RE: Forte On Clustered Systems

Hi,
This is a very interesting and useful concept. We are using DEC Alpha
TruCluster and may borrow your concept to improve our fail-over procedure of
our environment manager. Many Thanks.
Peter Sham.
-----Original Message-----
From: Shell, Charlie [SMTP:[email protected]]
Sent: Friday, March 19, 1999 2:59 AM
To: '[email protected]'
Subject: Forte On Clustered Systems
To the Group,
I've noticed a couple of folks have asked about running Forte on a
clustered
system. As far as I know Forte does not specifically support
clustered
environments, however since Oct 1998 we have been running Forte
3.0J1 on a 2
node Sun Enterprise 4000 cluster running Solaris 2.6 with the Sun
High-Availability v 2.1 product.
In our configuration only one node "owns" the logical host (the
shared disk
storage array). The other node operates in standby mode,
essentially doing
nothing. Since the Env Mgr repository has dependencies upon the
physical
machine on which it is run, (which are stored in the repository) you
cannot
use the repository from a different machine. To overcome this we
have
devised an approach, brute force in nature that allows us to support
the
clustered environment. In the interest of time, there are many
details I
won't go into here but the conceptual view of how we've configured
Forte
support a failover (failover being defined as the transfer of the
logical
host to the standby system by the HA software) is as follows:
The standard forte installation directories for a central server
have been
defined on the shared disk array. In addition to that we have
created 2
template Env Mgr repositories (one for each node). Each repository
contains
the server node entry for the specific host machine and all of the
model
node definitions we use. The template repositories had to be
created by
starting the Env Mgr once on each node and then making a copy of the
repository for use as a template.
When an failover occurrs, HA transfers the logical host (shared disk
arrays)
to whichever node is currently in standby mode. At an appropriate
point in
the reconfiguration process it runs our private script which copies
the
correct Env Mgr template repository into /sysdata/envrepos. Then it
runs
escript to load and install all of the applications that have been
distributed into the environment during the normal course of
day-to-day
production operations. Then the it starts the Env Mgr and any
required
ftexec servers. The downside to this is the EnvMgr cannot rebind to
previous client sessions, and since we have re-installed all of the
applications, it may download client images to the clients when they
first
re-connect to the environment.
Charlie Shell
Information Services, Bell Atlantic Mobile
2000 Corporate Drive, Orangeburg NY 10962
E-Mail: [email protected]
Tel: (914) 365-7927 Cell: (908) 770 -0096
Pager: 1-800-SKY-8888 pin 6300432
To unsubscribe, email '[email protected]' with
'unsubscribe forte-users' as the body of the message.
Searchable thread archive
<URL:http://pinehurst.sageit.com/listarchive/>
To unsubscribe, email '[email protected]' with
'unsubscribe forte-users' as the body of the message.
Searchable thread archive <URL:http://pinehurst.sageit.com/listarchive/>

