[e-nsp] extreme-nsp Digest, Vol 74, Issue 9
Dudek, Michael
michael.dudek at warema.de
Thu Sep 30 11:03:27 EDT 2010
I configured the switch like you adviced. The only difference is that I configured the mirroring on a transit switch, because there I've enough gigabit-ports. But it will be no problem to make an EAPS-Master out of the switch. Nevertheless I can see a steady flow of health packets. Now I'm nervous until the trouble comes again.
-----Ursprüngliche Nachricht-----
Von: Robert Jacobs [mailto:rjacobs at phonoscope.com]
Gesendet: Mittwoch, 29. September 2010 23:50
An: extreme-nsp at puck.nether.net
Cc: Dudek, Michael
Betreff: RE: extreme-nsp Digest, Vol 74, Issue 9
I would set up a mirror on the eaps master switch and mirror one of
your problematic control vlans. You should see a nice steady flow of
health packets and nothing else on the vlan. Default would be one
second apart. They also have a sequence number and they should be in
order. If you are seeing them in order and are dropping some or have
other traffic on your control vlan then you have a problem.
Robert Jacobs Network Director
Phonoscope
6105 Westline Drive
Houston, TX 77036
832-615-7742 cell 281-830-2092
-----Original Message-----
From: extreme-nsp-bounces at puck.nether.net
[mailto:extreme-nsp-bounces at puck.nether.net] On Behalf Of
extreme-nsp-request at puck.nether.net
Sent: Wednesday, September 29, 2010 11:00 AM
To: extreme-nsp at puck.nether.net
Subject: extreme-nsp Digest, Vol 74, Issue 9
Send extreme-nsp mailing list submissions to
extreme-nsp at puck.nether.net
To subscribe or unsubscribe via the World Wide Web, visit
https://puck.nether.net/mailman/listinfo/extreme-nsp
or, via email, send a message with subject or body 'help' to
extreme-nsp-request at puck.nether.net
You can reach the person managing the list at
extreme-nsp-owner at puck.nether.net
When replying, please edit your Subject line so it is more specific
than "Re: Contents of extreme-nsp digest..."
Today's Topics:
1. Re: EAPS - Fail-timer-exp flag set. Domain state: Complete
(Fabian)
----------------------------------------------------------------------
Message: 1
Date: Wed, 29 Sep 2010 15:56:00 +0200
From: "Fabian" <fabian.extr at gmail.com>
To: "'Dudek, Michael'" <michael.dudek at warema.de>
Cc: extreme-nsp at puck.nether.net
Subject: Re: [e-nsp] EAPS - Fail-timer-exp flag set. Domain state:
Complete
Message-ID: <4ca34576.06df8e0a.7395.1132 at mx.google.com>
Content-Type: text/plain; charset="us-ascii"
Not sure that the counter is linked to the EAPS issue :
Rx-Invalid-Vlan-Intf -
Vlan interface for incoming vlan cannot be found.
Rx-Undersize-Pkt -
Packet length is less than the length of the header.
Rx-Invalid-8021Q-Tag -
VlanTypeLength field in Ethernet header does not match the default
Ethernet
value for 8021Q Tag.
Rx-Invalid-SNAP-Type -
SNAP value in Ethernet header is invalid.
Rx-Invalid-OUI -
OUI value in Ethernet header does not match 00:E0:2B.
Rx-EEP-Unsupported-Version -
Unsupported EEP version. EEP version should be 1
Rx-EEP-Invalid-Length -
Length in EEP header is greater than the packet length.
Rx-EEP-Checksum-Invalid -
EEP checksum is invalid.
Rx-Domain-Invalid -
Control vlan of incoming PDU does not have any EAPS domain associated
with
it.
Rx-Lif-Invalid -
LIF for ingress port cannot be found.
I would suggest to open a case at TAC.
You can also use the command : debug eaps check config (on EXOS) on each
EAPS switch to check if there is some configuration anomaly.
Regards, Fabian.
From: Dudek, Michael [mailto:michael.dudek at warema.de]
Sent: mercredi 29 septembre 2010 13:17
To: Fabian
Cc: extreme-nsp at puck.nether.net
Subject: AW: [e-nsp] EAPS - Fail-timer-exp flag set. Domain state:
Complete
Hello Fabian,
first of all I checked whether there are rx or tx errors on the common
links
(2 BD8810), result: no errors
Then I checked the common links with all possible commands like "sh eaps
counters shared-port 4:8". Result:
On all EAPS-Domains are only Rx-Seg-Health and tX-Seg-Health packets.
