Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

instrumentation and uninstrumentation is triggered too frequently when the collector is restarted #1085

Open
esara opened this issue Mar 22, 2024 · 0 comments
Labels
bug Something isn't working

Comments

@esara
Copy link

esara commented Mar 22, 2024

Describe the bug
The gateway or the collector process may be killed for various reason in a shared environment (even if we assign a request to memory). During the restart of the collector process, the instrumentor removes the instrumentation from all instrumented applications and after the collector come back it adds it back again.
This is fairly disruptive, especially in a cluster with several applications that have relatively heavy startup load (like java applications), restarting every instrumented app (potentially twice) all at the same time creates even more problems and potentially kill again the collector, causing some significant, maybe even ongoing churn.

To Reproduce
Create an environment with somewhat limited (or well sized?) resources, where the applications can all run at the same time as well as the odigos pipeline, but occasional spike can disrupt one or so process (which by itself should be able restart without impacting others).

Expected behavior
Tolerate disruptions in the pipeline, maybe wait before uninstrumenting applications and even then possible stagger the instrumentation/uninstrumentation

Additional context
We also discussed besides adding a memory limiter also setting a default (and configurable) request/limit on the collector. Which is good, but does not completely guarantee no disruptions.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bug Something isn't working
Projects
None yet
Development

No branches or pull requests

2 participants