Update prometheus helmfiles and rules to better use metric labels #304
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
using .75 for now as anecdotally seems to work. Ideal number could be more or less.
MekWarrior
approved these changes
Mar 28, 2020
willgraf
added a commit
that referenced
this pull request
May 23, 2020
* Update prometheus-operator, prometheus-adapter, and promethes-redis-exporter helm charts and remove stale default values * Relabel redis-exporter with `deployment=$QUEUE-consumer` and change key to be `$QUEUE` * Rename zip-consumer to segmentation-zip-consumer to match labels. * Using .75 instead of .9 for backoff coefficient.
willgraf
added a commit
that referenced
this pull request
May 23, 2020
* Update prometheus-operator, prometheus-adapter, and promethes-redis-exporter helm charts and remove stale default values * Relabel redis-exporter with `deployment=$QUEUE-consumer` and change key to be `$QUEUE` * Rename zip-consumer to segmentation-zip-consumer to match labels. * Using .75 instead of .9 for backoff coefficient.
willgraf
added a commit
that referenced
this pull request
May 23, 2020
* Update prometheus-operator, prometheus-adapter, and promethes-redis-exporter helm charts and remove stale default values * Relabel redis-exporter with `deployment=$QUEUE-consumer` and change key to be `$QUEUE` * Rename zip-consumer to segmentation-zip-consumer to match labels. * Using .75 instead of .9 for backoff coefficient.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
In an effort to improve the autoscaling, each prometheus-related chart has been updated to the latest version. In updating the charts, I found that some of the values became stale (especially the image version), so I removed all values that we did not need/want to set ourselves. They just inherit from the default value set, with a link provided in the helmfile.
Additionally, prometheus metrics use labels to do math on 2 or more time series. To get our labels to match up, I made a few changes:
redis-exporterto have the key name bequeueandqueue-zipinstead ofqueue_image_keysandqueue_zip_keys.metric_relabel_configto theredis-exporterprometheus job to take thequeuename and include a new label,deployment="queue-consumer".zip-consumertosegmentation-zip-consumerin order to have the labels match the queue name.Also, I updated our rules:
consumer_key_ratioandconsumers_per_gpuwhich use the new labels to calculate stats for all deployed consumers (if their nameX-consumermatches the queueX).tf_serving_gpu_usagemetric.avg_over_timecalls, they are outputting discrete points instead of a nice line.