Cleanup BTF

[grcevski]

The Linux BTF memory parsed by ebpf-go holds onto all the symbols. On my machine this is 30 MB. From this issue I found that we can cleanup this memory manually after we've loaded the programs cilium/ebpf#1063. It's hard to tell exactly when for performance reasons, but I want to see if I can pass the tests with manual cleanup on every tracer install.

Beyla started without any programs, resident state size of the memory after pprof which triggers GC:

Before:

VmRSS:	           83884 kB
RssAnon:	   50924 kB
RssFile:	   32960 kB

After:

VmRSS:	           43312 kB
RssAnon:	   10032 kB
RssFile:	   33280 kB

From 83MB to 43MB.

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 80.16%. Comparing base (0c7d38f) to head (d87f51b).

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1011      +/-   ##
==========================================
- Coverage   80.17%   80.16%   -0.01%     
==========================================
  Files         137      137              
  Lines       10930    10932       +2     
==========================================
+ Hits         8763     8764       +1     
- Misses       1685     1686       +1     
  Partials      482      482              

Flag	Coverage Δ
integration-test	55.00% <100.00%> (-0.15%)	⬇️
k8s-integration-test	57.64% <100.00%> (-0.08%)	⬇️
oats-test	37.32% <100.00%> (+0.01%)	⬆️
unittests	50.01% <0.00%> (-0.03%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[mariomac]

That's great!

[marctc]

Nicee!!