Potential 100% CPU hang in some circumstances

[driis]

A service we have in production using EventFlow started exhibiting 100% CPU hangs periodically.

We have identified that this happens due to concurrent, not synchronized, access to the dictionaries used in ReadModelSqlGenerator. A single instance of this class is shared between all threads and requests.

This call stack (redacted) was found on multiple threads in a hang dump taken from the process:

0:023> !clrstack
OS Thread Id: 0x4cc0 (23)
        Child SP               IP Call Site
000000033173c9d0 00007ffc27c0669e System.Collections.Generic.Dictionary`2[[System.__Canon, mscorlib],[System.__Canon, mscorlib]].FindEntry(System.__Canon) [f:\dd\ndp\clr\src\BCL\system\collections\generic\dictionary.cs @ 298]
000000033173ca40 00007ffc27c05e64 System.Collections.Generic.Dictionary`2[[System.__Canon, mscorlib],[System.__Canon, mscorlib]].TryGetValue(System.__Canon, System.__Canon ByRef) [f:\dd\ndp\clr\src\BCL\system\collections\generic\dictionary.cs @ 499]
000000033173ca80 00007ffbd3d38c1a EventFlow.Sql.ReadModels.ReadModelSqlGenerator.CreateUpdateSql[[System.__Canon, mscorlib]]()
000000033173cb30 00007ffbd3d244a6 EventFlow.MsSql.ReadStores.MssqlReadModelStore`1+d__7[[System.__Canon, mscorlib]].MoveNext()

Replacing the dictionaries in the ReadModelSqlGenerator with ConcurrentDictionary<,> should fix this issue. I intend to provide a pull request for this shortly.

[rasmus]

There's a blog post on MSDN here on the subject as well
https://blogs.msdn.microsoft.com/tess/2009/12/21/high-cpu-in-net-app-using-a-static-generic-dictionary/

[rasmus]

Automatically closed when the PR got merged. Releasing 0.66 with the fix now