You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@mesos.apache.org by "Dennis (JIRA)" <ji...@apache.org> on 2018/01/22 13:54:01 UTC

[jira] [Commented] (MESOS-7503) Consider improving the WebUI failed to connect dialog.

    [ https://issues.apache.org/jira/browse/MESOS-7503?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16334283#comment-16334283 ] 

Dennis commented on MESOS-7503:
-------------------------------

Same here, Mesos is running in Microsoft Azure behind a load balancer and /master/state and metrics/snapshot are using the internal IP of the current master which leads to the "failed to connect" errors...

 

> Consider improving the WebUI failed to connect dialog.
> ------------------------------------------------------
>
>                 Key: MESOS-7503
>                 URL: https://issues.apache.org/jira/browse/MESOS-7503
>             Project: Mesos
>          Issue Type: Improvement
>          Components: webui
>    Affects Versions: 1.2.0
>            Reporter: Anand Mazumdar
>            Priority: Major
>              Labels: mesosphere, webui
>         Attachments: Capture d’écran 2017-05-12 à 15.06.07.png
>
>
> Usually, when your Mesos Master is behind a reverse proxy/LB; the keepalive timeout value would be set to a small value e.g., 60 seconds for nginx. This results in the persistent connection between the browser and the Mesos master breaking resulting in the connection lost dialog (see attached screenshot). This is very inconvenient when debugging using the Web UI.
> We should consider making the error dialog less intrusive e.g., update an element to signify that a reconnection is in progress similar to what other online services like gmail etc. do.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)