[GitHub] [incubator-druid] vogievetsky commented on issue #8686: Router management Proxy routing should have a way to route to any node in the cluster
vogievetsky commented on issue #8686: Router management Proxy routing should have a way to route to any node in the cluster URL: https://github.com/apache/incubator-druid/issues/8686#issuecomment-542968137 Yeah just to be clear I never pictured it to be a proxy to an arbitrary destination only to other Druid servers. If there was some server uuid we could use that instead of a host port combo but right now from the sys table the best server id I can see is host:port  The issue with having this check be encapsulated by a single backend API is that it might give someone the idea to hit it as part of a routine health monitoring which limits how heavy these tests might should be. I am very much picturing a 'deep' test suite (that can take its sweet time) and might be intensive on the cluster. Something you run a handful of times in the clusters lifetime (in leu of posting to the user-group). Something to catch issues like: https://groups.google.com/forum/#!topic/druid-user/w3Uy-dXZSII This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services - To unsubscribe, e-mail: commits-unsubscr...@druid.apache.org For additional commands, e-mail: commits-h...@druid.apache.org
[GitHub] [incubator-druid] vogievetsky commented on issue #8686: Router management Proxy routing should have a way to route to any node in the cluster
vogievetsky commented on issue #8686: Router management Proxy routing should have a way to route to any node in the cluster URL: https://github.com/apache/incubator-druid/issues/8686#issuecomment-542959319 The proxy could only forward to Druid nodes themselves so it is not open to everything. Also there is already an open proxy in the sampler (http firehose). I tried to implement it that way and it works for non TLS clusters, (for TLS clusters we would need to address: https://github.com/apache/incubator-druid/issues/8668 From a security perspective my model is that if you have admin access to the management proxy enabled router you can pretty much do anything anyway at that point. From a scalability perspective I do not think it is a problem if this is a user initiated action (with a warning in the UI). Also remember that it would be OK for this to be slow, taking a 100ms pause between every request. People would be fine with it as long as there is visual feedback with a ticking `Contacting node 12 out of 343` that shows that progress is being made. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services - To unsubscribe, e-mail: commits-unsubscr...@druid.apache.org For additional commands, e-mail: commits-h...@druid.apache.org