Hello Charlie,
We are about to go live with a high availablility application on an
identical platform to the one you describe. Are you using Oracle 7.3.4
Parallel Server? If so I would like to ask you a few more questions.
David Campbell,
CSC Australia
To: [email protected]
cc: (bcc: David Campbell/AUST/CSC)
From: [email protected]
Date: 03/18/99 06:58:55 PM GMT
Subject: Forte On Clustered Systems
To the Group,
I've noticed a couple of folks have asked about running Forte on a
clustered
system. As far as I know Forte does not specifically support clustered
environments, however since Oct 1998 we have been running Forte 3.0J1 on a
2
node Sun Enterprise 4000 cluster running Solaris 2.6 with the Sun
High-Availability v 2.1 product.
In our configuration only one node "owns" the logical host (the shared disk
storage array). The other node operates in standby mode, essentially doing
nothing. Since the Env Mgr repository has dependencies upon the physical
machine on which it is run, (which are stored in the repository) you cannot
use the repository from a different machine. To overcome this we have
devised an approach, brute force in nature that allows us to support the
clustered environment. In the interest of time, there are many details I
won't go into here but the conceptual view of how we've configured Forte
support a failover (failover being defined as the transfer of the logical
host to the standby system by the HA software) is as follows:
The standard forte installation directories for a central server have been
defined on the shared disk array. In addition to that we have created 2
template Env Mgr repositories (one for each node). Each repository
contains
the server node entry for the specific host machine and all of the model
node definitions we use. The template repositories had to be created by
starting the Env Mgr once on each node and then making a copy of the
repository for use as a template.
When an failover occurrs, HA transfers the logical host (shared disk
arrays)
to whichever node is currently in standby mode. At an appropriate point in
the reconfiguration process it runs our private script which copies the
correct Env Mgr template repository into /sysdata/envrepos. Then it runs
escript to load and install all of the applications that have been
distributed into the environment during the normal course of day-to-day
production operations. Then the it starts the Env Mgr and any required
ftexec servers. The downside to this is the EnvMgr cannot rebind to
previous client sessions, and since we have re-installed all of the
applications, it may download client images to the clients when they first
re-connect to the environment.
Charlie Shell
Information Services, Bell Atlantic Mobile
2000 Corporate Drive, Orangeburg NY 10962
E-Mail: [email protected]
Tel: (914) 365-7927 Cell: (908) 770 -0096
Pager: 1-800-SKY-8888 pin 6300432
To unsubscribe, email '[email protected]' with
'unsubscribe forte-users' as the body of the message.
Searchable thread archive <URL:http://pinehurst.sageit.com/listarchive/>
To unsubscribe, email '[email protected]' with
'unsubscribe forte-users' as the body of the message.
Searchable thread archive <URL:http://pinehurst.sageit.com/listarchive/>

Similar Messages

Is there a difference in behaviour between clustered systems vs. standalone

OpenVMS, MessageQ (could not find another place to ask :().
Suppose 2 nodes, same DMQ bus.
Node 1: Application X starts up, binds to group Y (not yet existing and so creates connection)
Node 2: Application X starts up, binds to group Y.
If node1 and node2 are in one VMS cluster, Node2::X would now be unable to do anyting since group Y is now owned by Node1::X and so node2::X will get into a wait state, to become active only if, for whatever reason, Node1::X gives up.
But what happens if node1 and node2 are stand-alone systems, only connected using a DECNet (or TCPIP) connetcion:
1: Node2::X waits until Node1::X stops and group Y is released (as if the systems were clustered)
2: Node2::X will startup and access group Y so does not enter a wait state. If so, can this be forced to a wait (like (1))?
(You may assume that the DMQ configuration is correct)

Most people either use one app or the other, and have very limited knowledge of the one they don't use. But from what I can find from a quick look round the web, they seem to be the same.
Review: Photoshop Elements 13 gets Photomerge and content-aware fill | Macworld

BRCONNECT Error in clustered system

Hi,
Getting below mentioned error, when i schedule Check Database or Update Optimizer through DB13. Landscape is ECC6.0 on HP-UX and Oracle 10.2.0.2.0.
Job started
Step 001 started (program RSDBAJOB, variant &0000000000017, user ID RAKESH)
Execute logical command BRCONNECT On host dbSID
Parameters: -u / -jid CHECK20100308172918 -c -f check
BR0801I BRCONNECT 7.00 (44)
BR0252W Function remove() failed for '/oracle/SID/102_64/dbs/sapSID.ora' at location BrInitOraCopy-7
BR0253W errno 13: Permission denied
BR0805I Start of BRCONNECT processing: cecswqyx.chk 2010-03-08 17.29.19
BR0252E Function fopen() failed for '/oracle/SID/sapcheck/cecswqyx.chk' at location main-11
BR0253E errno 13: Permission denied
BR0121E Processing of log file /oracle/SID/sapcheck/cecswqyx.chk failed
BR0806I End of BRCONNECT processing: cecswqyx.chk2010-03-08 17.29.21
BR0280I BRCONNECT time stamp: 2010-03-08 17.29.22
BR0804I BRCONNECT terminated with errors
External program terminated with exit code 3
BRCONNECT returned error status E
Job finished
Regards,
Ranjith
Edited by: Ranjith Kumar D on Mar 9, 2010 5:09 AM

