[Observability] Use metric filter to generate the clustermgtd heartbeat metric.#7219
Merged
gmarciani merged 3 commits intoaws:developfrom Feb 5, 2026
Conversation
gmarciani
changed the title
gmarciani
force-pushed
the
wip/mgiacomo/3150/clustermgtd-alarm-fix-0204-1
branch
2 times, most recently
from
14c5fba to
cd40ce8
Compare
…ndpoint for CloudWatch Metrics (com.amazonaws.$region.monitoring), which is now required by clustermgtd to put metrics." This reverts commit 3aacea3. We must revert this commit because the VPC Endpoint for CloudWatch Monitoring is not needed anymore. We introduced it when we introduced for the first time a call to PutMetricData from clustermgtd. Now we moved to an approach where we make no rewquests to CloudWatch Monitoring so the endpoint is not required anymore.
…the head node policy so that clustermgtd is able to emit metrics." This reverts commit 9329636. We must revert this commit because we do not need the permission cloudwatch:PutMetricData` anymore. We introduced such extra permissions when we introduced a PutMetricData request from the head node to let clustermgtd publish a metric. However, we now changed the approach moving to metric filters rather than explicit PutMetricData, so the permissionb is not required anymore.
gmarciani
force-pushed
the
wip/mgiacomo/3150/clustermgtd-alarm-fix-0204-1
branch
from
cd40ce8 to
ef2eab7
Compare
gmarciani
force-pushed
the
wip/mgiacomo/3150/clustermgtd-alarm-fix-0204-1
branch
from
ef2eab7 to
ba3c5e2
Compare
hanwen-cluster
approved these changes
Feb 5, 2026
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Description of changes
Use metric filter to generate the clustermgtd heartbeat metric rather than metric pushed by clustermgtd.
In #7209 we introduced a new alarm for clustermgtd not running which was based on a metric pushed by clustermgtd. We decided to change the approach and replace the explicit metric publishing form clustermgtd with a metric filter.
The advantage of this new approach compared to the previous one is that:
cloudwatch:PutMetricDataUX
Here is the alarm and widget visible on the dashboard. We did not change the user experience of the dashboard compared to what we had before. This is just to show that it is functional.
Notes
test_cluster_in_no_internet_subnet. We do not need such VPC Endpoint anymore with the new approach.cloudwatch:PutMetricDatathat is not required anymore with the new approach.Q&A
Yes, it would be enough to have clustername dimension in that metric. However, our default behavior so far is to include instance id in whatever metric filter. Being stick to such default has an advantage in terms of coding simplicity and also in terms of user experience. When the user browses the metrics within ParallelCluster namespace they will only see one metric folder "ClusterName, InstanceId". If we change from the default, then they would see two folders: one "ClusterName, InstanceId" and another one "ClusterName" which could be confusing. Also this is more future proof in case we want to add head node high availability in future where the head node could run on multiple instances.
Tests
test_monitoringon a cluster that is also using the change [Observability] Clustermgtd to emit heartbeat into event logs rather than explicit CW metric aws-parallelcluster-node#687By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.