On Mon, May 11, 2026 at 05:28:15PM +0200, Krzysztof Hałasa wrote:

> Now what should we do with this?
> a) should we increase the 200 ms timeout? Would it still fail eventually?
> b) should the timeout be configurable, maybe by the platform code?
> c) some other way?

If raising the timeout resolves the issue just raise the timeout, this
is not something that should ever happen in production and is just there
to ensure we don't completely deadlock on catastrophic errors.