Hello Ranjith,
it seems things are getting clearer.
Actually I was asking about physical hosts. But if I understand correctly, database and central instance are not on the same physical host; correct?
So on the database host all Oracle related file systems are mounted, but from central instance host files like /oracle/SID/102_64/dbs/sapSID.ora are not available, and that's what the error message is telling you. Correct?
And the three dialog instances are located outside of the cluster, I presume?
So you have got several options available. Unfortunately I can't help with details, but can only give you a rough idea:
Make Oracle file systems available on the central instance host, by means of an NFS mount. This has to be integrated with your cluster, and should only be done by someone very experienced in HP-UX clusters.
Create a way how to execute SAP commands on a remote host, like gateway instance or remote shell calls. Read details in the installation guides or SAP help manual, under terms like distributed installation. Probably you will also have to consider the integration in your cluster here.
Maybe one of these options is already (partially) installed, but you will have to double check the details.
hope this helps

Cloning Clustered system

We're trying to setup a staging environment which mirrors our production cluster. The production cluster contains two systems, the first is running the admin server and managed server 1, the second system is running managed server 2.
After cloning both of these systems and changing the multicast address yesterday and starting the new (staging) admin server, the 2nd Managed Server in the production cluster detached itself from the prod admin server and attached to the staging admin server.
After a bit of Googling I saw Unicast mentioned in Weblogic 10 docs, but we are using WL8. Based on the unicast documentation though, it sounds as if it may not be possible to run both on the same network segment, or at least not without renaming everything. I started to go that route and renamed the Cluster and Admin server, but now when starting the newly renamed Admin server I get an error about boot.properties "Boot identity not valid" so I figure there's something I'm still missing. Any ideas how to do this (assuming you can have multiple clusters on the same subnet with WL8)?
TIA!

Hi, There.
It sounds like you have more of a configuration issue, then a clustering one.
When you copy your files from prod, you need to do a few things:
1. remove running-managed-servers.xml from your config directory. This is what's causing your problems with detaching.
2. use a different multicast ip
3. obviously, reset your database urls and passwords. you can put them in 'plain text' into the config.xml file. weblogic will rename them
4. setup appropriate ListenAddress and Machine properties in config.xml
This will get you going!
Cheers,
-Adrian

How to distrubute the load to another Dialog instance in clusterred system

Hi All,
I have cluster environment(ABAP Stack) where load has been put on CI which lead to performance issue. Is there possibility to change the load to another DI dynamically or if not please tell me how can i reduce the load on the system.
Cheers
V.

By using logon groups... simply consigure SMLG.
http://help.sap.com/saphelp_nw70/helpdata/en/c4/3a64b4505211d189550000e829fbbd/content.htm
Regards
Juan

Forte System Management

Hi, I'm writing this note under a colleague's e-mail address.
Do you have to build a forte application with system management in mind
from the start, if so what features.
Failure to do so appears to leave Forte as a 'black box'.
Also what are the potential risks of using/not using system management
features.
Regards
Steve Gordon
To unsubscribe, email '[email protected]' with
'unsubscribe forte-users' as the body of the message.
Searchable thread archive <URL:http://pinehurst.sageit.com/listarchive/>

We have connected two environments. Environment A has
an application with an Environment Visible SO. This SO
gets triggered at a certain moment to create a certain
anchored object and publish this object with the Name-
Service. Applications in environment A and B can then
access the anchored object.We have a number of connected environments each of which publishes several
anchored objects in the forte name service.
When environment B is offline and the partition containing
the EVSO in environment A has just been started, the
first attempt to register the anchored object with the Name-
Service fails. The error says:
USER ERROR: Failed to create object (/sub/tst) in a remote
environment AF5C47B0-61F5-11D2-B9FC-627494AEAA77.
Class: qqsp_SystemResourceException
The UUID in this error belongs to environment B.This reads to me as if Env A is trying to create the object in environment B
which doesn't sound like a good idea.
The second attempt succeeds, even though environment B
is still offline.
Can anyone explain what the cause of this error is? Is
this normal behaviour?Hmm, we don't see this behaviour because we make sure that the creation of
the anchored object is done in the environment that it lives in. Is there a
reason environment A is trying to create environment's B anchored object?
Cheers
David McPaul
Lumley Technology
This message has been Content Filtered by MailMarshal.
---------------------------------------------------------------------

