kern/57732: Some testbed installs time out during set extraction

[gson%gson.org@localhost (Andreas Gustafsson)]

>Number:         57732
>Category:       kern
>Synopsis:       Some testbed installs time out during set extraction
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Wed Nov 29 10:00:00 +0000 2023
>Originator:     Andreas Gustafsson
>Release:        NetBSD-current since about June 2023
>Organization:
  
>Environment:
System: NetBSD
Architecture: i386
Machine: i386
>Description:

Recently, a number of test runs on the i386 testbed on
babylon5.netbsd.org have stalled and timed out during the set
extraction phase of installation.  The first of these failures was
using -current sources from 2023-07-28.  Before that, there had not
been similar failures since 2018.

The symptoms are similar to those in PR 47030, but since the present
issue appeared only recently and wapbl is not involved (as far as I
know), I'm treating it as a new issue and filing a separate PR.

Logs can be found at:

  [failures from before 2018-08 omitted]
  http://releng.netbsd.org/b5reports/i386/#2018.08.12.10.43.04
  http://releng.netbsd.org/b5reports/i386/#2018.12.01.13.32.55
  [almost five trouble-free years passed]
  http://releng.netbsd.org/b5reports/i386/#2023.07.28.04.27.30
  http://releng.netbsd.org/b5reports/i386/#2023.08.09.21.09.09
  http://releng.netbsd.org/b5reports/i386/#2023.08.23.19.16.14
  http://releng.netbsd.org/b5reports/i386/#2023.09.08.12.01.56
  http://releng.netbsd.org/b5reports/i386/#2023.09.11.22.09.28
  http://releng.netbsd.org/b5reports/i386/#2023.09.21.14.00.34
  http://releng.netbsd.org/b5reports/i386/#2023.09.24.10.13.44
  http://releng.netbsd.org/b5reports/i386/#2023.09.25.07.11.08
  http://releng.netbsd.org/b5reports/i386/#2023.09.30.21.00.43
  http://releng.netbsd.org/b5reports/i386/#2023.10.01.21.04.10
  http://releng.netbsd.org/b5reports/i386/#2023.10.06.14.40.06
  http://releng.netbsd.org/b5reports/i386/#2023.10.13.19.30.28
  http://releng.netbsd.org/b5reports/i386/#2023.10.19.23.43.40
  http://releng.netbsd.org/b5reports/i386/#2023.10.24.18.01.31
  http://releng.netbsd.org/b5reports/i386/#2023.10.25.06.01.29
  http://releng.netbsd.org/b5reports/i386/#2023.10.25.17.18.53
  http://releng.netbsd.org/b5reports/i386/#2023.10.28.21.06.05
  http://releng.netbsd.org/b5reports/i386/#2023.10.29.14.55.16
  http://releng.netbsd.org/b5reports/i386/#2023.11.05.17.52.09
  http://releng.netbsd.org/b5reports/i386/#2023.11.05.21.13.06
  http://releng.netbsd.org/b5reports/i386/#2023.11.07.13.31.26
  http://releng.netbsd.org/b5reports/i386/#2023.11.14.03.03.18
  http://releng.netbsd.org/b5reports/i386/#2023.11.15.14.03.19
  http://releng.netbsd.org/b5reports/i386/#2023.11.17.22.14.16
  http://releng.netbsd.org/b5reports/i386/#2023.11.18.19.46.55
  http://releng.netbsd.org/b5reports/i386/#2023.11.21.19.59.07
  http://releng.netbsd.org/b5reports/i386/#2023.11.24.17.31.03
  http://releng.netbsd.org/b5reports/i386/#2023.11.27.10.03.40

There has also been one similar failure on sparc:

 http://releng.netbsd.org/b5reports/sparc/#2023.09.24.20.41.52

Mostly for my own reference, the commands I ran on babylon5 to
identify the failed i386 installs were:

  cd /bracket/i386/results
  find . -name install.log.gz | xargs zgrep -a ETA.*Traceback | cat -v | perl -pe 's/^.......//; s/(?<=.{20}).{1,}(?=.{45}$)/.../' | sort

>How-To-Repeat:

>Fix: