Hey Mohammad, Thanks for your interest and gumption in moving Knox forward. We always encourage collaboration and contributions of all kinds. I would be surprised if Jeffrey objected to any work or help with a patch. I¹m sure he would be in fact happy to review and/or work with you on a patch. I would recommend tagging Jeffrey in a Jira comment and going forward with working on a contribution. I say this also because the Jira has been created a while ago and could certainly use some progress.
Thanks again! Sumit On 11/2/16, 8:27 PM, "Mohammad Islam" <[email protected]> wrote: >Thanks Sumit for the pointer. >Is there any plan to work on it soon? The reason I'm asking is : if I get >some guidance, I can also stab on it if Jeffrey (current Assignee) is >busy. >Regards,Mohammad > > > On Wednesday, November 2, 2016 4:43 PM, Sumit Gupta ><[email protected]> wrote: > > > Hi Mohammad, > >You are right we have HA support for various services that is somewhat >generic and based on connectivity failures. Our WebHDFS HA support is a >bit more specific and maps to some of the nuanced states that the >Namenodes can be in. We have yet to put in specific support for Yarn and >like you point out there is a redirect that occurs that needs to be >handled. > >There is an open JIRA to add support for Yarn¹s resource manager in HA >mode. > >https://issues.apache.org/jira/browse/KNOX-719 > >Sumit > >From: Mohammad Islam <[email protected]<mailto:[email protected]>> >Reply-To: "[email protected]<mailto:[email protected]>" ><[email protected]<mailto:[email protected]>>, Mohammad Islam ><[email protected]<mailto:[email protected]>> >Date: Wednesday, November 2, 2016 at 6:57 PM >To: "[email protected]<mailto:[email protected]>" ><[email protected]<mailto:[email protected]>>, >"[email protected]<mailto:[email protected]>" ><[email protected]<mailto:[email protected]>> >Subject: Re: HA support for YARN and NameNode UI level > >Just to add more specific use-case of YARN UI. > >BTW my KNOX service was running behind another FE proxy. > >When I typed https://<MY_FE_PROXY>/gateway/sandbox/yarn in my browser, I >saw the message " This is standby RM. Redirecting to the current active >RM: >http://<PRIMARY_DIRECT_YARN_UI_HOST>:8088/cluster<http://<PRIMARY_UI_HOST> >:8088/cluster>". > >At this time, my browser also tried to connect to >"http://<DIRECT_YARN_UI_HOST_PRIMARY>:8088/cluster<http://<PRIMARY_UI_HOST >>:8088/cluster>" but it's timed out because the YARN internal host was >>behind the firewall and was not directly accessible through browser. > > >Looks like Knox was sending a response to FE proxy with a header >"Refresh: 3; url=http://<DIRECT_YARN_UI_HOST_PRIMARY>:8088/cluster". My >question is : is there a way for Knox to replace the internal Yarn Host >or endpoint with something like knoxy endpoint. > >Regards, >Mohammad > > > > >On Tuesday, November 1, 2016 12:24 PM, Mohammad Islam ><[email protected]<mailto:[email protected]>> wrote: > > >Hi, >I'm wandering if there is any built-in support in Knox for any (NN/RM) UI >to automatically forward to active NN/RM. > >I saw there is HA support in WebHDFS level. But not sure if I need to do >anything special for HDFSUI. >The same is true for YARNUI. > >Regards, >Mohammad > > > > > >
