cinder-volume is marked down during image uploads

[Fabian Zimmermann]

cinder-volume-Agent is marked as down if upload of volume needs longer than heartbeat-timeout.

steps to reproduce:

1. create a big Volume (f.e. 100G)
2. fill the volume with data to avoid too fast upload
3. tell cinder to upload the volume as image
4. watch cinder-volume-agent being marked as DOWN, because noch heartbeat-update is done during upload.

the last line in cinder-volume.log is the curl-CMD to upload the image. As soon as the upload completed the agent will reconnect the lost mysql-connection and is back active.

[Fabian Zimmermann (dev-faz)]

Version: PIKE

[Fabian Zimmermann (dev-faz)]

Could validate, the problem is *new* since Pike. Rebuild setup with Ocata and agent remains reporting state during upload.

[Fabian Zimmermann (dev-faz)]

if I install cinder-volume via pip under python3 it working as expected, so working setups

* Ocata with python2.7
* Pike with python3.5

not working

* Pike with python2.7

I would prefer running in python2.7, because in python3.5 the RBDDriver is failing with "JSON/Bytes"-errors and I had to do a really dirty qick-hack to get the volume-agent started.

[Fabian Zimmermann (dev-faz)]

Looks like problem is caused by ubuntu or one of the pkg ubuntu is using in 16.04. Just installed the latest stable/pike-branch directly with pip and even python 2.7 is working as expected.

Feel free to close/invalidate this bug. I will try to create a suitable bug in ubuntu-pkg-bugtracker