Operating system call writev failed (error no. 32 )

[Former Member]

Hi!

We are running a CRM 5.0 system under AIX 5.3. The system is a HA scenario (database and ASCS instance on host A, Dialog instance on host B and C)

Kernel patch 113

Basis supportpack 10 (n2004s)

We have the following error message in our syslog, Operating system call writev failed (error no. 32 ) ,we don't know whats causes it. Every day we about 40 entry on each dialog instance (diffrent users). The error occurs on all types of processes, DIA, BTC, UP2, UP1 etc.

Detail

Operating system call writev failed (error no. 32 )

Technical details

File Offset RecFm System log type Grp N variable message data

83 409680 m Error (Function,Module,Row) Q0 I writev 32 NiIWritwritev nixxi.cp3803

No documentation for syslog message Q0 I exists

Additional specifications for error number 32

Name for errno number EPIPE

No documentation available for error EPIPE

Further details for this message type

Module nam Line Error text Caller.... Reason/cal

nixxi.cp ; 3803 writev 32 NiIWrit writev

Anyone who has seen this error before?

[markus_doehr2]

writev() is a function, that writes to a file or to a socket.

The most common case for this kind of error is,that the terminals (user SAPGUIs) are no more reachable when the server tries to send something to them.

Markus

[Former Member]

Thanks for your answer Markus.

So that means that I have some kind of communication problem between appserver and the database server? Because most of the errors is not communication from a SAPGUI, they are batch jobs.

Is there a way to trace this error on the network?

Anton

[Former Member]

Hi,

check sap note Note 34116 - Syslog: "Operating System call xxxx failed" for trace

regards,

kaushal

[Former Member]

In dev trace I found:

M ***LOG Q0I=> NiIWrite: writev (32: Broken pipe) nixxi.cpp 3803

M *** ERROR => NiIWrite: SiSendV failed for hdl 2 / sock 15

(SI_ECONN_BROKEN/32; I4; ST; 10.10.10.10:3260) nixxi.cpp 3803 IP of ASCS server - CHANGED

M *** ERROR => EncFragSend():: received error -6 encomi.c 1062

E *** ERROR => EnsaCliDoRequest: DoRequest failed (rc=-1): 504 bytes on con. 0 in

E ENSA_DoRequest (): Reconnect

E *** ERROR => ENSA_Connect(): AlHARegisterComponent() failed with error 254 [ens

M EnsaCliRegisterCom: found layer EncCli

M EnsaCliRegisterCom: found layer EncTh

Anton

[Former Member]

Hi,

(SI_ECONN_BROKEN/32; I4; ST; 10.10.10.10:3260) nixxi.cpp 3803 IP of ASCS server - CHANGED

check the ip address of ASCS Server

regards,

kaushal

[Former Member]

hmm... I changed the IP to 10.10.10.10.. did not want to publish our PRD IP on SDN... thats why I marked it as changed.

Anton

[Former Member]

Hi,

ohh ok,

can you also provide the trace file for gateway (dev_rd)

tcode SMGW -> goto ->trace->gateway->display file

I'm not sure but upgrding kernel might solve this problem

regards,

kaushal

[markus_doehr2]

You changed the IP on the server to hide it here on SDN?

Markus

[Former Member]

No, I did NOT change the IP on the server only when I pasted the log message at SDN.

Anton

[Former Member]

i had the same issue, plz let me know how to solve it

[Former Member]

hey Anton,

I had the same issue & my bckgrd jobs failed. Plz let me know how solve the issue.

***LOG Q0I=> NiIWrite: writev (32: Broken pipe) [nixxi.cpp 3948]

• ERROR => NiIWrite: SiSendV failed for hdl 4 / sock 16

(SI_ECONN_BROKEN/32; I4; ST; XXXXXXXXX:3200) [nixxi.cpp 3948]

• ERROR => EncFragSend():: received error -6 [encomi.c 1062]

• ERROR => EnsaCliDoRequest: DoRequest failed (rc=-1): 472 bytes on con. 0 in layer 0 (EncCli) (1/1) [ensacli.c 406]

ENSA_DoRequest (): Reconnect

• ERROR => ENSA_Connect(): AlHARegisterComponent() failed with error 254 [ensaclnt.c 1099]

EnsaCliRegisterCom: found layer EncCli

EnsaCliRegisterCom: found layer EncTh

thanks

[markus_doehr2]

Did this happen during a failover in a cluster? Did you configure operation modes (SE63)?

Markus

[Former Member]

sorry not sure - what u mean by failover in cluster.

yes configured operation modes - sm63

[markus_doehr2]

An ASCS instance is installed in a cluster (hardware/high availibility). If you have the same error message as originally posted that would be the case.

So my question was: Did a failover to another node happen at that time?

And the other idea were the operation modes - did they maybe change at the time that error occured?

Markus

[Former Member]

No,

& system is ECC 6.0, ABAP & JAVA.

Does it has anything to do with enqueue.

[markus_doehr2]

Yes - it seems that the enqueue server tried to send some data somewhere and this failed (system overload, queue full..)

So your log message is not the same as the original poster

Markus

[Former Member]

This could be problem with you kernel files. Try download and apply a higher/lower version of the kernel and check.