spurrious test failures on s390x

[martinetd]

While backporting time64 fixes in https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/104520 the s390x ran into these test failures:

---- daemon::ntp_source::tests::test_timeroundtrip stdout ----
thread 'daemon::ntp_source::tests::test_timeroundtrip' (4916) panicked at ntpd/src/daemon/ntp_source.rs:707:35:
called `Option::unwrap()` on a `None` value
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace---- daemon::ntp_source::tests::test_deny_stops_poll stdout ----
thread 'daemon::ntp_source::tests::test_deny_stops_poll' (4914) panicked at ntpd/src/daemon/ntp_source.rs:748:13:
assertion failed: timestamp.is_some()
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

For these two, imestamped socket's recv() didn't provide a timestamp for some reason, I couldn't understand why in the short time I started at it

---- daemon::server::tests::test_server_serves stdout ----
thread 'daemon::server::tests::test_server_serves' (4920) panicked at ntpd/src/daemon/server.rs:330:14:
called `Result::unwrap()` on an `Err` value: Elapsed(())

This one might just be the builder being slow given it looks like a timeout, but it only ran once and I didn't try fiddling with the timeout as I don't have direct access to the builder, sorry.

The tests have been disabled on alpine for now

[martinetd]

For what it's worth, I don't have access to the builder any more than you, but if you have patches you'd like to try you can open a MR in aports with something like this to test:

diff --git a/testing/ntpd-rs/APKBUILD b/testing/ntpd-rs/APKBUILD
index abda79834e48..86b3d8d94f1a 100644
--- a/testing/ntpd-rs/APKBUILD
+++ b/testing/ntpd-rs/APKBUILD
@@ -1,10 +1,10 @@
 maintainer="Achill Gilgenast <achill@achill.org>"
 pkgname=ntpd-rs
 pkgver=1.9.0
-pkgrel=1
+pkgrel=2
 pkgdesc="Full-featured implementation of the Network Time Protocol, including NTS support"
 url="https://docs.ntpd-rs.pendulum-project.org/"
-arch="all"
+arch="s390x"
 license="Apache-2.0 OR MIT"
 makedepends="
 	cargo-auditable
@@ -18,6 +18,7 @@ source="$pkgname-$pkgver.tar.gz::https://github.com/pendulum-project/ntpd-rs/arc
 	ntpd-rs.initd
 	ntpd-rs-metrics-exporter.initd
 	bump-clock-steering-time64.patch
+	new-patch-to-test.patch
 	"
 options="net" # downloading rust crates
 
@@ -33,11 +34,7 @@ build() {
 check() {
 	# test failures reported here:
 	# https://github.com/pendulum-project/ntpd-rs/issues/2286
-	cargo test --frozen --release -- \
-		--skip daemon::spawn::nts::tests::allow_srv_direct_name_resolution \
-		--skip daemon::ntp_source::tests::test_timeroundtrip \
-		--skip daemon::server::tests::test_server_serves \
-		--skip daemon::ntp_source::tests::test_deny_stops_poll
+	cargo test --frozen --release
 }
 
 package() {
@@ -77,4 +74,5 @@ e4e1b053223bce6633b216c818f8f13bac520631e256ebace6e07d1a2b8d8290bb908d8ea9949b78
 0073d186837898a42e4adde5b6a06be12732ae3ca43e5d726b594ea774377b3ace61010965d821301d4773ae4ba0e79b2e52fb8258e4b1e9075a77bdeb6a51bd  ntpd-rs.initd
 0a75a0d2ac77e0754f180df880dc7b0ef2c525cde5c575ffccfafef4ba06a59cbcd7bd0a1a313649a0674def8353f3333be692054b7416f4b9302124f5be3aeb  ntpd-rs-metrics-exporter.initd
 1c7cb4564f069faee7de64920ace8f12aa366e46affaaa948c1f787aaaeb58c92dbc135fd45675bda72be826e69d265d0f3124e7f876e55d6be29d01f52cc41f  bump-clock-steering-time64.patch
+cf83e1357eefb8bdf1542850d66d8007d620e4050b5715dc83f4a921d36ce9ce47d0d13c5d85f2b0ff8318d2877eec2f63b931bd47417a81a538327af927da3e  new-patch-to-test.patch
 "

(If you post me a diff here I can do it for you, but I'm not adding much value here)

[davidv1992]

Thank you for splitting this off into its own issue. I might look into it at some point, but s390x is not exactly an important architecture for us, and as you note somewhat hard to test on (I am still looking at where we might get some access to systems running on that actual hardware more directly).