Minimize upgrade downtime in HA environment

Hi upgrade gurus,
I have 3 instances (1 CI , let say PR1_00, 2 DI, let say PR1_01 and PR1_02, DB on seperate box, let say PR1_DB). HA setting is used for both PR1_00 and PR1_DB. (Microsoft Cluster). some are running on top of
Now I need to patch my system from SPS14 to SPS19.
To minimize the downtime, I would like to first apply that patch on the SLAVE of my CI.
1.During the patch I expect the DB is synchronized from my PR1_00_PRIMARY to PR1_00_SLAVE is on going but not vice versus, while I am applying the patch on PR1_00_SLAVE. This will not cause any stopage of my business.
2.After the patch on PR1_00_SLAVE is completed. I switch the PR1_00_SLAVE from passive mode to active mode. I will apply patch on PR1_00_PRIMARY during this time. I expect same synch. is applied as step 1.
3. restart the DIs respectively.
Is this feasible ? will step 1 and step 2 work ?
Kindly comments.
thanks & Best Regards,
Aaron

Hi Aaron,
Let me phrase what i understand from you is you have CI and DB on HA environment. And CI and DB are mirrored. So you want to patch slave first so that master will be available to users. Then once slave patching is completed you will make slave as master and then do patching of earlier master.
But i don't think that this a good approach. As what about the difference in data because when master is in production mode there will be change in data as users will use that..And when you will use slave as master after upgrade then there will also be change in data...
How will you synchronize both ?
Thanks
Sunny

Similar Messages

SAP EPR6.0 Upgrade (Downtime).

Dear Sir/Madam,
Currently we are doing SAP ERP6.0 Upgrade, Can you please guide me what are the steps we need to adapt in order to Minimize System downtime.
Regards.
Hubert.

Hi Hubert,
Regarding the upgrade times, there are a lot of factors like the table
contents, DB size, tp and R3trans versions and so on that can affect the
performance. As to my experience, it depends mostly on database performance.
Please refer to the following SAP notes:
838725 - Oracle 10.2: Dictionary and system statistics
830576 - Parameter recommendations for Oracle 10g
1171650 - Automated Oracle DB parameter check
983548 - Long runtimes during SAP Upgrade using Oracle 10
558197 - upgrade hangs in PARCONV_UPG, XPRAS_UPG, SHADOW_IMPORT_UPG2
Additional information:
1. if the location of the datafiles (upgrade directory) is a local
   directory or a mounted remote directory. For a remote directory
   the connection speed to it is important;
2. Even if the upgrade directory is local, network layer is used and
   thus network configuration is important. A typical difference
   between production and test systems can be that production is often
   high availability and thus can use e.g. virtual hostnames or other
   methods. Maybe the network setup is then different to non-production
   systems;
3. Also hardware issues, e.g. if the database or the upgrade directory
   are on fast disks and if caches are used are important;
4. the database speed (parameters, memory, disks, ....);
5. if the database is on the host where SAPup and R3trans run or on a
   different host might influence the speed but even a local database
   access uses network layer and thus network configuration for the
   database host is also an issue.
Best Regards
Julia

Minimize upgrade time

Dear all,
One of my client has a database around 6TB, and they are planning to upgrade the database version from 10.2.0.4 to 11.2.2 . OS is Solaris 10. How do I minimize the downtime of the database upgrade ? . In this scenario, how to proceed with huge database and scenarios like banking,telco where they can't afford for such a huge downtime ?
Please guide
Kai

KaiS
Handle:     KaiS
Email:     [email protected]
Status Level:     Newbie
Registered:     Nov 20, 2007
Total Posts:     1,232
Total Questions:     513 (389 unresolved)
Name     Kai Schikum
Location     Back in NL
And still trying to outsource as much of his work as possible. Tell me, who are the clients who hire you?
Sybrand Bakker
Senior Oracle DBA

11.5.10.2 to R12 - Reducing Upgrade Downtime

Hi,
We'r in the process of upgrading 11.5.10.2 instance to R12. Please anybody who has already done the upgrade, can tell us the way by which we can reduce the Upgrade DOWNTIME to a considerable extent (apart from what Oracle Recommends). Looking forward for early response as this is very URGENT.
Thanx in Advance
Regards
Mahesh

