[jbosstools-issues] [JBoss JIRA] (JBIDE-22603) Sometimes multiple OpenShift watch managers are periodically created and finished

[Jeff MAURY (JIRA)]

    [ https://issues.jboss.org/browse/JBIDE-22603?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13259064#comment-13259064 ] 

Jeff MAURY edited comment on JBIDE-22603 at 6/30/16 4:46 AM:
-------------------------------------------------------------

OK seems I am able to reproduce the problem with a single watched connection. Seems the web socket is closed or timeout expired after a certain period.
Anyway, reading the code, I found several pbs:
* watchers are registered using Openshift project as key so if two different connections have the same project name, ....
* there is one job per project and resource type where the REST client allows to have a single watcher for multiple resource type
* in some rare case, the initial watcher cannot be started for a resource type but this is hard to reproduce and may disappear when a single watcher per project will be set

Given the amount of work, I propose to delay this issue to the next sprint


was (Author: jeffmaury):
OK seems I am able to reproduce the problem with a single watched connection. Seems the web socket is closed or timeout expired after a certain period.
Anyway, reading the code, I found several pbs:
* watchers are registered using Openshift project as key so if two different connections have the same project name, ....
* there is one job per project and resource type where the REST client allows to have a single watcher for multiple resource type
* in some rare case, the initial watcher cannot be started for a resource type but this is hard to reproduce and may disappear when a single watcher per project will be set

Given the amount of work, I propose to delay this issue to the next one

> Sometimes multiple OpenShift watch managers are periodically created and finished
> ---------------------------------------------------------------------------------
>
>                 Key: JBIDE-22603
>                 URL: https://issues.jboss.org/browse/JBIDE-22603
>             Project: Tools (JBoss Tools)
>          Issue Type: Bug
>          Components: openshift
>    Affects Versions: 4.4.0.Final
>            Reporter: Marián Labuda
>            Assignee: Jeff MAURY
>            Priority: Critical
>              Labels: explorer, openshift_v3
>             Fix For: 4.4.x
>
>         Attachments: watchers.png
>
>
> Sometimes when I am working with OpenShift toolings, there are created multiple OpenShift Watch Managers which are getting finished, disappearing from Progress view and another new ones are created and this repeats nonstop. Sometimes there are 2-3 running, sometimes even more. I am working with openshift-dev user, where is visible default project (still, until patch in upstream get effect) and my own project with application created from eap 6.4 basic s2i template.
> !watchers.png!
> It would not be a problem but I have a hunch it is the problem breaking automatic update of OpenShift explorer view to reflect current state and existence of resources on OpenShift server. E.g. under a service there is build table visible all the time and no application pod is shown, even build is finished and there is an application pod running on OpenShift. Another example when it is not working is when I am scaling application up/down - it does not reflect real amount of application pods.



--
This message was sent by Atlassian JIRA
(v6.4.11#64026)