Acess to web too slow

[fwynyk]

I ran the airflow.all.yaml in my kubernetes cluster but when I try to access the web interface, it takes too long. Some minutes after, I can navigate, but it also takes too long.

The cluster is full of resources (CPU/memory). No errors were found in any pod, just these warnings in 'web':

[2017-10-30 13:51:17,936] [53] {models.py:167} INFO - Filling up the DagBag from /usr/local/airflow/dags
[2017-10-30 13:51:18 +0000] [32] [INFO] Handling signal: ttou
[2017-10-30 13:51:47 +0000] [49] [INFO] Worker exiting (pid: 49)
[2017-10-30 13:52:17 +0000] [32] [INFO] Handling signal: ttin
[2017-10-30 13:52:17 +0000] [54] [INFO] Booting worker with pid: 54
/usr/local/lib/python2.7/dist-packages/flask/exthook.py:71: ExtDeprecationWarning: Importing flask.ext.cache is deprecated, use flask_cache instead.
  .format(x=modname), ExtDeprecationWarning
[2017-10-30 13:52:17,995] [54] {models.py:167} INFO - Filling up the DagBag from /usr/local/airflow/dags
[2017-10-30 13:52:18 +0000] [32] [INFO] Handling signal: ttou
[2017-10-30 13:52:18 +0000] [50] [INFO] Worker exiting (pid: 50)
[2017-10-30 13:52:49 +0000] [32] [INFO] Handling signal: ttin
[2017-10-30 13:52:49 +0000] [55] [INFO] Booting worker with pid: 55
/usr/local/lib/python2.7/dist-packages/flask/exthook.py:71: ExtDeprecationWarning: Importing flask.ext.cache is deprecated, use flask_cache instead.
  .format(x=modname), ExtDeprecationWarning

Do you have any idea? Does anyone tried to run it on a real Kubernetes cluster?

[chribsen]

I'm using Airflow on k8s in preproduction - overall it works fine, but the web pod seems to halt after some time. However, I managed to fix this with a health check (the health check regularly pings the web page and restarts the pod if it doesn't respond).
Does this problem occur immediately after deploying kube-airflow?