There is no other way to reduce the downtime other than suggested by oracle. May be you can club patches like applying rdbms patches in sequence, applying help patches once you release the instance and merging all other product patches
regards
Pravin

Changing the IP address of CI before the upgrade downtime

Hello,
I have posted a similar question in another forum, but have specific DB2 question in this forum.
Our CI server is in windows and the database is mainframe DB2 z/OS. We are upgrading from 46C to ERP Unicode (CUUC)
We have just finished the upgrade uptime 'Upgrade Phase Group: Shadow Import' - which took twice as previous benchmark. This is because the network response between the CI windows server and the mainframe database is very slow.
Now the plan is to move the CI server to the same subnet as the database server BEFORE we start the 'Upgrade Phase Group: Downtime phases I: Switch tables and Kernel'. Plan below
1. we have reached the START OF PHASE MODPROF_TRANS and we exit (our schedule dowtime is on weekend)
2. Once the scheduled downtime is reached, we stop SAP, stop the DB2 system
3. Then move the CI windows server - still the same physical and logical server name, just different IP address
4. Start DB2, Start SAP in the CI windows servef
5. We then resume the SAPUP in the MODPROF_TRANS, proceed with the upgrade
6. When the non-Unicode upgrade is complete, we use LINUX server for the Unicode conversion to use the DB2 fastload
- while UC is happening, we move back the CI windows server to original place and original IP address (several reasons why we have to do it.
7. So when UC is complete, we install the 7.0 Unicode CI in the windows server (using the previous IP address)
Several question
- do we need to change anything in the DB2 side (e.g. ICLI)
- do you see any problem with this plan?
Thanks,

Hello Terris,
I cannot make any statement concerning the different Upgrade phases and your upgrade plan itself.
But as far as ICLI is concerned you do not have to re-configure anything.
You do not even need to stop the ICLI server. As soon as the CI is restarted (using the new IP address),
it will re-connect to the ICLI server. The ICLI server does not do any validation / caching based on IP addresses.
best regards, Harald

Database upgrade in Dataguard Broker Environment

Hi All,
OS: Wndows 32 bit
DB: 11.2.0.1
We have an upgrade activity due, in this, we have a windows Primary Database and it has 2 Physical Standby databases. There is no concept of Near DR and Far DR. Both the databases are getting sync'd directly from the Primary database.We are also using Dataguard Broker. Following is the skeletol POA for the same:
1. Install an out of place DATABASE Software (11.2.0.3 ) on both the Standby databases and also the Primary database.
2.Disable the configuration of the database broker on all the 3 servers
3.Copy the files init/spfile, tnsnames.ora, listener.ora, sqlnet.ora and the password file from the old oracle home to the new oracle home location on all the 3 servers.
4.Take a clean shutdown of the Standby databases only and then stop the services and delete the services as well.
5.Get into the new oracle home and create the oracle services for both the standby databases
6.Once, the services are created, mount both the standby databases and enable their recovery.
7. Now, we will connect to the Priamry database server and start the upgrade here. I will be using the manuall method of upgrade for the database here which all includes all the pre and post tasks.
8. Once the upgrade is done succesfully on the primary databases, the changes will be automatically replicated on its 2 Standby databases as well.
9. Post this, we will enable the broker configuration and check for the syncronziation by doing some switch's.
10. Lastly we will remove the old oracle home's.
Any suggestions?
Regards,
SPhinx

After the upgrade the home directory will change and all I need to do is change the home directory path in the extract/replicate prm file correct ?
this what I will do :
1. stop GG services on both source and target
2. upgrade the db.
3. change the path of the db in the prm files as required.
4. start the extract and replicate process on source and target respectively.

