Re: [nft PATCH] tests: shell: Fix for spurious fail in maps/typeof_maps_add_delete

[Phil Sutter <phil@nwl.cc>]

On Fri, Oct 23, 2026 at 01:56:26PM +0200, Florian Westphal wrote:
> Phil Sutter <phil@nwl.cc> wrote:
> > Testing on a (probably slow) aarch64 machine yielded dump failure since
> > ruleset listing had '# count 1' appended to map size statement. The 3s
> > delay supposedly wasn't long enough for the element to be reapped after
> > disabling.
> > 
> > The test itself seems to perform all desired ruleset checks though, so
> > just drop the stored dumps to avoid this problem.
> 
> Annoying, but spurious failures suck more:

Well, there might be another way involving timeout and gc-interval
tweaks, but AIUI the test didn't intend to check gc reliability or
"performance". Moreover, delayed execution of gc due to high system load
is expected, right?

Sadly, it's not the only test which becomes unstable with system load
and delays. Looking at estimate_timeout in sets/0044interval_overlap_0,
variance in system load may also reduce test stability. :(

> Acked-by: Florian Westphal <fw@strlen.de>

Thanks, Phil