Session problem on clustered sun fire with SP6.1

Hi All,
I am having a problem with a webapplication on a clustered system.
In the following I describe the current system setup.
I will then describe the setup of the web application.
Following this, I explain the trouble I'm having with sharing sessions in this environment.
Finally, I pose the key question.
SYSTEM SETUP:
Our group has a clustered server setup. I believe the two machines are Sun Fire systems running Solaris 9. Let me know if you need more information concerning the HW.
To manage the cluster, we use Sun Cluster software. Let me know if you need to know the version number. I do not have it handy. We just recently setup the system (in the past 6 months).
Also, we have the newest Sun Web Server installed 6.1. We used the default installation. The only thing we changed was the dynamicreloadinterval variable from "-1" to "60".
WEB APPLICATION:
We have a fairly simple MVC web application setup, which works well on a single-server system. Essentially, the web application (1) asks for a certain input via a webform, (2) takes this input, and uses it to get information from a database, (3) puts this information into a httpsession attribute, (4) redirects the response to a JSP. The JSP gets the httpsession attribute (populated in step 3) and prints it out.
So, HTML with form posts to Servlet, Servlet gets data from database and populates a session attribute, JSP gets that session attribute and prints it out.
Between each transition from HTML, Servlet and JSP, the currently processing server may switch. I.e. the HTML may be served, for instance, by ClusterServer1, the Servlet may be handled, for instance, by ClusterServer2, and the JSP may be served at random by either.
THE TROUBLE I'M HAVING
Sometimes, my JSP is able to find the session attribute and print out whatever I put into the session during the Servlet step. However, sometimes, the JSP will print out "null" (the session attribute isn't available to the JSP).
For the HTML and the JSP, I have the page print out what server it is being served from. Thus, I can tell that the HTML is served from, for instance, ClusterServer1 while JSP is served from, for instance, ClusterServer2. Sometimes, the same server serves both.
KEY QUESTIONS:
Why would this be happening?
How can I have a session stick to the user and be shared across both servers?
What other information would you need to provide an answer concerning this issue?
I appreciate your efforts very much!
Matthias Edrich
dailysun

Hi All,
Hi Elving,
I read through the documentation and have the following questions:
(1) It seems that I can share sessions amongst both servers if I
configure the web server to store sessions in a persistant manner such as in a file or in a database. Is this correct?
(2) To enable this persistent storage of a session, I would need to change the Session Manager used. Is this correct?
(3) If yes, I have the choice between the following managers. Please correct me if I have misunderstood the options.
- PersistentManager: Instead of securing session information
in memory, this manager saves session information within
a file on the server in a directory, which I specify within
sun-web.xml
- IWSSessionManager: With this manager, I can store sessions
in a defined database or file on the server.
- MMapSessionManager: This manager also stores sessions in
a file on the server
(4) What is the difference between PersistentManager and
MMapSessionManager if my web server is running in
single-process mode?
(5) If MMapSessionManager is a file-based manager, where do I
specify to what directory the related file is stored to as I
do in PersistentManager?
(6) What are the advantages and disadvantages of
Persistent/IWS/MMap managers?
(7) In looking at SessionManagers, am I even barking up
the right tree? It seems like these would help me share
sessions across clustered servers.
(8) Finally, I guess my plan of action would be the following:
-> Identify what session manager to use
-> Include a sun-web.xml file in /WEB-INF containing
the manager-specific info
-> Reload my web application
-> done...
Is this correct?
Elving, I appreciate your help. Thanks!
dailysun

MS SQL 2005 clustered database instance.

Short Text
MS SQL cluster database instance - ONLY!
Long Text
Hi,
If possible could you please confirm that our planned landscape
migration is supported by SAP.
At present our PRD SAP system is a standalone instance that consists of
the database & central instance running on the same "HOST".
Current SAP PRD version:
ERP 2004 ECC 5.0 32 bit
MS SQL 2000 32 bit
both running on same host with OS Windows 2003 32 bit
We plan to upgrade/migrate our PRD SAP system to new 64 bit hardware
AND to move to a "Distributed" landscape installation i.e. database
instance & central instance installed on different HOSTS. Also, the
database instance WILL be clustered but the central instance will NOT
be clustered. There are also multiple NON-SAP MS SQL databases that
will run in the same cluster as the PRD SAP database instance.
SAP PRD version after migration:
SAP ERP 2004 ECC 5.0 64bit (running on X86_64 AMD64 hardware) on HOST'X'
MS SQL 2005 64 bit (running on IA64 hardware) on clustered HOST'Y'
Both running on Windows 2003 64 bit
My questions are:
1) Does SAP support a clustered MS SQL 2005 database instance that is
running on a clustered HOST along with NON-SAP MS SQL databases?
2)During the migration our PRD SAP MS SQL database will be upgraded
from MS SQL 2000 32 bit to MS SQL 2005 64 bit IA64, is this also
supported in a clustered istallation (database instance ONLY)?
3) The central PRD SAP instance will be running on x86_64 AMD64
hardware & the database instance will be running on 64 bit IA64
hardware. Is this configuration supported?
Thanking you inadvance.

