Well folks, as happens many times the resolution is a simple one...but it took hours for support to find it.
Data in file qusrsys/qatocifc caused our issue.
SELECT INTERNET, CNNTYPE FROM qusrsys/qatocifc returned the data below.
If you have P's at connection type for ip's other than the VETH's then "Proxy ARP's" can occur, which is what "flooded" or overwhelmed the network. We ended up recreating the interfaces but we probably could have resolved this by changing the connection type for the non VETH's to V or N.
Display Data
Data width . . . . . . : 33
Position to line . . . . . Shift to column . . . . . .
....+....1....+....2....+....3...
Internet Address Connection Type
V5R5M0
127.0.0.1 N
172.18.1.54 V
172.18.1.55 V
172.18.3.54 V
172.18.3.55 V
172.18.1.252 P
172.18.3.252 P
******** End of data ********
-----Original Message-----
From: DrFranken <midrange@xxxxxxxxxxxx>
To: midrange-l@xxxxxxxxxxxx
Sent: Sun, May 15, 2011 6:43 pm
Subject: Re: V6 Upgrade Trouble
Doesn't smell like IBM i here, really.
Does this still happen when the IP interface is shut down?
How about when the Ethernet line description is varied off?
Try this just for sanity. Edit your Ethernet Line description and put
n an override for the local adpater address such as 020202020202.
Actually any value from 020000000000 to FEFFFFFFFFFF. ) Then bring the
ine back up.
I think this is a network issue where something in the network has hung
p on this MAC address/IP Address pair and that it's not IBM i doing it,
ather the conflict in address stuck in some network gear.
Can the network guys tell you where the other device with the same
ddress claims to be?
Going back to V5R4 just seems very drastic to me!
- Larry "DrFranken" Bolhuis
On 5/14/2011 10:54 PM, fbocch2595@xxxxxxx wrote:
After some work we started up IP again and we're getttin TCP1B21 for and
thernet line on a different mac address. We ran a trace this time and
opefully IBM can make sense out of it. IBM said they'd find it tough to
esolve since when we endsbs the network guys still can't connect to the
etwork.
I hate to say it but it looks like we might go back to V5R4, hopefully the
ewtwork connection will be fine after that.
-----Original Message-----
From: DrFranken<midrange@xxxxxxxxxxxx>
To: midrange-l@xxxxxxxxxxxx
Sent: Sat, May 14, 2011 8:39 pm
Subject: Re: V6 Upgrade Trouble
Well the evidence does seem to point to the system I must agree. Since
he 5706 must be assigned to a single partition that means there are no
ther partitions that can be blamed.
What type of packets are they seeing? If they can tel they are IPV6
tuff they should be able to tell us what they are.
You said you've told the system not to start IPV6 and it still happens?
So the obvious next question is have you spoken to IBM yet?
- Larry
n 5/14/2011 8:29 PM, fbocch2595@xxxxxxx wrote:
Hi Larry, thanks for responding. When we endsbs *all the flooding stops.
t's a 5706.
Since the flooding started at 3:30 am and that's when we were upgrading (lic
robably done by then) the network guys are saying it must have been us (which
eems correct)
-----Original Message-----
From: DrFranken<midrange@xxxxxxxxxxxx>
To: midrange-l@xxxxxxxxxxxx
Sent: Sat, May 14, 2011 7:59 pm
Subject: Re: V6 Upgrade Trouble
I have done MANY of these and never seen IPV6 flood a network. Actually
've never seen IPV4 floods from an i either for that matter.
Does the flood stop while the system is IPLing or does it continue
hroughout? Frankly the only part of IPV6 that actually is configured
y default is the loopback address ::1. There is no IPV6 configured on
ny Ethernet line unless you did so so this on the surface doesn't make
ny sense at all.
I'm assuming that the network guys are saying 'from this LPAR' because
he MAC address is a match to that of your Ethernet line. You didn't say
hat sort of Ethernet card the machine has (i.e. 5706 or IVE card for
xample)
Can we be sure the switch itself isn't at fault here? Tried any other
orts on the switch?
- Larry "DrFranken" Bolhuis
On 5/14/2011 5:55 PM, fbocch2595@xxxxxxx wrote:
Hi Folks, overnight we upgraded from V5R4 to V6R1M1 (LIC at V6R1M1) and
fter
n upgrade of about 6 hours (including the object conversion) our network
tarted to be flooded by IPV6 traffic from the newly upgraded lpar and it's
ausing slowdowns all over the network and affecting users and production.
as
nyone else had similar circumstances or issues like this? I did not know or
ealize that IPV6 would start so we changed the cmd default so STRTCP would
ot
tart IPV6 and we also CHGIPLA to Start TCP/IP *NO and we've ipl'd several
ime
nd our network folks continue to say the newly upgraded system is flooding
he
etwork with traffic.
If anyone has any help or info to share on this type of situation, plz let
e
now
Management wants to go back to the prior release!
Thanks
-
his is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing list
o post a message email: MIDRANGE-L@xxxxxxxxxxxx
o subscribe, unsubscribe, or change list options,
isit:
http://lists.midrange.com/mailman/listinfo/midrange-l
r email: MIDRANGE-L-request@xxxxxxxxxxxx
efore posting, please take a moment to review the archives
t
http://archive.midrange.com/midrange-l.
-
his is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing list
o post a message email: MIDRANGE-L@xxxxxxxxxxxx
o subscribe, unsubscribe, or change list options,
isit:
http://lists.midrange.com/mailman/listinfo/midrange-l
r email: MIDRANGE-L-request@xxxxxxxxxxxx
efore posting, please take a moment to review the archives
t
http://archive.midrange.com/midrange-l.
midrange.com
