[nSLUG] Ping cares about my kernel

D G Teed donald.teed at gmail.com
Tue Jan 28 10:30:37 AST 2014


This has been around for a long time, but I've only noticed it today.

Sometimes, ping, in the middle of not reaching
a host, will print:

WARNING: kernel is not very fresh, upgrade is recommended.

I had this happen on Redhat EL 6.4 yesterday, and it is a closed bug
from 2003 in redhat bugzilla.

https://bugzilla.redhat.com/show_bug.cgi?id=73660

It is also an ancient Debian bug, with a redo in 2013:

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=709052

You wouldn't think it is a tough bug to squash.  Just tell upstream to get
it out of there,
it can't have any idea what kernel level is appropriate.  But that is the
view
of me wanting to remove an annoyance, which is also the focus of these bug
reports.

What kind of thinking was behind making this warning in ping?

Looking at the code for iputils from http://www.skbuff.net/iputils/, I see:


                                if (error_pkt) {

acknowledge(ntohs(icp1->un.echo.sequence));
                                        if (working_recverr) {
                                                return 0;
                                        } else {
                                                static int once;
                                                /* Sigh, IP_RECVERR for raw
socket
                                                 * was broken until 2.4.9.
So, we ignore
                                                 * the first error and warn
on the second.
                                                 */
                                                if (once++ == 1)
                                                        fprintf(stderr,
"\rWARNING: kernel is not very fresh, upgrade is recommended.\n");
                                                if (once == 1)
                                                        return 0;
                                        }
                                }

It prints the warning during the error processing, and makes some assumption
this is a kernel related issue.  This is really a case of "This shouldn't
happen"
style of error, but judging by the comment, it seems to be related
to a known kernel bug from back in the days of 2.4.  The sporadic nature
of the warning combined with the lack of a good error could mean
we'll never learn what is going wrong when this happens.  Maybe it should
dump out some of the raw info it has at that point.  The current error is a
distraction from possibly discovering something else.

Didn't we cover this in some episode of Star Trek?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://nslug.ns.ca/pipermail/nslug/attachments/20140128/ae62c768/attachment-0001.html>


More information about the nSLUG mailing list