Hi Greg,
Have you figured out the answer? We are trying do the exactly the same thing -- Move our SAP WAS 6.40 clustered central instance to a 64-bit hardware/SQL 2005/Clustered system. And this system has several non-SAP database on there. The biggest concern we have is whether SAP support this architecture. I posted a message here last year and several experts like Clas Hortien, N. P. C, in this form suggest NOT go to this architecture because SAP will not support it. One example, is that SAP/SQL requires server collation setting to SQL_Latin1_General_CP850_BIN2 and other databases may work on different collations.
Migrate SAP SQL Server database to co-exist with others
Anyway, if you or anyone in this form have some more information please share here.
Thanks,
Yujun Ran

SUN Fire V65 - System Status LED - non-critical condition

Hi guys
We have a Checkpoint Cluster which is made up of 2 Sun Fire V65 servers.
The front panel of the servers are showing the System Status LED lights as Blinking Amber , which sun documenation has listed as
"Blinking amber: Indicates the system is in a non-critical condition" ref: http://dlc.sun.com/pdf/817-2024-12/817-2024-12.pdf
We have resetead the hardware, cleared ,logs, even change HW etc. The light goes away for some time and then returns.
Can anyone advise on what to do?
Sun has the following listed in their Troubleshooting pdf i ref before
"Non-Critical Condition
A non-critical condition is indicated with a blinking amber status LED and signifies
that at least one of the following conditions is present:
¡ Temperature, voltage, or fan non-critical threshold crossing.
¡ Chassis intrusion.
¡ Satellite controller sends a non-critical state, via the Set Fault Indication
command, to the BMC.
¡ A Set Fault Indication command from the system BIOS. The BIOS may use the Set
Fault Indication command to indicate additional, non-critical status such as
system memory or CPU configuration changes"
Is there any Unix / Linux / SPLAT logs I can look at or debugging we can look at.
Thanks in advance

Your initial description mentions that you have swapped various pieces of hardware around. I presume that means you used current parts instead of new-never-used components. Are both of the clustered systems showing the same symptoms?
I can only think of a couple of additional things to try.
Firstly, those are old boxes. V65's went EOL in January 2005, so yours are at least that old if not older. They are x86 architecture. Have you replaced the CR2032 CMOS battery that's on the systemboard? Have you additionally checked for voltage fluctuations in the data center? A marginal power supply can "mostly work" but it can be effected by an AC mains issue.
If that doesn't change anything, then you may need to contact Checkpoint. They may be able to point you toward some sort of heartbeat function that is tripping a fault level when it flip-flops between the two systems.
That's a software investigation and would be outside the scope of this forum.

Re: Forte & Lotus Notes

Has anyone developed an interface between Forte and email systems,
specifically Lotus Notes? We want to, among other things, send email
messages to named users from within a Forte application. We are running
AIX on our Forte servers. Our client machines are mainly Windows 95.
Any help or hints to point us in the right direction would be appreciated.
Also, if you have any information on directly accessing Lotus Notes from
Forte without going through email it would be appreciated.We wrote a OLE Automation server which raps the Notes API. We partition
that
service object on the Notes Mail server and send mail. We have the method
to
get the address book and use it to send the mail. If all you want to do is
send mail
then this is a great solution.
Hope this helps
Kamran Amin

