You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Subject: Update on Performance Testing with tmpfs Mounts
👋,
I’d like to provide a brief summary regarding the recent performance testing request.
We successfully executed the tests using tmpfs mounts.
Testing branch: test/in-mem-volume
Note: The volume type was modified for all services, not just the database.
Results:
Unfortunately, we did not observe any positive performance improvements. The CI execution time remained unchanged.
Additionally, there was no noticeable performance increase for local development either.
If I overlooked anything or if you'd like to review the results further, please feel free to let me know. I'm happy to provide additional details or assist with further checks! 😊
Additional Observations:
While tmpfs may provide performance benefits in some scenarios, I wasn’t anticipating significant improvements here, as approximately 95% of the execution time is consumed by application-specific processes (e.g., composer install, console command execution, etc.).
Based on the following documentation it's cool to add the possibility to map database files into RAM.
https://docs.gitlab.com/runner/executors/docker.html#mount-a-directory-in-ram
`[runners.docker]
For the main container
[runners.docker.tmpfs]
"/var/lib/mysql" = "rw,noexec"
For services
[runners.docker.services_tmpfs]
"/var/lib/mysql" = "rw,noexec"`
As docker/sdk generates a docker-compose.yml, here we have a volume or tmpfs mount definition, https://docs.docker.com/compose/compose-file/compose-file-v3/#long-syntax-3
The idea is to add a configuration to switch on/off mounting map database files into RAM.
The text was updated successfully, but these errors were encountered: