CAS fails randomly even though it shouldn't

[SeverinAlexB]

I am building a republisher. Following the instructions for CAS in the publish() method, I resolve the most recent packet first before I publish the packet again.
Problem: publishing fails randomly even though no newer packet has been published. See this test which reproduces the problem.

Output:

0 Ok
1 Ok
2 Ok
3 Failed to publish packet. Compare and swap failed; there is a more recent SignedPacket than the one seen before publishing
thread 'publisher::tests::republish_cas_test' panicked at pkarr-publisher/src/publisher.rs:197:17:

[Nuhvi]

Republishers don't run the risk of overwriting anything, so they don't need to use cas at all.

[SeverinAlexB]

Sure. Issue is still here though

[Nuhvi]

This turns out to be a bug with how the cas is sent to relays as an HTTP Date, which doesn't have enough resolution (seconds instead of microseconds). Still thinking if I can salvage If-Unmodified-Since or if we have to use a custom header.