I am presently investigating implementing a Forte' Shared Library using
Lotus Notes Hitest C API (ver. 4.5).
You can download Hitest API Toolkit from the Lotus Notes Application
Development Tools's WEB Site:
(http://www2.lotus.com/developers/tools.nsf).
The Lotus Notes HiTest C API is an alternative higher level C interface
to the standard Lotus Notes C API. Program development is significantly
faster and requires a fraction of the API code needed with other APIs.
The Hitest API are available for all the platforms supported by Forte'
(except for Apple Macintosh).
Hope this helps.
Giuseppe Sorce
CSI-PIEMONTE
Cso Unione Sovietica 216
10134 Torino ITALY
tel: +39 11 4618515
fax: +39 11 4618212
e-mail: [email protected]
David Wilbur wrote:
>
To: [email protected]
cc:
Subject: Forte & Lotus Notes
Has anyone developed an interface between Forte and email systems,
specifically Lotus Notes? We want to, among other things, send email
messages to named users from within a Forte application. We are running
AIX on our Forte servers. Our client machines are mainly Windows 95.
Any help or hints to point us in the right direction would be appreciated.
Also, if you have any information on directly accessing Lotus Notes from
Forte without going through email it would be appreciated.
David Wilbur email: [email protected]
University of Windsor phone: 519-253-3000 ext. 2779
Windsor, Ontario Canada

Export for system copy fails - Error 9 (Bad file number)

Hello all,
I am attempting to perform the export phase for a hetergenous system copy on our BI system (BI 7.0 - NW2004s). This system runs on HP-UX 11.11 using Oracle 10.2.0.2 and will be migrated to Windows Server 2003 and MS SQL Server 2005, once the export is complete.
While running sapinst to export the data from the source system, there are several input dialog screens that are skipped, which means I am not able to enter all the paramters used for the export. After seeing this behavior from sapinst, I reviewed the sapinst.log and sapinst_dev.log files and find the warning/error "System call failed. Error 9 (Bad file number)". I have included a portion of these files below.
I have reviewed notes 546006 and 583861, but these seem to refer to an Oracle error ORA-27121, but I am not receiving any Oracle errors in sapinst.log or the Oracle alert log. I have looked for possible permission or ownership problems with any of the SAP or Oracle files, but have not found any that were incorrect. Also, these errors do not specify which file sapinst is having problems with, so I do not have a specific file or area to check.
If anyone has seen something similar to this and has any suggestions, I would greatly appreciate any help.
sapinst.log details:
INFO 2009-02-14 15:15:51.67
Execute step collect of component |NW_Export|ind|ind|ind|ind|0|0|NW_GetSidFromProfiles|ind|ind|ind|ind|1|0|NW_readProfileDir|ind|ind
|ind|ind|0|0
INFO 2009-02-14 15:15:52.737
Account bwpadm already exists.
WARNING 2009-02-14 15:15:52.747
System call failed. Error 9 (Bad file number) in execution of system call 'getgrent' with parameter (), line (292) in file (syuxcuse
r.cpp).
INFO 2009-02-14 15:15:52.757
Account 0 already exists.
WARNING 2009-02-14 15:15:52.777
System call failed. Error 9 (Bad file number) in execution of system call 'getgrent' with parameter (), line (292) in file (syuxcuse
r.cpp).
INFO 2009-02-14 15:15:52.777
Account 200 already exists.
WARNING 2009-02-14 15:15:52.827
System call failed. Error 9 (Bad file number) in execution of system call 'getgrent' with parameter (), line (292) in file (syuxcuse
r.cpp).
INFO 2009-02-14 15:16:04.937
Creating file /tmp/sapinst_instdir/NW04S/LM/COPY/ORA/EXP/CENTRAL/AS/EXP/ypcat.log.
INFO 2009-02-14 15:16:05.007
Output of ypcat passwd.byname is written to the logfile ypcat.log.
sapinst_dev.log details:
TRACE      2009-02-14 15:15:52.737
2009-02-14 15:15:52.743 NW._getSAPMount() done: /sapmnt
INFO       2009-02-14 15:15:52.737 [syuxcaccmg.cpp:227]
           bool CSyAccountMgtImpl::isExisting(ISyAccountMgt::eAccountType accountType, iastring sName)
Account bwpadm already exists.
TRACE      2009-02-14 15:15:52.747 [syuxcuser.cpp:1943]
           CSyUserImpl_getOsInfos(ILogBook *m_logBook, iastring sName, iastring sID, tSyUserInfo& msUserinfo)
getspnam(bwpadm) returned NULL.
WARNING    2009-02-14 15:15:52.747 [syxxsyshlp.cpp:78]
           syslib::printOSError(const iastring &, int, DWORD, const iastring &, CMessage::eLogMessage)
System call failed. Error 9 (Bad file number) in execution of system call 'getgrent' with parameter (), line (292) in file (syuxcuse
r.cpp).
Regards,
Rusty Robbins

