10054 : WSAECONNRESET: Connection reset by peer

[former_member201258]

Dear All,

ECC 6.0 Linux(RHEL5)/ Maxdb7.6

We are frequently getting error "Connection reset by peer" in the SAPGUI. Even Application server and database is running fine still we are facing this after some time and whenever we are working from customer premises.

We have gone through the SAP Notes 155147, 26086,500235 etc and try to find out whether there is any network related problem.

We have run following commands

At application server

niping -S 3299 -s -I t=0 (Application server)

At presentation server ( USER desktop)

niping -S 3299 -c -h 192.168.9.39 -L 3600 -D 1000

We go the error message from the above niping command which we have run at the presentataion server.Shows that it may be due to network problems. But network team is saying that the network is working fine.

As suggested by SAP We have set the parameter rdisp/keepalive to 300 still we are facing the same error

The session opened at the customer desktop after some idle time the session is not responding for any t-code . When we clieck on the GUI after some idle time its getting disconnected.

If we again connect to the system we can see the last session active at the server side. Could not understand what is the issue? Is there any one face this issue?

Thanks

Tom

[former_member201258]

Hello All,

Our network team had ajusted some firewall setting .Now the disconnection error is happening very rarely.

Thanks

Tom

[sayalib25]

Hi Tom,

Could you please tell which Firewall settings were adjusted?

Regards,

Sayali

[Former Member]

Hi TOM,

This is purely network problem, just ping the server from one of the desktop in client site and find the latency it was showing there(time to send and recieve in ms).

If this value is high obviously there will be a connection break happening at client GUI.

run the ping command and post the values here. One thing the latency should be less than 300ms for good connectivity.

Regards,

Raju.

[former_member201258]

Hi,

The output of ping to the sap system from the client machine

Reply from 192.168.9.39: bytes=32 time=39ms TTL=57

Reply from 192.168.9.39: bytes=32 time=36ms TTL=57

Reply from 192.168.9.39: bytes=32 time=39ms TTL=57

Reply from 192.168.9.39: bytes=32 time=41ms TTL=57

Reply from 192.168.9.39: bytes=32 time=40ms TTL=57

Reply from 192.168.9.39: bytes=32 time=39ms TTL=57

Reply from 192.168.9.39: bytes=32 time=1049ms TTL=57

Regards

Tom

[Former Member]

Hi,

check the value

Reply from 192.168.9.39: bytes=32 time=1049ms TTL=57

why was there sudden jump in latency from 32ms to 1049, this can be one of reason for connection drops, sudden congestion in network which was shown in by sudden high latency, you need to investigate this along with your network team.

Regards...

[former_member201258]

Dear All,

The problem was related with the firewall. We have rectified it now the disconnection is very rarely happening.

Thanks

Tom

[former_member201258]

Dear All,

The disconnection problem is still happening rarely . We are tring to further analyse the root cause of the issue. Pls see the error log the user is getting, If any one know abt this issue pls help us to solve this as this escalated to the top level.

ed Nov 12 20:05:36 2008

**LOG Q0I=> NiIRead: recv (104: Connection reset by peer) [nixxi.cpp 4424]

• ERROR => NiIRead: SiRecv failed for hdl 17 / sock 14

(SI_ECONN_BROKEN/104; I4; ST; 192.168.50.8:2776) [nixxi.cpp 4424]

etwork error of client T47, NiBufReceive (-6: NIECONN_BROKEN), dp_tm_status=3

lient address of T47 is 192.168.50.8(192.168.50.8)

**LOG Q04=> DpRTmPrep, NiBufReceive (4463 MANGESH 47 comp-196 ) [dpxxdisp.c 11618]

M-T47, U4463, 400 MANGESH, comp-196, 19:41:05, M0, W0, , 2/0

hu Nov 13 09:27:01 2008

oftCancel request for T52 U4633 M0 received from REMOTE_TERMINAL

hu Nov 13 09:39:35 2008

oftCancel request for T38 U4634 M0 received from REMOTE_TERMINAL

hu Nov 13 09:43:35 2008

• WARNING => DpRqServiceQueue: timeout of HIGH PRIO msg, return DP_CANT_HANDLE_REQ

hu Nov 13 09:43:59 2008

• WARNING => DpRqServiceQueue: timeout of HIGH PRIO msg, return DP_CANT_HANDLE_REQ

hu Nov 13 09:44:31 2008

• WARNING => DpRqServiceQueue: timeout of HIGH PRIO msg, return DP_CANT_HANDLE_REQ

hu Nov 13 09:44:32 2008

• WARNING => DpRqServiceQueue: timeout of HIGH PRIO msg, return DP_CANT_HANDLE_REQ

hu Nov 13 09:44:33 2008

• WARNING => DpRqServiceQueue: timeout of HIGH PRIO msg, return DP_CANT_HANDLE_REQ