BW Upgrade 3.0 to 7.0 ( Data Migration for Prod)

Hello Experts,
We are doing BW Upgrade from 3.0 to 7.0, I need your suggestions.
We are going to upgrade BW Development, Quality and Production landscape to 7.0 and once technical upgrade is done we will go ahead with Function upgrade.
Now point is, once we done with all developments in Production environment then how we will take Historical data?
We are planning to connect Old BW system (3.X) as Source system to new BW 7.0 and then we are going to pull data.
But as we need to move data from Standard data sources like LO we need downtime for Production environment. Our challenge is to minimize this downtime for production environment.
Does any body having idea about this? I am aware of one IBM toll TDMF for data migration which migrate data from one data base to one data base with hardly 2 hours of downtime , but again for this licensing cost and etc is involved.
Please share your experiences and suggestions.
Thanks in advance.
Regards,
Pankaj Naik

You would not need to da a data migration if you are upgrading the existing 3.0 system with the data.
Edited by: Henry Janz on Jun 9, 2011 5:09 AM

Hardware Migration

Dear All,
We are in progressing of Hardware Migration for All prod servers.
Below is the ECC system detail .
CI and DB are separate nodes and 4 App servers were installed.
ECC EHP5 6.0 , AIX , DB2
Active and Active cluster setup.
Now can you suggest how we can proceed a new system build on new hardware in with cluster environment ( CI and DB separate nodes with Active and active cluster setup )
thanks

Hi John,
As you mentioned export/import is one of the options which when optimized can minimize the downtime on Production environment.
Second option which I could suggest is to perform offline backup restore and then perform in-place upgrade using db2se upgrade .
Found some links to guide you on this.
http://support.esri.com/en/knowledgebase/techarticles/detail/40798
http://www.dangerousdba.com/upgrading-from-db2-luw-9-7-to-10-1/
Hope this helps.
Regards,
Deepak Kori

Minimizing Downtime during Upgrade

Hi All,
We are in process of upgrade from R/3 4.7 EE 2.00 to ECC 6.0 SR3.
The details of legacy and new landscape is as follows :
Legacy Landscape
OS       Solaris 9
DB       Oracle 9.2.04
SAP     R/3 4.7 EE 2.00
New Landscape
OS       Solaris 10
DB       Oracle 10g
SAP     ECC 6.0 SR3
We have taken the offline backup of Live PRD and restored it on to Stand by server. We have also upgrade oracle 9i to 10g, which has taken around 20 hours.
My question is, we want to minimize the downtime by directly installing Oracle 10g with R/3 4.7 EE and the take the offline/online backup of live PRD system and restored back to oracle 10g enviroment. By doing this we can reduce the downtime by aprrox 15 hrs.
Is this scenario is possible ? Please give your valuable suggestion from your last experiences.
Thanks,
Kshitiz Goyal

> My question is, we want to minimize the downtime by directly installing Oracle 10g with R/3 4.7 EE and the take the offline/online backup of live PRD system and restored back to oracle 10g enviroment. By doing this we can reduce the downtime by aprrox 15 hrs.
Yes - this is possible.
I suggest you get the newest installation CD set for 4.7 which supports Oracle 10g (using sapinst, not R3SETUP):
See note "969519 - 6.20/6.40 Patch Collection Installation : Oracle/UNIX", Section 3:
3.) Retroactive Release of Oracle 10.2
Markus

