Red Hat Bugzilla – Bug 77803
ping gives warning: time of day goes back
Last modified: 2015-03-04 20:11:40 EST
From Bugzilla Helper:
User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; .NET CLR
Description of problem:
When pinging hosts, intermittently, the error 'Warning: time of day goes back,
taking countermeasures.' Countermeasures, in this case, appears to be
returning a trip time of 0 usecs.
I just recently upgraded the machine from kernel-2.4.9-31 to kernel-2.4.18-
17.7.x (both RH7.2 errata kernels, stock). The problem did not exist before
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. upgrade kernel
2. ping -U -s 56 host 10
Actual Results: Warning given, ping returns bad 0 value for the time. After
first return, subsequent warnings take a long time (> 5secs) to appear.
Expected Results: Ping doesn't give warning and gives correct ping times.
This bug has been filed under 19952 (CLOSED) for RH7.0 and as a duplicate of
that bug (60765) for RH 7.2. I'm filing separately because the old bug has
been closed for over six months, and the kernel was released in October. This
may be different issue, or it may be a regression, but I want it seen.
I upgraded to the latest rawhide version of iputils (I normally don't like to
upgrade past anything outside the distro and its errata, but I really need
this), and the warnings have gone away (presumably because they're only printed
in verbose mode now; see changelog).
I'm still extremely nervous about those 0usec ping times, though, when the
errors was printed. We're measuring ping times for latency calculations, so
0usec ping times are actually a big deal to us.
Example output showing what we're seeing, even with the new code:
$ ping -U -v -s 56 220.127.116.11
PING 18.104.22.168 (22.214.171.124) from 126.96.36.199 : 56(84) bytes of data.
64 bytes from 188.8.131.52: icmp_seq=1 ttl=254 time=0.609 ms
64 bytes from 184.108.40.206: icmp_seq=2 ttl=254 time=0.756 ms
64 bytes from 220.127.116.11: icmp_seq=3 ttl=254 time=0.664 ms
64 bytes from 18.104.22.168: icmp_seq=4 ttl=254 time=0.635 ms
64 bytes from 22.214.171.124: icmp_seq=5 ttl=254 time=0.744 ms
64 bytes from 126.96.36.199: icmp_seq=6 ttl=254 time=0.906 ms
64 bytes from 188.8.131.52: icmp_seq=7 ttl=254 time=0.616 ms
64 bytes from 184.108.40.206: icmp_seq=8 ttl=254 time=0.695 ms
64 bytes from 220.127.116.11: icmp_seq=9 ttl=254 time=0.634 ms
64 bytes from 18.104.22.168: icmp_seq=10 ttl=254 time=0.720 ms
Warning: time of day goes back (-7806011us), taking countermeasures.
64 bytes from 22.214.171.124: icmp_seq=11 ttl=254 time=0.000 ms
Warning: time of day goes back (-7806078us), taking countermeasures.
64 bytes from 126.96.36.199: icmp_seq=12 ttl=254 time=0.000 ms
64 bytes from 188.8.131.52: icmp_seq=13 ttl=254 time=0.722 ms
64 bytes from 184.108.40.206: icmp_seq=14 ttl=254 time=0.833 ms
64 bytes from 220.127.116.11: icmp_seq=15 ttl=254 time=0.630 ms
Do you have ntp running on that host by chance? This might affect the time
setting of your machine and cause the effect. With the latest iputils on a RH 9
i haven't see the effect at all anymore (where i was able to reproduce it with
older releases and kernels).
As you've been using the latest iputils package i can only think of 2 causes:
either ntp running on your machine or that the 7.2 errata kernel still has
issues with the precise time interface.
Read ya, Phil
I've tried various setups and installations over the last few months and the bug
didn't appear on any of the systems anymore, so i'm closing the bug as fixed in
the current release.
Read ya, Phil