DEV Community

Aykut
Aykut

Posted on

Stackify Retrace shuts down on 31 March 2027: a practical .NET (and IIS) move to OpenTelemetry

BMC Helix has formally set 31 March 2027 as the end of life for Stackify Retrace. The official customer letter is unambiguous: Retrace is SaaS-only, so after that date you cannot log in, ingest data, or read stored data. There is no extended support, and there is no automated conversion off Retrace's agents and profilers.

If you are a .NET team, that means a real inventory-and-replace job: NuGet packages, Stackify.json, log appenders, custom ProfileTracer operations, custom metrics, and (on Linux or in containers) the profiler GUID in environment variables.

I built a small free tool for the inventory part.

What is going away (and what is not)

  • Retrace, the production APM, shuts down on 31 March 2027. Until then it is in maintenance mode: no new features, only critical fixes.
  • Prefix, the local profiler, is not being switched off. Premium support and updates for Prefix end on the same date; BMC says Prefix premium features are already free. Prefix is not a production monitoring replacement.
  • Exporting history is your job. Logs can be exported in files of up to 100,000 entries or 50 MB each. Custom dashboards can be saved as PDFs. Alert rules have no documented export — write them down.
  • BMC recommends Helix AIOps with OpenTelemetry and, in the same guide, says it is not a like-for-like replacement for Retrace. Many small and mid-size .NET teams will be better served by OpenTelemetry plus a backend they already use, such as Azure Monitor or Grafana.

Ownership, for context: Netreo acquired Stackify in 2021; BMC announced the Netreo acquisition in April 2024.

What OpenTelemetry will and will not copy

BMC's own comparison of OpenTelemetry auto-instrumentation with the Retrace profilers is the honest starting point:

  • Comparable: incoming request tracing, outbound HTTP, W3C distributed context, unhandled exceptions.
  • Partial: database spans (depends on the client), log-framework integration (maturity varies; NLog / log4net / Serilog need an explicit bridge).
  • Different: method-level call trees, first-chance (caught) exceptions, query parameters and row counts, continuous profiling, web-server pipeline breakdowns.

If your team lives in the Retrace method tree, plan extra manual ActivitySource work or a vendor agent for those services. Do not promise a silent, identical lift.

The good news for older apps: OpenTelemetry .NET zero-code instrumentation supports .NET Framework 4.6.2+ and IIS (Register-OpenTelemetryForIIS). For new ASP.NET Core services landing on Azure Monitor, Microsoft's current distro is the Microsoft.OpenTelemetry package — do not initialize it together with the older Azure Monitor OpenTelemetry Distro.

A free scanner: Retrace.Exit

Retrace.Exit is a dotnet tool. Point it at a solution. It walks project files, config, and source, and writes a Markdown or HTML report plus starter OpenTelemetry snippets. It does not rewrite your code, it does not send data over the network, and it never prints API keys.

dotnet tool install -g Retrace.Exit
retrace-exit ./src --format markdown --out retrace-exit-report.md
retrace-exit ./src --backend azure-monitor --emit-snippets ./otel-snippets
Enter fullscreen mode Exit fullscreen mode

--backend can be azure-monitor (default), grafana, otlp, or signoz.

What it looks for today:

  • Official Stackify NuGet and packages.config references (StackifyLib, StackifyMiddleware, Serilog.Sinks.Stackify, NLog.Targets.Stackify, StackifyLib.log4net, Stackify.AzureWebApps, and others), matched by package id, not by prefix (so StackifyLib.AspNetCore is not reported as StackifyLib).
  • Stackify.json — including AppName, Environment, CaptureAllExceptions, and Instrumentation[] counts.
  • appsettings*.json Stackify sections and Web.config / App.config keys (Stackify.ApiKey, Stackify.AppName, Stackify.Environment).
  • Log hooks: WriteTo.Stackify(), ConfigureStackifyLogging, StackifyAppender, StackifyTarget.
  • Code: ProfileTracer.CreateAsOperation, StackifyLib.Metrics (including unqualified Metrics.Count / Metrics.Time when using StackifyLib is in scope), RequestTracerMiddleware, StackifyHttpModule.
  • The Retrace profiler GUID cf0d821e-299b-5307-a3d8-b283c03916da in Dockerfiles and environment files.

A scan of the two fixtures that ship with the repo (an ASP.NET Core shop and a .NET Framework 4.6.2 / IIS-style project) currently reports 24 findings (15 critical, 9 warning). The fake API keys in those fixtures do not appear in the report.

Source and issues: github.com/aykutmayali/retrace-exit (MIT). The package id is Retrace.Exit; the command is retrace-exit.

A migration order that keeps every app monitored

This is the order BMC itself recommends, and it is the one I use:

  1. Inventory each app, runtime, host, and the Retrace features you actually rely on.
  2. Export logs and dashboard PDFs you must keep.
  3. Pilot OpenTelemetry (or a vendor agent) on one representative service. Keep Retrace running in parallel.
  4. Move the rest in waves. Remove the Retrace profiler from a service only when the new path covers it.
  5. Uninstall the Retrace agents well before 31 March 2027.

Use an OpenTelemetry Collector if you want to dual-export to two backends while you decide.

If you want help with the actual move

The scanner is free. If you would rather have someone walk an ASP.NET Core or IIS / .NET Framework app through the cut-over — inventory, pilot, Azure Monitor or Grafana, then removal of the Retrace agents — I offer a fixed-price migration package.

15-minute fit check: cal.com/aykut-ovvudn/15min

Service page: aykutmayali.github.io/launch-audit

Top comments (0)