b10c

I'm taking a look at this.

``` node1 [httpworker.2] [rpc/request.cpp:181] [parse] [rpc] ThreadRPCServer method=addpeeraddress user=__cookie__ node1 [httpworker.2] [logging/timer.h:58] [Log] [addrman] CheckAddrman: new 2, tried 1, total 3 started node1 [httpworker.2] [logging/timer.h:58] [Log] [addrman] CheckAddrman: completed (0.60ms)...

@mzumsande is there a reason why `addpeeraddress` should return true if it fails to put in in the table we asked it to do? Sounds like a bug to me....

Improving `addpeeraddress tried=true` failure behavior in #28998

Had a brief look. The `mempoolrej` log message shows that we indeed rejected the transaction but the test didn't see the reject. Can't reproduce this locally with about 20 test...

Other CI runs with similar behavior: - https://cirrus-ci.com/task/6650565413830656?logs=ci#L3531

I've been trying to reproduce this by re-running the CI on master https://github.com/bitcoin/bitcoin/commit/49b87bfe7e2799d25ce709123ecafa872b36e87a: https://cirrus-ci.com/task/4749392670883840 So far all runs passed..

Thanks, didn't know `rr`. Especially the chaos mode (https://robert.ocallahan.org/2016/02/introducing-rr-chaos-mode.html) seems to be helpful here. However, rr and bpf don't work well together yet. `rr` asserts when running the python USDT...

Gave this another look. No luck with `rr` and haven't been able to reproduce this locally or in the CI.

I'll give this another look next week.