I'm using kube-prometheus-stack
via Helm release on my AWS EKS cluster and I have started seeing the following error via alert manager:
In Prometheus logs, I see the following error: fallback_media_type="" err="non-compliant scrape target sending blank Content-Type and no fallback_scrape_protocol specified for target"
Helm release version: 67.4.0
Prometheus Image Version: v3.0.1
I have checked both the components on GitLab and they are working fine. It has something to do with Prometheus scraping.
How can I fix this?
I have found the solution. I upgraded the kube-prometheus-stack
Helm release to 68.1.0
and Prometheus image v3.1.0
and after that, I updated the Prometheus configuration and added fallback_scrape_protocol: PrometheusText0.0.4
in the following 2 jobs as follows:
...
- job_name: gitlab_exporter_database
fallback_scrape_protocol: PrometheusText0.0.4
metrics_path: /database
static_configs:
- targets:
- __REDACTED__
- job_name: gitlab_exporter_sidekiq
fallback_scrape_protocol: PrometheusText0.0.4
metrics_path: /sidekiq
static_configs:
- targets:
- __REDACTED__
- job_name: gitaly
...
After that, it worked.
Reference for solution is here.