Sporadically I get 100 to 200 redis timeout exceptions. Because i use sessions on redis (via the official MS package) these become "500s"/yellow-screen-of-death to the client.
Investigating those timeouts using application insights, i found that the root cause are some "Eval" commands. I used the term "some" because the keyHashSlot reported on the exception message varies.
The exception also reports IOCP and Worker threads available (min is 100 and there are usually more than 10 free)
So, how can i dive even deeper and figure out why those Evals are dimming out? Are they timing out because of their content (the script that it's been executed)? or they are wasting too much time on the queue?
Have you taken a look at this? https://gist.github.com/JonCole/db0e90bedeb3fc4823c2#file-diagnoserediserrors-clientside-md.
Especially, the threadpool throttling bit: https://gist.github.com/JonCole/db0e90bedeb3fc4823c2#burst-of-traffic
Could you be experiencing high CPU on your client or threadpool throttling?