You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@heron.apache.org by GitBox <gi...@apache.org> on 2020/08/20 04:16:42 UTC

[GitHub] [incubator-heron] nicknezis commented on issue #3554: Kubernetes Scheduler does not scale properly

nicknezis commented on issue #3554:
URL: https://github.com/apache/incubator-heron/issues/3554#issuecomment-677010647


   I was able to set the `--container-number` after making my code edits. It also fixes the json issue when setting `--component-parrallelism`. 
   
   I still ran into a different issue when testing with the `acking` example topology.
   `heron update heronkube acking --component-parallism exclaim1:3`
   
   Here is the error message I see in the `acking-0` TMaster log.
   ```
   I0820 03:47:17.628922   125 tmaster.cpp:323] Topology read and validated
   I0820 03:47:17.628945   125 zkclient.cpp:228] Getting zknode /heron/pplans/acking
   I0820 03:47:17.629315   125 heron-zkstatemgr.cpp:788] Setting watch on packing plan succeeded: ok
   E0820 03:47:17.630455   125 tmaster.cpp:497] There was no existing physical plan
   I0820 03:47:17.630913   125 httpserver.cpp:49] Starting Http Server bound to 0.0.0.0:6002
   I0820 03:47:17.630980   125 httpserver.cpp:49] Starting Http Server bound to 0.0.0.0:6003
   I0820 03:47:21.054697   125 tmaster.cpp:697] Got a register stream manager request from stmgr-1
   I0820 03:47:26.723330   125 tmaster.cpp:697] Got a register stream manager request from stmgr-2
   I0820 03:47:26.723366   125 tmaster.cpp:752] All stream managers have connected with us
   I0820 03:47:26.723526   125 zkclient.cpp:192] Creating zknode /heron/pplans/acking
   E0820 03:47:26.726553   125 heron-zkstatemgr.cpp:510] Creating Physical Plan failed because zk was not setup properly
   E0820 03:47:26.726657   125 tmaster.cpp:807] Error writing assignment to statemgr. Error code is 2000
   ```
   
   I suspect that the original issue is fixed with this PR, but perhaps there is a different issue. It seemed that the TMaster was in a constant crash loop and the `heron update` command never returned. I was forced to `heron kill heronkube acking` to resolve the failed state.


----------------------------------------------------------------
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:
users@infra.apache.org