Add CPU profile collection to diagnostics handler #4394
Conversation
michel-laterman
added
enhancement
|
Pinging @elastic/elastic-agent-control-plane (Team:Elastic-Agent-Control-Plane) |
|
This pull request does not have a backport label. Could you fix it @michel-laterman? 🙏
NOTE: |
michel-laterman
force-pushed
the
diag-handler-cpu
branch
from
7c232ad to
f251b0d
Compare
|
fleet-server changes are here: elastic/fleet-server#3333 in the future we can also specify a duration parameter as part of the action, but that will need support from the elastic-agent-client-libs repo in order for us to actually pass it (especially to |
| } | ||
| diags = append(diags, client.DiagnosticFileResult{ | ||
| Name: "cpuprofile", | ||
| Filename: "cpu.pprof", |
There was a problem hiding this comment.
I copied metadata from the control server: https://github.com/elastic/elastic-agent/blob/main/pkg/control/v2/server/server.go#L214-L221
We use cpu.pprof, but all other traces use *.pprof.gz.
Should we alter diagnostics.CreateCPUProfile in order to compress the resulting trace?
There was a problem hiding this comment.
It is probably correct to change the name to whatever comes back from a GET to http://localhost:6060/debug/pprof/profile?seconds=30
The parameters here could also be pulled out into a shared set of constants much like the profile duration.
changelog/fragments/1710153529-Add-CPU-profile-collection-to-diagnostics-action-handler.yaml
Outdated
Show resolved
Hide resolved
internal/pkg/agent/application/actions/handlers/handler_action_diagnostics.go
Show resolved
Hide resolved
internal/pkg/agent/application/actions/handlers/handler_action_diagnostics.go
Outdated
Show resolved
Hide resolved
| } | ||
| diags = append(diags, client.DiagnosticFileResult{ | ||
| Name: "cpuprofile", | ||
| Filename: "cpu.pprof", |
There was a problem hiding this comment.
It is probably correct to change the name to whatever comes back from a GET to http://localhost:6060/debug/pprof/profile?seconds=30
The parameters here could also be pulled out into a shared set of constants much like the profile duration.
internal/pkg/agent/application/actions/handlers/handler_action_diagnostics.go
Show resolved
Hide resolved
|
81.1% Coverage on New Code
0.0% Duplication on New Code
|
If there is a Fleet API to let us request this we could write an E2E test for diagnostics requests in this repository. @michel-laterman is there an API for this? Is there a change on the Kibana side to follow? |
What does this PR do?
Add the optional CPU profile collection to the diagnostic action handler.
CPU profiles will be collected if the REQUEST_DIAGNOSTICS action has the
optional additional_metrics parameter list contains "CPU".
Why is it important?
Will make diagnostics requested through fleet feature-complete (compared to cli options)
Checklist
I have made corresponding changes to the documentationI have made corresponding change to the default configuration files./changelog/fragmentsusing the changelog toolI have added an integration test or an E2E testTesting
Changes have been tested with an e2e workflow, see elastic/fleet-server#3333 (comment)
Related issues