@jujubot
Contributor

jujubot commented Nov 13, 2017

Status: merge request accepted. Url: http://ci.jujucharms.com/job/github-merge-juju

@jujubot jujubot merged commit 093a970 into juju:develop Nov 13, 2017
1 check failed

1 check failed

continuous-integration/jenkins/pr-merge This commit cannot be built
Details

@howbazaar howbazaar deleted the howbazaar:pubsub-connection-timeout branch Nov 13, 2017

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment