This repository has been archived by the owner on Sep 30, 2024. It is now read-only.
GraphQL requests are ambiguous in Prometheus, making it impossible to monitor HTTP metrics #4246
Assignees
Labels
Milestone
Comments
slimsag
added
ops & tools & dev
customer
slimsag
added a commit
that referenced
this issue
Jun 6, 2019
Today our HTTP metrics are pretty nice, but when it comes to any GraphQL requests they are all grouped under the `graphql` route which makes them mostly useless. For example, you can see HTTP latency is quite bad or that there are lots of error codes on GraphQL requests -- but it is impossible to know if those are search requests or some super-slow request made on e.g. the admin-only repository list page. I chose to include this in the route name (`graphql: Search`) instead of adding a different field because it is nice to not have to deal with the separate field in dashboard queries and because the route field would be mostly useless on GraphQL requests otherwise. Fixes #4246
slimsag
added a commit
that referenced
this issue
Jun 6, 2019
Today our HTTP metrics are pretty nice, but when it comes to any GraphQL requests they are all grouped under the `graphql` route which makes them mostly useless. For example, you can see HTTP latency is quite bad or that there are lots of error codes on GraphQL requests -- but it is impossible to know if those are search requests or some super-slow request made on e.g. the admin-only repository list page. I chose to include this in the route name (`graphql: Search`) instead of adding a different field because it is nice to not have to deal with the separate field in dashboard queries and because the route field would be mostly useless on GraphQL requests otherwise. Fixes #4246
slimsag
added a commit
that referenced
this issue
Jun 6, 2019
* pkg/trace: include GraphQL ?Type parameter in route name Today our HTTP metrics are pretty nice, but when it comes to any GraphQL requests they are all grouped under the `graphql` route which makes them mostly useless. For example, you can see HTTP latency is quite bad or that there are lots of error codes on GraphQL requests -- but it is impossible to know if those are search requests or some super-slow request made on e.g. the admin-only repository list page. I chose to include this in the route name (`graphql: Search`) instead of adding a different field because it is nice to not have to deal with the separate field in dashboard queries and because the route field would be mostly useless on GraphQL requests otherwise. Fixes #4246 * CHANGELOG
slimsag
added a commit
that referenced
this issue
Jun 6, 2019
* pkg/trace: include GraphQL ?Type parameter in route name Today our HTTP metrics are pretty nice, but when it comes to any GraphQL requests they are all grouped under the `graphql` route which makes them mostly useless. For example, you can see HTTP latency is quite bad or that there are lots of error codes on GraphQL requests -- but it is impossible to know if those are search requests or some super-slow request made on e.g. the admin-only repository list page. I chose to include this in the route name (`graphql: Search`) instead of adding a different field because it is nice to not have to deal with the separate field in dashboard queries and because the route field would be mostly useless on GraphQL requests otherwise. Fixes #4246 * CHANGELOG
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
GraphQL requests currently show up in Prometheus as just "GraphQL" instead of e.g. "Search" or "Repositories".
This makes it impossible for us / site admins to know if baseline HTTP perf is bad or if just e.g. a site admin endpoint has poor perf.
Required for upgrading at https://app.hubspot.com/contacts/2762526/company/407948923
The text was updated successfully, but these errors were encountered: