RE: Bad NIC Card ? -- MIDRANGE-L

Thanks Scott and that was my mistake on the Keep Alive deal. I found
info IBM, got interrupted, did the prompted the CHGTCPA command, got
interrupted AGAIN and when I came back hit the wrong parameter and the
rest is history.

This is ONE PC doing this; never dropping active iSeries connections,
etc. Really strange but I'll check the cabling, etc.

And thanks so much for all of the other information; it was very
educational!

Chuck

-----Original Message-----
From: midrange-l-bounces@xxxxxxxxxxxx
[mailto:midrange-l-bounces@xxxxxxxxxxxx] On Behalf Of Scott Klement
Sent: Thursday, July 09, 2009 1:56 PM
To: Midrange Systems Technical Discussion
Subject: Re: Bad NIC Card ?

Hi Chuck,

First of all, I don't see what this has to do with keepalives. If you
were getting "reason code 3" then I could see keepalives being an issue,

but you're not!! You're getting "reason code 2". Reason code 2 means
that the R2 threshold has been exceeded.

If you don't understand what is meant by "R2" threshold, then let me
explain:

TCP tries to send a packet to a remote system. It gets no response.
What should it do? Well, the TCP protocol is supposed to be "reliable",

so if anything goes wrong it's supposed to detect it an retry. So it
retries... but how long should it retry before it gives up?

There's an R1 threshold, which by default is set to 3. That means that
it will try 3 times, and if it fails after 3 times, it'll ask the system

if maybe the route to the host has changed. It'll try to find a new
route to the host, and then go back to retrying.

Then there's the R2 threshold. That's the "time to give up" threshold.
By default, after it has tried 16 times (which means it's tried
requesting a new route via the R1 threshold 5 times as well -- assuming
default values) it'll finally give up entirely and assume this TCP
connection is not gonna work.

You can view your R1 and R2 threshold values by prompting the CHGTCPA
command. You can change them with that command as well.

So... you'll get a TCP2617 with reason code 2 when a host stops
responding. This could mean that it's powered off. It also could be a
host you're trying to connect to that's been disconnected from the
network or powered off.

For example, maybe you have a RMTOUTQ that always sends to a printer or
another computer. If you unplugged that printer, you might get TCP2617
on a regular interval as the system keeps trying to send spooled files
to that printer.

All the message means is that the system isn't responding. It doesn't
tell you WHY it's not responding. It could be a bad network card, it
could be a bad cable, bad switch, it could simply be disconnected or
turned off... And it isn't necessarily on your i -- the problem could

be on the other side, or anywhere in between.

I'd watch the message and see if it happens consistently on a particular

IP address or port. If so, you may have a problem on that IP address or

port.

However, if this is a sporadic message with no real pattern that seems
to show up occasionally, it could simply be that users PCs are locking
up and they're powering them off. Or that someone powered off a printer

to fix a jam, or something like that. In which case, these are normal
messages that can be ignored.

In V5R2 and earlier, OS/400 never reported these errors. They were
added in V5R3. If you prefer not to see them, you can turn them off
with CHGTCPA LOGPCLERR(*NO)

Good luck

Chuck Lewis wrote:

Hi Folks,

We keep getting this message on qsysopr every couple of minutes:

TCP/IP connection to remote system 172.16.10.79 closed, reason code 2.

Additional Message Information

Message ID . . . . . . : TCP2617 Severity . . . . . . . : 20

Message type . . . . . : Diagnostic

Date sent . . . . . . : 07/09/09 Time sent . . . . . . :
10:43:31

Message . . . . : TCP/IP connection to remote system 172.16.10.79
closed,
reason code 2.

Cause . . . . . : The TCP/IP connection to remote system

172.16.10.79

has
been closed. The connection was closed for reason code 2. Full
connection
details for the closed connection include:

- local IP address is 172.16.10.11

- local port is 20663

- remote IP address is 172.16.10.79

- remote port is 3766

Reason codes and their meanings follow:

1 = TCP connection closed due to expiration of 10 minute Close

Wait

timer.
2 = TCP connection closed due to R2 retry threshold being run.

More...

This IP address is a specific PC.

I've Googled the message and got info from IBM (first hit !) about

using

CHGTCPA to change the Keep Alive setting and it's at 120 so I changed

to 360 and still get it.

I'm beginning to wonder if there is a NIC card going bad in this PC ?

doesn't drop the 5250 emulation sessions or anything thing.

Any idea/comments ?

Thanks !

Chuck