Re: NIC in FAILED status -- MIDRANGE-L

Default action when a network interface fails is an inquiry message to QSYSOPR that needs to be answered with an R (Retry). I chased my tail with this a number of years ago and finally automated the reply so it didn't hang.

Mike Cunningham wrote:

We ran all week without a problem, including a nightly ENDTCP/STRTCP but then had this problem occur again this past weekend when we did a full powerdown/restart. But this time we found out the problem and we think we have it fixed but I wanted to share the resolution for the archives.

The cardinal rule of networking CHECK THE CABLES!!!

This time, we switched the ip address that was having problems to a spare NIC in the iSeries so we had time today to investigate. The server admin said the cisco switch was reporting it saw a 1GB full duplex connection but the iSeries never showed anything but "VARY ON PENDING" and "FAILED". Since this was only happening on a power up I suspected something was up with the auto negotiate process so I tried all hardcoded combinations and found ...

Auto/auto did not work
100MB/Full worked
100MB/half worked
100MB/auto worked
Auto/full did not work
Auto/half did not work
1GB/full did not work

Forced 100MB and any duplex are the only ones that worked.

The network port was a 100MB/1GB port but it appeared to only run at 100MB. Set everything back to auto and swapped the cable and it varied on immediately and at 1GB. Did 10 more vary on/offs and each time it connected at 1GB immediately. When this started the server admin had also moved the connection to a different switch port (and forgot to tell me) in addition to upgrading the server switch IOS. He kept thinking the IOS upgrade caused the problem. The cable either was bad all along and we were just lucky or moving it was enough to break a wire.

-----Original Message-----
From: midrange-l-bounces@xxxxxxxxxxxx [mailto:midrange-l-bounces@xxxxxxxxxxxx] On Behalf Of lbolhuis@xxxxxxxxxx
Sent: Monday, January 07, 2008 2:36 PM
To: Midrange Systems Technical Discussion
Subject: RE: NIC in FAILED status

Mike,

For GbE stuff we strongly recommend *AUTO *AUTO for speed and duplex on both ends of the connection. With 10/100 stuff we still recommend forcing speed and duplex on BOTH the switch and in i5/OS. If the switch is not GbE capable then set both for the highest speed supported. But also VERIFY that the switch AND i5/OS are reporting what you've requested.
Occasionally we DO see even with the settings forced that one side does *FULL duplex and the other side does *HALF. In these cases settings for *HALF on both ends seem the only way to get them to make nice. I don't believe this is your problem at this point.

What is your setting for the CMNRCYLMT on the Ethernet line? If it is
set to (1 0) then the line will retry forever (a good thing) while if it's anything else it will only try some for a while and then go FAILED. With the (1 0) Setting it does appear to magically heal itself. Frankly I think your network guys did 'something' but how can you prove that?

- Larry

Larry Bolhuis IBM Certified Advanced Technical Expert -
System i Solutions
Vice President IBM Certified Systems Expert:
Arbor Solutions, Inc. System i Technical Design and
Implementation V5R4
1345 Monroe NW Suite 259 eServer i5 iSeries LPAR Technical
Solutions, V5R3
Grand Rapids, MI 49505 IBM Certified Specialist
System i Integration with BladeCenter and System x V1
(616) 451-2500 System i IT Simplification: Linux
Technical V5R4
(616) 451-2571 - Fax iSeries System Administrator for OS/400
V5R3
(616) 260-4746 - Cell

If you can read this, thank a teacher....and since it's in English, thank a soldier.

Mike Cunningham <mcunning@xxxxxxx>
Sent by: midrange-l-bounces@xxxxxxxxxxxx
01/07/2008 11:09 AM
Please respond to
Midrange Systems Technical Discussion <midrange-l@xxxxxxxxxxxx>

To
"'Midrange Systems Technical Discussion'" <midrange-l@xxxxxxxxxxxx> cc

Subject
RE: NIC in FAILED status

It is the standard NIC in a 520. Will run at 1GB and shows that it is connected at 1GB right now. Connected to a Cisco switch (not sure of the model). Cisco admin says his side was set to auto also (although I have read that switch ports should not be set to auto, our switch admins insist it is "best practice" to set auto on all ports). 520's come standard with a pair of copper nics and one was working when the second one was not.

-----Original Message-----
From: midrange-l-bounces@xxxxxxxxxxxx
[mailto:midrange-l-bounces@xxxxxxxxxxxx] On Behalf Of Chris Bipes
Sent: Monday, January 07, 2008 11:02 AM
To: Midrange Systems Technical Discussion
Subject: RE: NIC in FAILED status

What is the NIC on the iSeries? Can it run at 1GB? What is the switch that is was plugged in to? Was it set at 1GB FDX and the iSeries NIC is older and can only run at 100MB FDX or less?

Chris Bipes
Director of I.S.
CrossCheck, Inc.

-----Original Message-----
From: midrange-l-bounces@xxxxxxxxxxxx
[mailto:midrange-l-bounces@xxxxxxxxxxxx] On Behalf Of Mike Cunningham
Sent: Monday, January 07, 2008 7:58 AM
To: 'Midrange Systems Technical Discussion'
Subject: RE: NIC in FAILED status

Both are set to *AUTO
--
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing list To post a message email: MIDRANGE-L@xxxxxxxxxxxx To subscribe, unsubscribe, or change list options,
visit: http://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxx Before posting, please take a moment to review the archives at http://archive.midrange.com/midrange-l.

--
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing list To post a message email: MIDRANGE-L@xxxxxxxxxxxx To subscribe, unsubscribe, or change list options,
visit: http://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxx Before posting, please take a moment to review the archives at http://archive.midrange.com/midrange-l.