prometheus metric_version = 2 & make scrape url as a configurable tag #5767
Conversation
|
Fixed the below issue as well as part of this same outstanding PR. (updated PR title to reflect the same) |
There was a problem hiding this comment.
I think I need to work on #3670 before merging, and specifically a way to have per field metric types. This will allow us to parse:
# HELP haproxy_up Was the last scrape of haproxy successful.
# TYPE haproxy_up gauge
haproxy_up 1
# HELP haproxy_left Was the last scrape of haproxy successful.
# TYPE haproxy_left counter
haproxy_left 1
# HELP haproxy_down Was the last scrape of haproxy successful.
# TYPE haproxy_down untyped
haproxy_down 1
And have them show up on a single line:
prometheus haproxy_up=1,haproxy_left=1,haproxy_down=1
But still convert them back to typed metrics in the prometheus output.
@danielnelson - when is 1.12.0 expected to ship? |
It will be around the beginning of September, sorry. |
|
@danielnelson We need to think of a long term plan for these types of situations. Setting a metric version to 1,2,3 is not sustainable for large deployments and makes upgrading difficult |
|
@Esity I agree it is quite hard to make an upgrade even with these options. Do you have an idea of what could make schema changes go smoother to help get us thinking? |
There was a problem hiding this comment.
I'm going to push #3670 off for now so we can for sure get this in for 1.13.0, there are just a few minor changes I'd like made first:
--test output -- go test -short ./plugins/inputs/prometheus/... ok github.com/influxdata/telegraf/plugins/inputs/prometheus 0.034s
|
@danielnelson - i fixed all the changes requested a few weeks back. Do you know when this is expected to be merged & shipped ? Thank you! |
|
@vishiy Sorry, I've just been out of town, will take a second look once I get situated. Expecting to merge for 1.13. |
Thank you! |
Required for all PRs:
#4415
#5689
#5798