How to delete file systems from a Live Upgrade environment

How to delete non-critical file systems from a Live Upgrade boot environment?
Here is the situation.
I have a Sol 10 upd 3 machine with 3 disks which I intend to upgrade to Sol 10 upd 6.
Current layout
Disk 0: 16 GB:
/dev/dsk/c0t0d0s0 1.9G /
/dev/dsk/c0t0d0s1 692M /usr/openwin
/dev/dsk/c0t0d0s3 7.7G /var
/dev/dsk/c0t0d0s4 3.9G swap
/dev/dsk/c0t0d0s5 2.5G /tmp
Disk 1: 16 GB:
/dev/dsk/c0t1d0s0 7.7G /usr
/dev/dsk/c0t1d0s1 1.8G /opt
/dev/dsk/c0t1d0s3 3.2G /data1
/dev/dsk/c0t1d0s4 3.9G /data2
Disk 2: 33 GB:
/dev/dsk/c0t2d0s0 33G /data3
The data file systems are not in use right now, and I was thinking of
partitioning the data3 into 2 or 3 file systems and then creating
a new BE.
However, the system already has a BE (named s10) and that BE lists
all of the filesystems, incl the data ones.
# lufslist -n 's10'
boot environment name: s10
This boot environment is currently active.
This boot environment will be active on next system boot.
Filesystem fstype device size Mounted on Mount Options
/dev/dsk/c0t0d0s4 swap 4201703424 - -
/dev/dsk/c0t0d0s0 ufs 2098059264 / -
/dev/dsk/c0t1d0s0 ufs 8390375424 /usr -
/dev/dsk/c0t0d0s3 ufs 8390375424 /var -
/dev/dsk/c0t1d0s3 ufs 3505453056 /data1 -
/dev/dsk/c0t1d0s1 ufs 1997531136 /opt -
/dev/dsk/c0t1d0s4 ufs 4294785024 /data2 -
/dev/dsk/c0t2d0s0 ufs 36507484160 /data3 -
/dev/dsk/c0t0d0s5 ufs 2727290880 /tmp -
/dev/dsk/c0t0d0s1 ufs 770715648 /usr/openwin -
I browsed the Solaris 10 Installation Guide and the man pages
for the lu commands, but can not find how to remove the data
file systems from the BE.
How do I do a live upgrade on this system?
Thanks for your help.

Thanks for the tips.
I commented out the entries in /etc/vfstab, also had to remove the files /etc/lutab and /etc/lu/ICF.1
and then could create the Boot Environment from scratch.
I was also able to create another boot environment and copied into it,
but now I'm facing a different problem, error when trying to upgrade.
# lustatus
Boot Environment           Is       Active Active    Can    Copy
Name                       Complete Now    On Reboot Delete Status
s10                        yes      yes    yes       no     -
s10u6                      yes      no     no        yes    -        Now, I have the Solaris 10 Update 6 DVD image on another machine
which shares out the directory. I mounted it on this machine,
did a lofiadm and mounted that at /cdrom.
# ls -CF /cdrom /cdrom/boot /cdrom/platform
/cdrom:
Copyright                     boot/
JDS-THIRDPARTYLICENSEREADME   installer*
License/                      platform/
Solaris_10/
/cdrom/boot:
hsfs.bootblock   sparc.miniroot
/cdrom/platform:
sun4u/   sun4us/ sun4v/Now I did luupgrade and I get this error:
# luupgrade -u -n s10u6 -s /cdrom
ERROR: The media miniroot archive does not exist </cdrom/boot/x86.miniroot>.
ERROR: Cannot unmount miniroot at </cdrom/Solaris_10/Tools/Boot>.I find it strange that this sparc machine is complaining about x86.miniroot.
BTW, the machine on which the DVD image is happens to be x86 running Sol 10.
I thought that wouldn't matter, as it is just NFS sharing a directory which has a DVD image.
What am I doing wrong?
Thanks.

