|
I know there have been about two messages in the archives touching on this
subject... but I'm still left here curious.
Currently I'm using direct routing and things are working just fine... This
is on a RH6.2 system with kernel version 2.2.16-3 rebuilt by me using the
RedHat 2.2.16-3 source.
On my active LVS, I very often find many of these sort of messages on the
console and in the kernel.log:
IP_MASQ:reverse ICMP: failed checksum from 24.26.156.111!
I receive this for a variety of completely different addresses on completely
different networks from completely different portions of the Internet.
I guess my question is -- why would I get this message? I might be more
comfortable with it if it was a *FORWARD* ICMP, but reverse? Wouldn't
reverse ICMP errors be errors when generating ICMP packets to be directed
FROM real servers back TO end-users?
I guess my question is ... is there just a small bug somewhere within the
confines of Linux and IPVS that accidently lets an IPVS packet fall through
to a section of masquerading code it is not supposed to hit and that code
mistakes this message as one with an error? Perhaps the real servers are
trying to send a source quench message or something and it hits the LVS and
gets confused..<?>
Any ideas why this is happening? I know others have had this message and
generally it's understood that it's not a big deal, but I would still like
to know why it occurs... Ya' know?
Anyway -- not a big deal. I just wanted to see if anyone had any ideas.
All the best --
Ted
|