Hello Peter,
Thanks for the reply. I did confirm the points you made, and have the details below.
1. check filesystem, if you have enough space, especially /tmp
All filesystems have available space, and /tmp has 310 MB available, before the start of sapinst.
2. check again alert log, trace log for Oracle,
I could not find any errors in the alert_SID.log or any unusual trace files.
3. How did you run sapinst? Are you using "root" account.
I am running sapinst as user root and I am following all the recomendations in the system copy guide (at least as far as I can tell. ;p )
We have four BI systems in this landscape, and the other three have migrated successfully. The procedure I am using, is the same as for the other 3 systems and uses the same SAP media directory that was used on the other systems.
One other item that may be part of my problem here, is that this system is the production system, and was originally installed as a high-availability system, using HP-UX Service Guard for clustering. Currently, this system does not have fail-over enabled, but of course the cluster is still intact. One required piece of the Service Guard enviornment is the use of a virtual hostname. This seems to resolve successfully by sapinst, as I see the correct (virtual) hostname referenced several times. While this is a clustered system (for fail-over purposes), the CI and DB are installed on the same host. So it is not a distributed system, and if failover had been initiated, it would fail both the CI and DB instance, which is the same and only SAP instance.
This did raise one question in my mind, however. I am attempting to extract this using the following menu path within sapinst.
SAP NetWeaver 7.0 Support Release 3 --> Software Life-Cycle Options --> System Copy --> Oracle --> Source System Export --> Central System --> Based on AS ABAP and AS Java
Since this does use Service Guard, but is not a distributed system, should I be using a different path in the sapinst tree, for the export? I have only performed a migration using the above path, so I am not familar with the detailf of the other types, such as High-Availabitly or Distributed-System.
Regards,
Rusty

Conectivity between existent SAP Systems and migrated system to MSCS Cluste

Hi
I have an existing SAP NW PI 7.0 (Abap + Java on Windows 2003 - SQLServer 2005) and plan to migrate to a cluster with new hardware (MSCS).
I have to prepare for the installation by creating a homogeneous copy of the database but I'm a little bit confused.
-Can I use backup/restore method or only can use import/export with R3load tools? The target system is windows and sql server same versions.
-If I can use backup/restore... what is the rigth sequence to the installation? Can I use sapinst with mixed high availability options and System copy options?
-I need to use in the target system the same hostname and IP address of the actual system, so in the Follow-Up Activities we do not change the configuration between the actual system (SAP NW PI) and other existent systems in the landscape... but I do not know what is the location in the cluster to this hostname and IP:
*The central instance will be in the local disks in the node 1. Is the node 1 the rigth option to use the hostname and IP of the actual system?
*The SAP Cluster group has a virtual hostname and IP, it has message and enqueu services. Is the sap cluster group he rigth option to use the hostname and IP of the actual system?
How is the communication between the other systemas and the new clustered system?
Thanks for your help

Can I use backup/restore method or only can use import/export with R3load tools? The target system is windows and sql server same versions.
You should follow the SAP guidlines in performing a homogeneous system copy.
I need to use in the target system the same hostname and IP address of the actual system, so in the Follow-Up Activities we do not change the configuration between the actual system (SAP NW PI) and other existent systems in the landscape... but I do not know what is the location in the cluster to this hostname and IP
This is not possible. If you follow the SAP instructions for installing NW 7.0 in MSCS you will create two cluster groups: DBGroup and SAPGroup. You will therefore have two virtual IP addresses, one for SAP cluster group and one for DB cluster group. If you retain the current IP address/hostname for the virtual SAPGroup, you will need to follow the SAP guidelines for performing a SAP system copy to find where to change the database hostname/IP addresses to point to your new DBGroup.
The central instance will be in the local disks in the node 1. Is the node 1 the rigth option to use the hostname and IP of the actual system?
If you are meaning the ASCS then this should be installed on the disks that are part of the SAPGroup in the cluster.
If you mean the dialog instance (or the Central Instance of a Java stack) then this can be on a local disk for node 1, and not in the cluster.
The SAP Cluster group has a virtual hostname and IP, it has message and enqueu services. Is the sap cluster group he rigth option to use the hostname and IP of the actual system
Yes, this will be the message server host for ASCS.

