Open kaushique opened 2 years ago
👋 can you share the logs of the pods when this situation occurs? Preferably with log.level set to debug or trace?
We experienced the same and after deploying the Exporter like a million times through Helm with different options this worked for us (for the records, we've deployed GitLab EE through Omnibus on EL7):
[...]
config:
gitlab:
enable_tls_verify: true
enable_health_check: true
maximum_requests_per_second: 60
token: <token>
url: https://git.host.name/
[...]
We had config.gitlab.url in our values.yaml without the trailing Slash, but it seems to be required.
So mind the trailing Slash in the GitLab URL, that fixed it for us.
Interesting, I believe that by adding a trailing /
on the GitLab URL & not setting health URL the you are hitting a redirect which likely get's you to the login page with a 200.
Have you checked the output of https://git.host.name/-/health ?
This is what various requests gave me
# curl -sS -i https://git.host.name/-/health
HTTP/2 404
server: nginx
[...snipp...]
Not Found
# curl -sS -i https://git.host.name/-/health?token={{config.gitlab.token}}
HTTP/2 404
server: nginx
[...snipp...]
Not Found
# curl -sS -i https://git.host.name/-/health?token={{access token from /admin/health_check}}
HTTP/2 404
server: nginx
[...snipp...]
Not Found
# curl -sS -I https://git.host.name/-/readiness?token={{config.gitlab.roken}}
HTTP/2 404
server: nginx
[...snipp...]
{"error":"Not Found","status":404}
# curl -sS -i https://git.host.name/-/readiness?token={{access token from /admin/health_check}}
HTTP/2 200
server: nginx
[...snipp...]
{"status":"ok","master_check":[{"status":"ok"}]}
Can confirm, the trailing slash is the magic!
I'm experiencing the same issue when pulling metrics from gitlab.com. I've tested with multiple different configs, anyone have any suggestions or seen this too?
Thanks, this tip on the trailing / helped me as well!
Hi Team, i've setup gitlab ci pipeline exporter using helm chart way , but when i see the pod events it says "Readiness probe failed: HTTP probe failed with statuscode: 503" Because of this the endpoint is not coming and i'm unable to view metrics at /metrics at port 8080 I've made sure no other app service is exposed at 8080 . it's reading the data from gitlab but status of the pod remains 0/1 running how do i overcome it ? i tried increasing the initial delay and failure threshold for readiness probe but no luck
Any help is appreciated and Thanks in advance