Closed kinvaris closed 7 years ago
@redlicha what would your advice be?
BAM needs to reflect on wether we should do this.
@wimpers, isn't this just something of the healthcheck itself? As like how they should log their error?
@wimpers There is no need to include Arne. This is a decision of the healthcheck to display timeouts as warning instead of error (as it could mean that the system is healthy but slow!)
@khenderick so add status TIMEOUT as optional output status (instead of warning/critical)?
@wimpers, I have decided that @JeffreyDevloo and @kinvaris shall change logger.failure
to logger.warning
.
This is a decision of the healthcheck to display timeouts as warning instead of error
Actually and frankly it is not up to the team writing the healthcheck to make that decision. The healthcheck team should not on its own decide when OPS (and engineering if things really goes south) gets called out of bed in the middle of the night.
We always poll with OPS for the reason that you stated. Most of our changes are also requested by ops.
Maybe a suggestion: when creating a volume results in a time out, can we try again? Based on the 2nd attempt, I'd go for an error, as it will most likely point to something underneath which is too slow/not working/...
After discussion with @jtorreke :
Since check_Mk knows ok
, warning
, error
, unknown
let's use unknown
in case there is a timeout.
Volume creation is currently disabled.
Validation in: https://github.com/openvstorage/integrationtests/issues/447
Fixed by https://github.com/openvstorage/openvstorage-health-check/pull/291 -> openvstorage-health-check 3.2.1-rev.490.f2bbe59
Problem description
A timeout of a volume creation needs to be in warning not in error https://github.com/openvstorage/openvstorage-health-check/blob/master/ovs/extensions/healthcheck/volumedriver/volumedriver_health_check.py#L150
Additional information
Setup
Packages
// List packages in the code tags using dpkg -l | grep openvstorage