-
Notifications
You must be signed in to change notification settings - Fork 360
How to Perf: Finding and Fixing Bottlenecks
Eric Promislow edited this page Jan 3, 2018
·
2 revisions
-
Look at new relic reports for pointers to bottlenecks (login required)
-
Look at an environment, like
cc elsa.capi.land
-
Find hotspot endpoints
-
Turn on mysql logging in the CC and restart
-
Construct scripts that hit the endpoint with
hey
. Ifhey
isn't on your machine install it withgo get -u github.com/rakyll/hey
. -
Observe pre-fix times issued by database, and log average and 90%-ile times by
hey
-
Apply fix and restart CC
-
Rerun script and check the times for differences. If there's an improvement and tests pass, this fix is probably good.
-
Pipelines
-
Contributing
- Tips and Tricks
- Cloud Controller API v3 Style Guide
- Playbooks
- Development configuration
- Testing
-
Architectural Details
-
CC Resources
- Apps
- Audit Events
- Deployments
- Labels
- Services
- Sidecars
-
Dependencies
-
Troubleshooting
- Ruby Console Script to Find Fields that Cannot Be Decrypted
- Logging database queries in unit tests
- Inspecting blobstore cc resources and cc packages(webdav)
- How to Use USR1 Trap for Diagnostics
- How to Perf: Finding and Fixing Bottlenecks
- How to get access to mysql database
- How To Get a Ruby Heap Dumps & GC Stats from CC
- How to curl v4 internal endpoints with mtls
- How to access Bosh Director console and restore an outdated Cloud Config
- Analyzing Cloud Controller's NGINX logs using the toplogs script
-
k8s
-
Archive