New Install of MSCS for an Upgraded Abap System

We are currently on NW04S ECC 6.0 on WAS 700 in a MSCS, Oracle 10g. Our problem is that we are moving to new hardware and our current system uses the old MSCS (which is an upgraded system originally installed at 4.7 640 (perhaps earlier (4.6c)), which to me means that the Central Instance resides in the MSCS in the SAP Group. Our plan is to install a clustered system and do a DB restore or copy over to the new hardware. I have done the install and realize now that the ABAP MSCS install for NW04s ABAP creates the cluster a little differently in that it looks like the Java (Portal) cluster. Only the SCS and enqueue are in the MSCS and the CI and all Dis are outside of the MSCS. This creates a problem for me in that the SCS is using the 00 system number, that on our old system the CI had, so our gui is now pointing to what would be the wrong system number, as it looks like the gui must point to to the CI system number for connection to the message server. So, I guess this means a reinstall for me and switching the system numbers. I also dont know exactly how high availability would be created in this new setup unless you put an application switch in front of your CI and Dis, like you would with the Portal, to direct gui requests. With the old MSCS the cluster would know if the CI failed and move it to the other node, but since it resides outside of the cluster, you have to have some type of external director right?

Hello Jack,
even you have upgraded from 6.40 to 7.00 you will need to convert to new HA structure according the note 1011190 to avoid the problems with next upgrade. In 700 the CI and DI are not the SPOF (see also
<a href="https://www.sdn.sap.comhttp://www.sdn.sap.comhttp://www.sdn.sap.com/irj/sdn/go/portal/prtroot/docs/library/uuid/e0b8cd93-f1e9-2910-c186-86bfff3dac63">here</a>) and you can configure logon groups via the message server.
In case of hardware change better to take the system copy guide and install new HA architecture - it will save you manual conversion (SAP note 1011190).
best regards

Mounting a local directory in Forte 4.0 (Sun One?)

Okay, I'm relatively new to Forte and am probably missing some detail that is quite stupid, but for some reason I can't mount a new local directory in Forte (on win98 system). I select "Mount", choose "local directory", but then no matter what directory I select the "Next" and "Finish" buttons remained grayed out so I can't get anywhere! It's not a big crisis, but a little annoying just the same. Can anyone out there tell me what I'm missing? Thanks.

nice idea (opening a file)
but, File|New, File|Open, etc.,
all cause Forte to exit on my Win2K box
(yes I have uninstalled/installed/ad nausem for Forte)
any more ideas?
[email protected]

Benefits of RAC in large systems environment

Hi All,
One query regarding RAC
In a large systems environment where multiple CPUs and memory are dynamically available to be deployed for a database server, what are the benefits of a RAC implementation? Obviously no single point of failure is eliminated but how else is performance improved?
Thanks
Edited by: user10394804 on Aug 10, 2009 1:17 AM

A clustered system either DB or the system(OS) or any other form of clustering is designed for high availabilty only and not for performance improvement.
Please keep in mind that you need to check the performance of 2 clusterd and non-clusterd systems with the same level hardware and OS resources.
If your application is performing at level 'X' with 4 CPU and 4 GB RAM on a non-clusterd machine then its performance cannot be matched with the 2 machines with 4 CPU each and 4 GB RAM each, because thats the double the resources than available in non-clusterd machince.
Also just keep one more thing in mind that with the same resources now distributed in clustered nodes (ie 2 nodes with 2 CPU and 2 GB RAM v/s 1 node with 4 CPU and 4 GB ram), the single node will alawys perform better because there will be no inter-connect overhead that is there in RAC systems.
Amar.

RE: Forte On Clustered Systems

Similar Messages

Maybe you are looking for