Remote login in SM51 Fails / ECC 6.0 / HP-UX / Oracle 10

[Former Member]

Hi All,

We had a cluster, 4 physical servers / 14 SAP Instances / HP-UX - Oracle 10G / ECC 6 EHP 3.

Basically we have a general RFC problem in the system. You can reproduce it simply in transaction sm51: In most of the instances whenever you try to change to an instance that has the same instance number than the one you are logged on at this moment, you will come back to this instance, e.g.:

you are logged on at host1_SID_01, start sm51 and try to change to the instances:

host2_SID_01,

host3_SID_01 or

host4_SID_01

(or you try to display the system log or other). You will see (System - Status) that you don't change to these instances but will stay on host1_SID_01.

We found the SAP Note 662895 - Remote login using SM51 Fails. In this note they refer to the error: "Transaction termination 14 025".

Now we have this error (we upgraded from SAP 46C with another error in sm21).

The problem is that we really donu2019t understand exactly what they want to do in the SAP Note 662895:

Summary

Symptom

Remote login in SM51 by double-clicking R/3 instances fails

Solution

You can use transaction SM51 to log in to another application server using RFC, without entering a password. Server "A"

Server "A" Server "B"

<--

---

Remote logon via RFC

---

<--

---

GUI

sm51

-

---

Double-click on the server B entry --->

If host name "B" was entered on server "A" as the gw/alternative_hostnames profile parameter, the local SAP gateway from server "A" is used when you log on to server "B". This results in the syslog message "Transaction termination 14 025".

The same error occurs if you try to open a new external mode on server "A".

-

---

Who is the origin server? Host A? Host B? In fact sometimes you can go from A to B and vice versa. So changing the gw/alternative_hostnames parameter doesn't appear to be a solution.

I would appreciate some tips about this Note / Problem.

Best Regards,

Erick Ilarraza

[Former Member]

Hi, in SM59 the "Internal Connections" test works fine ?

Regrads.

[Former Member]

Hi Sergo, Thanks a lot for your reply.

We tested all the internal connections (here you can see al the SAP instances <host>_SID_01,... etc) all this work fine. You get a Warning:

Connection Type Internal Connection

On HP multiprocessor machines, the specified time values may be incorrect

The only one RFC connection with a problem is this:

RFC Destination BACK

Connection Type I Internal Connection Description

Description 1 Back to calling program

Description 2 Can only be used in programs called from remote sources

Description 3 (SAP standard entry)

The error is this:

Logon Connection Error

Error Detail Program no longer started via RFC. No return possible.

Last weekend we upgraded our SAP systems from 46C to ECC 6.0 EH3 however the SM51 Problem come from 46C. It was not caused for the Upgrade.

Best Regards,

Erick Ilarraza

[former_member220071]

Hi Erick ,

Ensure that the computer with the gateway process can access the specified computer by entering /etc/ping <host name> or ping <host name> through OS level.

Abhijeet

[Former Member]

Hi, you have 14 instances, this situation are from all instances or not?

For example login direct to APP server 7, and try to 10, after 6 to 10, and over combinations, to check the server you not need to see in "system" --> "status",

in the right corner you can see the SAP APPL server host name.

Regards.

[Former Member]

Hello, thanks a lot for your replyes.

Abhijeet Siras, the command /etc/ping <host> works prefectly.

Sergo,

Exactly we have 3 host each one with 4 dialog instances. 1 host with 2 dialog instances and one host with the central instance.

Not all instances have this problem, only the following combination:

Origin Destination

host4_SID_01 - > host1_SID_01

host2_SID_03 - > host3_SID_03

host1_SID_01 --> host3_SID_01

host1_SID_02 --> host3_SID_02

host1_SID_03 --> host3_SID_03

host1_SID_04 --> host3_SID_04

host2_SID_01 --> host3_SID_01

host2_SID_02 --> host3_SID_02

host2_SID_03 --> host3_SID_03

host2_SID_04 --> host1_SID_04

host2_SID_04 --> host3_SID_04

host3_SID_04 --> host1_SID_04

host4_SID_01 --> host1_SID_01

host4_SID_02 --> host1_SID_02

host4_SID_02 --> host2_SID_02

host4_SID_02 --> host3_SID_02

The problem is between *_NUMBER to *_NUMBER!

For example from Dialog Instance 3 running on host 2 to Dialog Instance 3 running on host 3

For example host4_SID_01 means the Dialog Instance 1 running on host 4.

Best Regards,

Erick Ilarraza

[Former Member]

Ok, then can you check for example on host1_SID_02 in SM59 the "internal connection"

to host3_SID_02

Open the connection and check the "Technical settings" target host and client are correct ?

It's also may be the Gateway problem ....

Regards.

[former_member220071]

Hi Erick ,

Have you checked the hosts file in OS level.

The file hosts must contain the name of the calling host .

Make sure each host containing remaining host name in hosts file.

Abhijeet

[Former Member]

Not sure about this, but I just stumbled across SAP note 29934.

Maybe two instances are using common parameter settings that actually should be different?

E.g. are instances host4_SID_01 and host1_SID_01 using the same instance profile, or different?

[Former Member]

Hi All, Thanks a lot for your replyes.

Sergo:

I tested the "internal connection" with the SM59: All works perfect, Connection Test OK.

Abhijeet:

I searched the /etc/hosts file on the Origin and Destination server: the file contains all the names

of the systems. host1, host2, host3 and host4.

ping is Ok.

rlogin is Ok.

Joe:

Each SAP Instance is started with its own START profile:

Example START_D02_host1 for the SAP Instance number 2 running on the server 1.

Inside the file START_D02_host1 file you can find the Instance profile <SID>_D02_host1.

What could be the problem?

We are looking again with the SAP note 662895 - remote login using SM51 fails, particulary the parameter gw/alternative_hostnames. In our case the parameter is set with all the instances:

Example:

For host1_D02 (Dialog Instance 2 running on host1) the parameter gw/alternative_hostnames is set with host1, host2 and host3 and C.I. Probably this is not correct and we should only set this parameter to the alternatives names of the current localhost (host1 in this case).

On the other hand, since this is a High Availability Cluster, in case of a failover for example the host1 could act as a backup of the host2 (if host2 fails host1 takes the cluster package of host2 to do the failover), maybe this is the reason for this set.

Best Regards,

Erick Ilarraza

[Former Member]

Erick,

I think you might be on the right track here.

We too have a high availability cluster. But unlike you, we gave different instance numbers to all the instances.

And your problems are confirming my opinion that we made a wise decision here.

And regarding backup hosts:

Are you aware that instances host1_SID_01 and host2_SID_01 will never be able to run on the same physical host, neither host1 nor host2, just because instance numbers are the same?

regards