Arne Welzel

> @awelzel I have tweaked the work-stealing scheduler to avoid spinlocks and yield(). Could you check out the branch topic/neverlord/work-stealing in your auxil/broker and see if that changes anything? It...

@timwoj - how about closing this and if profiling in production environments indicates there's issues fix these specifically?

> Another possible piece of data would be the VERSION file from any of the submodules, if one exists. The only problem I see with the submodule data I suggested...

> I made a very naive mistake and ended up with this crash/segfault This is now caught as a runtime error instead of a segfault. Reiterations of the same underlying...

> The same plugin might be installed multiple times in different versions; this could either produce "weird" behavior or even lead to a Zeek installation which cannot be started anymore...

With #2767 added, the current zkg change/proposal is at https://github.com/zeek/package-manager/pull/157

Possibly cause of OOMs reported by [h-mikami on Slack](https://zeekorg.slack.com/archives/CSZBXF6TH/p1692608209819579?thread_ts=1685587437.627989&cid=CSZBXF6TH).

These aren't exactly the same issues as the half-duplex scenario, but hopefully addressing state growth can address these, too. * A single dropped packet from the responder side triggers unbounded...

The original problem seems to actually have been fixed with #2329.

Chatting with @vpax who had a a IPv4 only use case here. He "hacked" this problem by encoding additional information into an IPv6 address (not sure if it was vlan-id...