XI lost connection to J2EE

[Former Member]

Hi,

a few time per month the XI system seems to loose connection to J2EE server, the RFC connections AI_DIRECTORY_JCOSERVER, AI_RUNTIME_JCOSERVER, etc. stops working and of course all messages fail.

Logs in dev_jcontrol showing errors like:

[Thr 3596] Thu Sep 15 18:34:24 2005

[Thr 3596] ***LOG Q0I=> NiPRead: recv (10054: WSAECONNRESET: Connection reset by peer) [ninti.c 785]

[Thr 3596] *** ERROR => MsINiRead: NiBufReceive failed (NIECONN_BROKEN) [msxxi.c 2488]

[Thr 3596] *** ERROR => MsIReadFromHdl: NiRead (rc=NIECONN_BROKEN) [msxxi.c 1652]

[Thr 3596] ***LOG Q0I=> NiPConnect2: SiPeekPendConn (10065: WSAEHOSTUNREACH: No route to host) [nixxi_r.cpp 8605]

[Thr 3596] *** ERROR => MsIAttachEx: NiBufConnect to sapserver4/3601 failed (rc=NIECONN_REFUSED) [msxxi.c 633]

[Thr 3596] *** WARNING => Can't reconnect to message server (sapserver4/3601) [rc = -100]-> reconnect [jcntrms.c 296]

[Thr 6612] ***LOG Q0I=> NiPRead: recv (10054: WSAECONNRESET: Connection reset by peer) [ninti.c 785]

[Thr 3596] Thu Sep 15 18:34:29 2005

[Thr 3596] *** ERROR => MsIAttachEx: NiBufConnect to sapserver4/3601 failed (rc=NIECONN_REFUSED) [msxxi.c 633]

[Thr 3596] *** WARNING => Can't reconnect to message server (sapserver4/3601) [rc = -100]-> reconnect [jcntrms.c 296]

Anyone else have similar problems? Suggestions?

Cheers,

Mattias

[Former Member]

Hi Mattias,

We see this issue often in our XI systems. To date, there is only one solution: restart the JCos. We have not been able to determine why these JCos lose their regirstration in the gateway, but they do.

The next time it happens, goto tcode SMGW and Goto -> Logged on clients. You should see your above mentioned RFC's logged onto the gateway. If you don't then open the VA on the XI server to stop then start each affected JCo. Once it's restarted, check the gateway again to see if they've registered. Your system should be good to go without restarting the entire J2EE engine (well, at least until it happens again)

We have asked SAP Support about this issue, and they answered to us what I just answered to you. Not the best answer, but it is workable in the meantime.

Good luck!

-Tim

[Former Member]

Dears,

The same problem occurs today with my PI enviroment. The rfc AI_DIRECTORY_JCOSERVER stop after restart PI works fine again.

My PI is kernel 700 level 185.