On Mon, Sep 21, 2026, at 19:57, Stuart Lamble wrote:
> On Mon, Sep 21, 2026, at 19:31, Stuart Henderson wrote:
>> On 2026/09/21 16:06, Stuart Lamble wrote:
>>> - Shutting down. Per Stuart Henderson's comments a couple of weeks ago, I
>>> have tested to see what happens. If I initiate the shutdown fairly soon
>>> after the server starts, it shuts down cleanly, no problem. If I wait a
>>> little bit, it seems to sit there. For my purposes, this is good enough,
>>> but it's not really acceptable for general release. (Logging is fine.)
>>
>> that's better than it was when I last tried updating things - it was
>> sitting there even when shutdown soon after start.
>
> I’ve dug into it a little bit. It’s looking like a race
> condition/deadlock, and I don’t have the C++ knowledge to try to figure
> it out quickly. I can try, but finding both the time and the
> inclination to work on it is going to be difficult for me for the next
> few months.
Further to this: I just shutdown my instance of Kea to replace it with a fresh
build (with the small fixes that have been made, including the now-verified
keactrl fix.)
It shutdown cleanly - it had been running for a few days at that point.
This adds more weight to the thought that it's a race condition that leads to a
deadlock. I've installed the debug- package; if/when it happens again, I'll try
to get a trace to feed to upstream. I doubt there's much more I can do than
that; I'm certainly not going to be able to trace through the code to try to
find the culprit. But the trace might point at some aspect of the code that
isn't doing the right thing from an OpenBSD perspective - I'll ask here for
assistance once I have the trace to see if we can track it down.
I think this is about as far as I can take the update for the time being; apart
from the hang on shutdown (not a small issue, I acknowledge, though we can
implement a workaround that just kills the errant process if it's still there
after a period of time), it looks good to me.
Thanks.
