Reset connection by peer

[ganimede_dignan]

Hi,

form 2 remote plant (first is 128K frame-relay with 4 fron-end and second is a cisco-vpn 1mb adsl with 10 front-end) we have many times "Reset connection by peer". It's not so nice.

Network manager say that all it's OK. Have you got any advice ? test ? to solve ?

Regards.

Ganimede Dignan.

[markus_doehr2 · Accepted solution]

This message means, that the peer (so your SAP server) sent an RST packet to the client because it was not reachable. Latency can be an issue here - especially if you run other applications over the same line (Windows network shares are know to take all they get).

How does the "network manager" check the connection? A working "ping" does not necessarily mean, that other connection oriented protocols, such as TCP which is used for SAP connections, don't have any problem...

Markus

[ganimede_dignan · Accepted solution]

Hi,

all this test ends without error...

niping -c -H <R3 IP> -B 20 -L 100

niping -c -H <R3 IP> -B 100000

niping -c -H <R3 IP> -B 10000 -D 100 -L 10

niping -c -H <R3 IP> -B 100000

niping -c -H <R3 IP> -B 20 -L 100

niping -c -H <R3 IP> -B 1000000 -L 100

What can I check, and what can I ask to our network manager ?

Regards.

[markus_doehr2 · Accepted solution]

The problem here is (as with all "ping" programs) - they create a connection, transfer data and close the connection. SAPGUI must have a permanent connection to the system (in both directions).

What you can do find out of that problem is on one or the other side is to use a tool like netcat (http://netcat.sourceforge.net/) to run on both sides a permanent connection test and see where it fails.

Markus

[ganimede_dignan · Accepted solution]

hi,

>The problem here is (as with all "ping" programs) - they create a connection, transfer data and close the >connection. SAPGUI must have a permanent connection to the system (in both directions).

OK,

>What you can do find out of that problem is on one or the other side is to use a tool like netcat

>(http://netcat.sourceforge.net/) to run on both sides a permanent connection test and see where it fails.

I open i SAPgui on R/3 system on fron-end PC, I run netcat on front-end as "nc.exe -v <ip r3 server> 32<SYSTEM NUMBER>"

After several minutes SAPgio ends with "reset connection by peer" and netcat still alive.

Regards.

[markus_doehr2 · Accepted solution]

> I open i SAPgui on R/3 system on fron-end PC, I run netcat on front-end as "nc.exe -v <ip r3 server> 32<SYSTEM NUMBER>"

>

> After several minutes SAPgio ends with "reset connection by peer" and netcat still alive.

This is really strange...

Do you see something in SM21 on the server when this happens? Also check the corresponding workprocess trace, who resetted the connection (also visible in SM21)

Markus

[ganimede_dignan · Accepted solution]

On SM21 there it on only line about it:

20:34:40 DP Q04 Connection to user 12309 (<USERNAME> ), terminal 133 (<PCNAME> ) lost

In details:

Details Page 2 Line 26 System Log: Local Analysis of mnsun201 1

Time Ty. Nr Cl. User Tcod MNo Text Date : 04.01.08

20:34:40 DP Q04 Connection to user 12309 (<USERNAME> ), terminal 133 (<PCNAME> ) lost

Details

Recording at local and central time........................ 04.01.2008 20:34:40

Task................ 05055 . 00 DP Dispatcher

User................

Client..............

Terminal............

Session............. 0

Transaction code....

Program name........

Problem class....... K SAP Basis problem

Development class... STSK

Further details for this message type

Module name......... dpxxdisp

Line................ 8106

Error text.......... 12309<USERNAME> 133 SAOFF003

Caller.............. DpRTmPr

Reason/called....... NiBufRe

Documentation for system log message Q0 4 :

The SAP Dispatcher (part of the application server) has lost the

connection to a terminal process. For example, this happens when the

terminal program terminates without correctly logging off the

application server. More detailed information about the error

context is not available here.

Technical details

File................ 000030

Position............ 0000200160

Entry type.......... m ( Error (function,module,line) )

Message ID.......... Q0 4

Variable parts...... 12309<USERNAME> 133 <PCNAME> DpRTmPrNiBufRedpxxdisp8106

[markus_doehr2 · Accepted solution]

Do you see anything in dev_disp?

Markus

[ganimede_dignan · Accepted solution]

at the end of dev_disp file:

Network error of client T133, NiBufReceive (-6: NIECONN_BROKEN), dp_tm_status=3

Client address of T133 is <IP_FRONTEND>(<FRONTEND_NAME>)

***LOG Q04=> DpRTmPrep, NiBufReceive (12309<USER> 133 <FRONTEND_NAME> ) [dpxxdisp.c 8106]

RM-T133, U12309, 020 <USER>, <FRONTEND_NAME>, 20:14:29, M0, W1, SM04, 2/1

Fri Jan 4 20:51:20 2008

• ERROR => DpCheckTerminals: NiCheck2(rc=-23) failed for tid: 104 -> disconnecting [dpxxdisp.c 8734]

[markus_doehr2 · Accepted solution]

So... it seems, that the connection is breaking from the server side because the server can't connect any more to the client. You have a network issue in that direction... are there any firewalls between those? If yes, check, that they don't tear down the TCP connections when they are inactive for longer time.

Markus

[ganimede_dignan · Accepted solution]

Hi,

on fron-end I run:

nc.exe -l -p 5555

on R/3 server I run:

nc -v <IP_FRONTEND> 5555

after severla minutes nc on R/3 server ends.

[markus_doehr2 · Accepted solution]

That confirms the problem.

You have a network issue from the serer to the clients. Check the active network components and the line itself.

Markus