Diagnostic techniques for .NET running on Linux and within containers
summary
Shows concrete commands and outcomes: use dotnet-counters for live EventCounter and Meter metrics (example CPU ~35% under load), capture CPU and GC with dotnet-trace (convertible to Chromium/speedscope flamegraphs), collect memory with dotnet-dump and inspect heaps (example 4,993 strings = 361 KB; a growth of 508 objects ≈108 MB between dumps), analyze in Visual Studio or SOS, run dotnet-monitor as a dotnet-monitor sidecar (requires ptrace / cap_sys_ptrace or kernel ≥4.8 and avoid tmpfs for large dumps).
AI-written summary · about
chapters (13)
- 0:00.NET diagnostics on Linux
- 1:37Diagnostic tools overview
- 4:06dotnet-* tools intro
- 8:53dotnet-counters demo
- 11:10dotnet-trace collection
- 14:54Trace analysis in VS
- 20:40Memory dumps collection
- 27:46Dumps analysis in VS
- 33:25Heap analysis insights
- 37:47Deadlock debugging
- 39:57Cloud diagnostics with dotnet-monitor
- 42:04dotnet-monitor in Kubernetes
- 48:53Closing & survey