Upgrade 10G to 11.1.1.5 in Clustered environment

I successfully ran the upgrade assistant on a single server install. Both RPD and Catalog upgraded with no issues. I scaled out to a 2 node cluster and attempted the upgrade with the assistant again. This time it fails. Here is log entries:
[2011-07-20T15:16:07.861-04:00] [Framework] [NOTIFICATION] [upgrade.Framework] Starting to upgrade components.
[2011-07-20T15:16:07.861-04:00] [Framework] [NOTIFICATION] [upgrade.Framework] Retain Source Ports: false
[2011-07-20T15:16:07.861-04:00] [Framework] [NOTIFICATION] [upgrade.Framework] Stop 10g Components: false
[2011-07-20T15:16:07.861-04:00] [Framework] [NOTIFICATION] [upgrade.Framework] Start 11g Components: false
[2011-07-20T15:16:07.876-04:00] [Framework] [NOTIFICATION] [upgrade.Framework] Managed server bi_server2 state is: RUNNING
[2011-07-20T15:16:07.876-04:00] [Framework] [NOTIFICATION] [upgrade.Framework] Starting to upgrade BIEE.
[2011-07-20T15:16:07.954-04:00] [Framework] [ERROR] [UPGAST-00138] [upgrade.Framework] upgrade exception occurred
[2011-07-20T15:16:07.954-04:00] [Framework] [ERROR] [upgrade.Framework] Cause: An unexpected upgrade exception has occurred. Action: See the secondary error message for additional details.
[2011-07-20T15:16:07.954-04:00] [Framework] [ERROR] [upgrade.Framework] UPGCMP-02712: Expected oracle.biee.admin:oracleInstance=*,type=BIDomain.OracleInstance,group=Service Oracle instance, found 2
[2011-07-20T15:16:07.954-04:00] [Framework] [ERROR] [upgrade.Framework] UPGCMP-02712: Expected oracle.biee.admin:oracleInstance=*,type=BIDomain.OracleInstance,group=Service Oracle instance, found 2
[2011-07-20T15:16:07.954-04:00] [Framework] [NOTIFICATION] [upgrade.Framework] Finished upgrading BIEE with status: Failure.
[2011-07-20T15:16:07.954-04:00] [Framework] [NOTIFICATION] [upgrade.Framework] Finished upgrading components.
[2011-07-20T15:16:07.954-04:00] [Framework] [NOTIFICATION] [upgrade.Framework] 0 components upgraded with success.
[2011-07-20T15:16:07.954-04:00] [Framework] [NOTIFICATION] [upgrade.Framework] 1 components upgraded with failure.
Question….can you use upgrade assistant in clustered environment?
Thanks!

You can only use the upgrade assistant to upgrade to a single instance Oracle BI 11g server. Typically this is a sandbox server environment or a VM that can simply be wiped clean after the upgrade. Then move the RPD,WebCat, etc. and AppRoles that get created during the upgrade process to a development server. The UA is looking for a single instance installation environment and that environment should be clean/fresh to avoid confusion and conflicts.
Please award points if helpful.

Steps to upgrade kernel patch in AIX cluster environment

Hello All,
We are going to perform kernel upgrade in AIX cluster environment.
Please let me know the other locations to copy the new kernel files ,
default location
CI+DB server
APP1
Regards
Subbu

Hi Subbu
Refer the SAP link
Executing the saproot.sh Script - Java Support Package Manager (OBSOLETE) - SAP Library
1. Extract the downloaded files to a new location using SAPCAR -xvf <file_name> as sidadm.
2. copy the extracted files to sapmnt/<SID>/exe
3. Start the DB & Application.
Regards
Sriram

Upgrade of 8.1.7.4 on Solaris 5.8 to Oracle 10g

Hi Gurus-
We are planning to upgrade some of the legacy databases to 10g.
They are all 8.1.7.4 instances on Solaris 5.8 with Standby configuration.
1)Is it possible to upgrade the instances on the same OS version(5.8) to 10g?
2)Where can I find the related documents which detail the supported OS versions pertaining to 10g?
3)Since the environment consists of Standby how can we minimise the downtime of upgrade? Any best practices and proven methods please.
Need to convert the existing DMTs to LMTs as well.
Appreciate your valuable inputs
Thanks,
newDBA

Hi newDBA,
to answer your questions:
1) Yes, according to Metalink ==> Certify ... Solaris 5.8 is certified to run an Oracle 10gR2 installation:
Operating System: Solaris Operating System (SPARC) Version 8
Oracle Database - Enterprise Edition Version 10gR2 64-bit
N/A Version N/A
Status: Certified
Product Version Note:
None available for this product.
Certification Note:
Existing Patch Sets:
10.2.0.2 Requires patch 5117016
10.2.0.3
10.2.0.4
2) See first of all the CERTIFY button on Metalink and make sure also to check the release notes and the installation guide for Oracle on Solaris:
http://download.oracle.com/docs/cd/B19306_01/relnotes.102/b15689/toc.htm
http://download.oracle.com/docs/cd/B19306_01/install.102/b15690/toc.htm
3a) Minimize the downtime with a standby
Your standby will be upgraded implicitly - this means you upgrade the production database and you restart your standby in MOUNT mode in the new environment you've installed on the standby machine as well. Once you switch your standby to ALTER DATABASE RECOVER MANAGED STANDBY DATABASE DISCONNECT; then it'll start synchronizing with your already upgraded production system.
During the regular upgrade either with DBUA or CATUPGRD.SQL there's not much possibility to shrink your downtime. You should create dictionary stats right before the upgrade but mainly the time it'll take is dependent on the number of options/components you've installed in your current database.
Recompilation by the way will be done automatically in parallel - so CPU power will save you some time here.
But always the question will be:
Can you accept approx 1-4 hours of downtime or not?
3b) Dictionary Managed Tablespaces to Locally Managed Tablespaces
As long as you don't touch the system tablespace this can be done online with no additional downtime. But as soon as you convert also the SYSTEM tablespace you have to startup your database at this point in RESTRICT EXCLUSIVE mode - which in fact means downtime.
Hope this helps - kind regards
Mike