In addition the global counters for EAPS Shared-Ports are all null.
The command "sh eaps counters global" shows maybe some interesting
things:
"Rx-Invalid-Vlan-Intf : 73" and "Rx-Lif-Invalid : 5"
"Rx-Invalid-Vlan-Intf : 150" and "Rx-Lif-Invalid : 88"
But I don't know how to interpret these counters.
Furthermore I compared again the eaps configurations but there seems to
be
no misconfiguration. Also EpiCenter
reports only that some vlans are not protected, but they are! It has
something to do with case-sensitve, how you are writing the names of the
vlans, like "PB4" or "pb".
It seems more that something is teasing or targeting the cpu on the
masters.
But there are 3 EAPS-Rings that doesn't have that problem.
Two of them have nothing to do with the investigated common links, and
there
is not the named vlan "stammhaus" on them.
But 1 EAPS-Master doesn't report any problems, he seems to be
unimpressed
even though the vlan "stammhaus" is configured on that switch, too.
Any further ideas, suggestions.
Regards, mikeld
_____
Von: Fabian [mailto:fabian.extr at gmail.com]
Gesendet: Dienstag, 28. September 2010 23:53
An: Dudek, Michael
Cc: extreme-nsp at puck.nether.net
Betreff: RE: [e-nsp] EAPS - Fail-timer-exp flag set. Domain state:
Complete
Hi Mikeld,
This message should appear when the EAPS Master is losing EAPS PDU that
it
has sent over the ring.
Loss of EAPS Health-Check Packets could happen:
- Frames in error in the ring
- A switch in the ring that is having some trouble and loosing
frames
- Lot of frames going to CPU on the Master
- ..
Wireshark will not really help because EAPS PDU are isolated in the
Control
VLAN.
If this happened on all rings, I will check the Common Link (or EAPS
Shared-Link) status/statistics.
Fabian.
From: extreme-nsp-bounces at puck.nether.net
[mailto:extreme-nsp-bounces at puck.nether.net] On Behalf Of Dudek, Michael
Sent: mardi 28 septembre 2010 09:08
To: extreme-nsp at puck.nether.net
Subject: [e-nsp] EAPS - Fail-timer-exp flag set. Domain state: Complete
No matter whether it is XOS Version 12.4.1.7 v1241b7-patch1-7 nor
EW-Version
7.8e.2.1, at a certain time, mainly between 12.00 to 01.00 pm I get
hundreds of messages in the syslogs like
09/27/2010 12:47:32.81 <Info:EAPS.DmnInfo> EAPSD DSK33 - Fail-timer-exp
flag
cleared. Domain state: Complete
09/27/2010 12:47:32.81 <Info:EAPS.DmnInfo> EAPSD DSK33 - Fail-timer-exp
flag
set. Domain state: Complete
09/27/2010 12:47:28.86 <Info:EAPS.DmnInfo> EAPSD DSK33 - Fail-timer-exp
flag
cleared. Domain state: Complete
09/27/2010 12:47:24.82 <Info:EAPS.DmnInfo> EAPSD DSK33 - Fail-timer-exp
flag
set. Domain state: Complete
09/27/2010 12:47:17.81 <Info:EAPS.DmnInfo> EAPSD DSK33 - Fail-timer-exp
flag
cleared. Domain state: Complete
09/27/2010 12:47:17.81 <Info:EAPS.DmnInfo> EAPSD DSK33 - Fail-timer-exp
flag
set. Domain state: Complete
There are nearly all EAPS-Master from 24 EAPS-Rings affected. My
investigations shows the following:
All the eaps-rings have no configuration errors.
There are no rx- or tx-errors on the uplinks.
Network performance is not affected.
It must be a certain vlan from which the interference comes, this vlan
is
known.
It happens only on weekdays and during office hours
I have no details found with wireshark, but I'm not an expert with
wireshark
I found some references about IPv6 neighbor discovery protocol, there
are
some W7 PCs in that vlan.
My guess is that something teases the cpu of the eaps-masters but how
can I
catch that thing?
Does anyone have a suggestion how I can get this under control with EW
or
XOS or wireshark or anything else?
Thanks a lot, Mikeld
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<https://puck.nether.net/pipermail/extreme-nsp/attachments/20100929/ddab
ce63/attachment-0001.html>
------------------------------
_______________________________________________
extreme-nsp mailing list
extreme-nsp at puck.nether.net
https://puck.nether.net/mailman/listinfo/extreme-nsp
End of extreme-nsp Digest, Vol 74, Issue 9
******************************************
More information about the extreme-nsp
mailing list