• WARNING => DpRqServiceQueue: timeout of HIGH PRIO msg, return DP_CANT_HANDLE_REQ

hu Nov 13 09:44:34 2008

• WARNING => DpRqServiceQueue: timeout of HIGH PRIO msg, return DP_CANT_HANDLE_REQ

• WARNING => DpRqServiceQueue: timeout of HIGH PRIO msg, return DP_CANT_HANDLE_REQ

hu Nov 13 09:44:35 2008

• WARNING => DpRqServiceQueue: timeout of HIGH PRIO msg, return DP_CANT_HANDLE_REQ

• WARNING => DpRqServiceQueue: timeout of HIGH PRIO msg, return DP_CANT_HANDLE_REQ

hu Nov 13 09:44:36 2008

• WARNING => DpRqServiceQueue: timeout of HIGH PRIO msg, return DP_CANT_HANDLE_REQ

• WARNING => DpRqServiceQueue: timeout of HIGH PRIO msg, return DP_CANT_HANDLE_REQ

hu Nov 13 09:44:37 2008

• WARNING => DpRqServiceQueue: timeout of HIGH PRIO msg, return DP_CANT_HANDLE_REQ

• WARNING => DpRqServiceQueue: timeout of HIGH PRIO msg, return DP_CANT_HANDLE_REQ

Thanks

Tom

[Former Member]

HI,

Is the end users and the servers all belong to same network or they connect through any third party network, if that was case ask them to establish a direct link.

if everything sits in same network ask to incresae the bandwidth.

Finally this was issue need to be dealt by network team, from the sys log we can say the reason was due to network problem or congestion. Though the network guys dont agree this but the issue need to be dealt by them, this was from my direct experience.

Regards,

Raju.

[former_member201258]

Hi all,

we are getting the above error msg.

Thanks

Tom

[former_member201258]

NiIWrite: hdl 4 sent data (wrt=265,pac=1,MESG_IO)

mode 0 sndok = FALSE

TM_LAYER TmWrite (0) <<

Mon Nov 03 15:07:46 2008

SiSelNNext: sock 1272 selected (revt=r--)

NiSelIListInsert: add hdl 4 [0] to sel-list (0) of set0

NiSelISelectInt: 1 handles selected (0 buffered)

TM_LAYER TmRead (0) >>

NiHsLGetServName: found port number 3205 in cache

NiIGetServName: port 3205 = servicename 'sapdp05'

***LOG Q0I=> NiIRead: P=192.168.9.39:3205; L=0.0.0.0:2916: recv (10054: WSAECONNRESET: Connection reset by peer) [nixxi.cpp 4854]

• ERROR => NiIRead: SiRecv failed for hdl 4/sock 1272

(SI_ECONN_BROKEN/10054; I4; ST; P=192.168.9.39:3205; L=0.0.0.0:2916) [nixxi.cpp 4854]

• ERROR => TmRead: NiRead (rc=-6) [dpxxtm.c 1129]

Mon Nov 03 15:08:11 2008

NiICreateHandle: hdl 17 state NI_INITIAL_CON

NiIInitSocket: set default settings for new hdl 17/sock 1124 (I4; DG)

NiDgHdlCreate: state of hdl 17 NI_DG_UNCONNECTED

NiGetLoopbackAddr: return 127.0.0.1

NiDgHdlSendTo: hdl 17 sent 1 of 1 bytes to 127.0.0.1:65000

NiICloseHandle: shutdown and close hdl 17/sock 1124

NiICloseHandle: shutdown and close hdl 9/sock 1300

SiSelNRemoveAll: removed all sockets

NiICloseHandle: shutdown and close hdl 4/sock 1272

NiIExit: free nitab 02BEA018

TmThrExit: free local resources

TmExit: free local resources