-
Notifications
You must be signed in to change notification settings - Fork 303
Description
Description:
The namedprocess_namegroup_states metric in process-exporter incorrectly displays one extra thread when the number of threads exceeds two digits. The total number of thread states should match namedprocess_namegroup_num_threads, but an extra thread is displayed when the thread count is 10 or more.
Expected Behavior:
The namedprocess_namegroup_states(all states, Running,Sleeping,Zombie,Other and Waiting) metric should match the total number of threads (namedprocess_namegroup_num_threads) precisely.
Actual Behavior:
namedprocess_namegroup_states displays one additional thread.
This issue occurs only in the Prometheus UI, while system metrics correctly report the number of threads and thread states.
Configuration:
Helm chart: https://mumoshu.github.io/prometheus-process-exporter
Version: 0.5.0
Image: ncabatoff/process-exporter
Tag: sha-626431b
Steps to Reproduce:
- Install process-exporter using the above Helm configuration.
- Enable {{.Cgroups}} in groups to get the container ID
groups:- name: "{{.Cgroups}}"
cmdline:- '.+'
3.Run process-exporter to collect metrics for a process with a thread count greater than 10.
4.Compare the values of namedprocess_namegroup_states and namedprocess_namegroup_num_threads.
- '.+'
- name: "{{.Cgroups}}"
- Observe that namedprocess_namegroup_states reports one additional thread.
6.Verify the actual thread count on the system, which correctly matches namedprocess_namegroup_num_threads.
Screen shots:




group Configuration:
Looking forward to your insights and a possible fix. Thanks in advance for your help!
Please let me know if any additional details are required.