Set -j 1 in macOS nix unit tests

[Anviking]

Issue Number

#2472 / ADP-970

Overview

• Set -j 1 for hydra Mac builds, in hope this alleviates unit timeouts

Comments

A hypothesis is that #2472 is caused by heavy load and unfocused
resources from running the tests concurrently, risking that the slowest
hspec runner - and thererefore the stdout - being silent for 900s causing
hydra to timeout.

Setting -j 1 should hopefully focus the resource we have in one place. It
should go silent less often, at the expense of the full run getting slower.

Testing locally

Tested locally with nix-build -A checks.cardano-wallet-core. Seems to work.

It now takes 399s instead of 299.7877 on my machine (Not that dramatic slowdown).

We have a -with-rtsopts=-N4 in the .cabal file. I'm not sure if that's good or not, but I suspect it doesn't matter as much as the -j option.

[Anviking]

bors try

[iohk-bors]

try

Build succeeded:

• buildkite/cardano-wallet
• ci/hydra-build:required

[rvl]

I believe that the main reason why tests were getting stuck on macos more is the problem fixed by #2734.
That problem is a simple case of readMVar waiting forever for a putMVar that never happens.
The exception condition (port in use, address unavailable, etc) happened more with macos.

[Anviking]

I believe that the main reason why tests were getting stuck on macos more is the problem fixed by #2734.
That problem is a simple case of readMVar waiting forever for a putMVar that never happens.

The entire LoggingSpec was marked pending on macOS in #2709, after which we have observed one proper bors timeout in #2724 (comment), and IIRC a couple more on normal PR non-bors builds.

To me it seems like marking it pending had little to no effect.

That problem is a simple case of readMVar waiting forever for a putMVar that never happens.

From ./scripts/bors-stats.rb list --tag "#2472" --details true there are no failures looking like direct LoggingSpec timeouts. If the LoggingSpecs spawned a lot of threads, and lots of work, perhaps this could contribute to the first hspec thread being slow. "readMVar waiting forever" doesn't sound like a lot of work though.

I think-j 1 is very much worth a shot.

[rvl]

OK Thanks @Anviking

[rvl]

This is very very minor, but just so you know, there's a handy lib.optionals function which does the same thing.

Suggested change

	unit.testFlags = if pkgs.stdenv.hostPlatform.isDarwin then ["-j" "1"] else [];
	unit.testFlags = lib.optionals pkgs.stdenv.hostPlatform.isDarwin ["-j" "1"];

Either way is perfectly fine though.

[rvl]

I think -j 1 is very much worth a shot.

Fair enough.

bors r+

[iohk-bors]

Build succeeded:

• ci/hydra-build:required
• buildkite/cardano-wallet