Comment # 2 on bug 1217029 from Thomas Blume

(In reply to Thomas Blume from comment #1)
> So, apparently the feedback from node1 to the container host is hanging.

Looks like there is a time keeping issue.
dmesg on the host shows:

-->
[  303.093179] clocksource: timekeeping watchdog on CPU0: Marking clocksource
'tsc' as unstable because the skew is too large:
[  303.094160] clocksource:                       'kvm-clock' wd_nsec:
480016638 wd_now: 47a9fb099a wd_last: 478d5e909c mask: ffffffffffffffff
[  303.095819] clocksource:                       'tsc' cs_nsec: 41654552965
cs_now: b3c0bc3362 cs_last: 9e6a91c09e mask: ffffffffffffffff
[  303.097036] clocksource:                       Clocksource 'tsc' skewed
41174536327 ns (41174 ms) over watchdog 'kvm-clock' interval of 480016638 ns
(480 ms)
[  303.098189] clocksource:                       'kvm-clock' (not 'tsc') is
current clocksource.
[  303.099101] tsc: Marking TSC unstable due to clocksource watchdog
--<

I can't see that on my private machine, where the test works.