-
-
Notifications
You must be signed in to change notification settings - Fork 107
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Setting redis as coordination backend #169
Conversation
Thanks a lot @arms11! 👍 Will try to run this setup and play with the config options. |
@test 'st2 key/value operations are functional' { | ||
run st2 key set foo bar | ||
assert_success | ||
|
||
run st2 key get foo | ||
assert_success | ||
assert_line --partial 'bar' | ||
|
||
run st2 key delete foo | ||
assert_line --partial '"foo" has been successfully deleted' | ||
assert_success | ||
|
||
run st2 key get foo | ||
assert_line --partial '"foo" is not found' | ||
assert_failure | ||
} |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Based on #167, included some tests here to verify the st2 K/V functionality.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Hey @arms11!
I just tested this config and it works OK on a local minikube 👍
Same for CircleCI e2e tests in Github, check the status:
✔️ https://app.circleci.com/pipelines/github/StackStorm/stackstorm-ha/906/workflows/c98fb252-0ac4-47f8-85b3-71a8f29ddd7f
Having said that, everything works well, it's great start!
Let's continue on connection string templating and removing the etcd dependency so we can merge this fix.
Thanks @armab . That's great and concerning at the same time. Not sure how this is working for you and not for me! I do have |
@armab - I have pushed the changes for Redis.
|
Besides of the comment above, we'll also need a changelog record for this PR and I think it will be good to merge! 👍 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Great work here 👍
Thanks a lot for the contribution!
@arms11 Following this major change, could you also update the official st2 documentation page for the The documentation page source code is available at https://github.com/StackStorm/st2docs/blob/master/docs/source/install/k8s_ha.rst |
This could not have come at a better time. A PKS/Kube host upgrade this weekend killed our etcd-cluster. After fighting half the day to get our obsolete stuff working again, I saw this merge in. Ran it a bit ago and it seems to be working perfect. One thing for documentation/awareness. We upgraded from chart 0.32.0 to 0.51.0. So we still had etcd-operator. I had commented out the etcd-operator section for the first run. Everything came up, but just as the final st2-mongodb-2 was starting up, etcd-operator appeared and st2-redis instances vanished. Once etcd-operator was in play, the mongo and rabbitmq pods started restarting and everything went downhill. For my second run, I explicitly set etcd-operator.enabled: false. The etcd-cluster & etc-operator pods went away, the new redis pods got created, and everything came up. It's possible that this was some sort of hidden failure and helm automatically rolling back to the prior chart, but it wasn't clear. Since I have leftover bits related to etcd, and anyone else upgrading will, I recommend that they explicitly disable etcd-operator for a release or two. Maybe disable it before doing the upgrade to 0.51.0.
|
@ytjohn - Point noted. Actually, this makes sense as you upgraded. Actually, this change should have been in the same PR where we moved from |
@arms11 I just got done reviewing. It turns out that what happened was a helm timeout. Our organization has been running into docker hub rate limits and the deploy took over 10 minutes. After the timeout, helm attempts to roll back. That is what I was seeing. It had nothing to do with the etcd-operator key like I originally thought. In fact, I don't see anything in the helm chart that would act upon such a value anyways. Everything else worked great. |
Thanks @ytjohn. Much appreciated the insight! That helps. |
Closes #94
Fixes #167
For the time-being...
st2ha
as the release name.sentinel=mymaster
in the connection string has been set assuming the name of the master is the name of the masterset.Below is the error I have been receiving...