Closed InformatiQ closed 9 years ago
Hi, it might worth check whether marathon http callback is enabled?
On 26 Jun 2015, at 15:39, Ramez Hanna notifications@github.com wrote:
according to https://github.com/QubitProducts/bamboo/blob/master/main/bamboo/bamboo.go#L102 the url for the marathon callback would be http://10.5.0.50:8000/api/marathon/event_callback which is what i set in marathon bamboo gets notified of changes and i can see it reflected in the API, but the haproxy doesn't cahnge and no logs to help
— Reply to this email directly or view it on GitHub.
It is enabled and bamboo api reflects the changes done in marathon but does not update haproxy
On Fri, 26 Jun 2015 17:27 Jing Dong notifications@github.com wrote:
Hi, it might worth check whether marathon http callback is enabled?
On 26 Jun 2015, at 15:39, Ramez Hanna notifications@github.com wrote:
according to https://github.com/QubitProducts/bamboo/blob/master/main/bamboo/bamboo.go#L102 the url for the marathon callback would be http://10.5.0.50:8000/api/marathon/event_callback which is what i set in marathon bamboo gets notified of changes and i can see it reflected in the API, but the haproxy doesn't cahnge and no logs to help
— Reply to this email directly or view it on GitHub.
— Reply to this email directly or view it on GitHub https://github.com/QubitProducts/bamboo/issues/142#issuecomment-115730049 .
You probably need to check your reload command is doing what you expect then.
@activars we are seeing this as well. We also see this when an application crashes or is redeployed, bamboo wont pick up the changes.
It's likely that marathon "restart" doesn't trigger callback in recent version.
"Redeploy" should always trigger changes.
Bamboo's log and built-in statsd are the best place to troubleshoot issues.
One good tips is always checking haproxy stats page. It shows whether the application is added to the load balancing pool and shows state of the application.
Applications are usually developed in different style or using many web frameworks. Sometime they don't provide correct http implementation, haproxy may not like those health check responses. (I've seen issues with some popular Python and Scala web frameworks)
On 26 Jun 2015, at 20:32, Austin Riendeau notifications@github.com wrote:
@activars we are seeing this as well. We also see this when an application crashes or is redeployed, bamboo wont pick up the changes.
— Reply to this email directly or view it on GitHub.
So the only problem I see that it is just not pulling the information. I am manually clicking the pencil, and update. This occurs with both with redeploy or restart.
Does the host running Bamboo be able to curl Marathon host? Marathon and bamboo should be able to communicate with each other.
On 29 Jun 2015, at 00:10, Austin Riendeau notifications@github.com wrote:
So the only problem I see that it is just not pulling the information. I am manually clicking the pencil, and update. This occurs with both with redeploy or restart.
— Reply to this email directly or view it on GitHub.
yes they can communicate @activars as i mentioned, when i make a change in marathon for example, a scale up increasing from 3 to 4 instances (no docker) i can see the change in reflected through the bamboo api, but the haproxy config is not updated and i see no logs related I am using latest release of bamboo not master
it was able to fix this by deleting all callback urls from marathon and restarting bamboo
@InformatiQ I dont know if this is a good solution because that doesnt solve the problem in the long run.
according to https://github.com/QubitProducts/bamboo/blob/master/main/bamboo/bamboo.go#L102 the url for the marathon callback would be http://10.5.0.50:8000/api/marathon/event_callback which is what i set in marathon bamboo gets notified of changes and i can see it reflected in the API, but the haproxy doesn't cahnge and no logs to help