Re: TCP goes down -- MIDRANGE-L

Hello, all:

On several AS/400s at several community colleges over the past fewyears, I found that it really helps to specify the CMNRCYLMT parameteron CHGLINETH command for your ethernet line (e.g. ETHLINE) as follows:

Recovery limits:Count limit . . .: 99

       Time interval ..: 5

Also, add a reply list entry (WRKRPYLE, then F6=Add) for CPA58EE,specifying a reply of "R" for Retry, and a compare value of *NONE. Thisway, if, after 99 retries on ETHLINE, it still fails, the system issuesCPA58EE, and this reply list entry tells the system to "retry" and thisstarts a new 99 tries every 5 minutes. :-)

With the combination of these two, these AS/400s stayed up, available tostudents on the internet, and recovered, even after events likemaintenance on the campus networks, etc., or power outages where theAS/400 was UPS-protected, and so it stayed up, but all of the campusrouters, etc. were not UPS protected ... :-o


Hope this helps someone out there ... ;-)

Cheers,

Mark S. Waterbury

> Joe Lewis wrote:

Sheri -
On my 170 when the Ethernet cable is disturbed, all TCP/IP traffic goesdown. If there's no message in QSYSOPR, you can try the old ENDTCP thenSTRTCP. Over the years, there has been a couple time where that didn'twork and had to do an IPL. Needless to say, we are very careful aroundthe ethernet cables and switch. Hope that helps.
- Joe
------------------------------

message: 4
date: Tue, 09 Jan 2007 11:47:34 -0500
from: Rich Loeber <rich@xxxxxxxxx>
subject: Re: TCP goes down

Sheri,

I've seen this happen when the cable connection is disturbed or the router (or
hub) it is connected to looses power for a very short time.  The first place
I've learned to look is at the QSYSOPR message queue.  If there is an error
there, then a simple 'R' to retry gets you back on track (assuming that the
physical problem has been corrected).

Rich Loeber
Kisco Information Systems
http://www.kisco.com
  ----------------------------------------------------------------------