Will there be any impact for OSD when upgrading ADK/MDT/R2?

We’re planning to upgrade our production ConfigMgr environment so that we can deploy Server 2012 and Windows 8.1.
Currently we have ConfigMgr 2012 SP1 CU3 with MDT 2012 Update 1 and Windows ADK 8.0. We have one Primary site and several secondaries.
My understanding is that I need to do the following:
Upgrade Windows ADK 8.0 to 8.1 (uninstall 8.0, install 8.1 on our Primary site)
Upgrade MDT 2012 U1 to MDT 2013 on our Primary site
Install ConfigMgr 2012 R2 on our Primary site
My primary concern with this upgrade is that we are in the middle of our Windows XP to 7 migration.
I am using an MDT 2012 U1 integrated deployment task sequence including the MDT 2012 toolkit, settings, USMT, and MDT database.
I am looking for feedback/links/resources/horror stories from others who have already been through this.
I would like to minimize OS deployment downtime and plan accordingly for outages.
Here are some concerns that I have:
Will uninstalling/reinstalling ADK have any effect on my existing production task sequence deployments?
They use a USMT 5 package from ADK 8.0
Will upgrading MDT from 2012 U1 to 2013 have any effect on my existing deployment task sequence toolkit, settings, boot image packages etc…
Are there any “gotchas” with the R2 upgrade that will impact my existing deployment task sequences or any ConfigMgr 2012 components (app deployment, software updates etc...)?
Any advice or links are appreciated. Thanks in advance!
-FP

We were in a similar boat, XP->Win7 migration and needed to do our R2 upgrade. We postponed the upgrade until just a month or two ago, and for good reason.
We found alot of issues with OSD, but have been able to resolve them fairly quickly.
WinPE 4.0 boot images must be rebuilt as WinPE 5.0 (this was expected)
Our WinPE 5.0 boot image could not contact the MP because of our HTTPS MPs in the DMZ. Had to stand up internal HTTPS MPs so boot clients could download Task Sequence policy.
All task sequences had to be recreated (we integrate with MDT and 1E Nomad) (this was also expected)
We are still fighting through our issues, but our biggest one just arose last week. Our OS images need rebuilt. We only have three, but we are going through that process now.
Microsoft has not confirmed the upgrade caused the OS image issues, but they recommend we rebuild to be safe.
The symptoms we saw were IE errors, desktop profile errors, DISM errors loading drivers, and black screen/white cursor hangs during Windows Setup. All issues were intermittent, one machine would build fine, another wouldnt.
I dont know what we messed up or configured that caused all this trouble, but the R2 upgrade was the trigger that caused it all to break.
Personally, I would definitely wait until your Win7 migration is complete, unless you can risk the possibility of having to rebuild everything if things go awry.
Daniel Ratliff | http://www.PotentEngineer.com

N5K - N2K Image Upgrade Single vs Dual Homed

Hi,
The upgrade distuptive behaviour is not explained well in the following documentation link . Especially about single homed FEX topology.
http://www.cisco.com/en/US/docs/switches/datacenter/nexus5000/sw/upgrade/503_N1_1/n5k_upgrade_downgrade_503.html
1. I have just upgraded a dual homed Active / Active N2K to N5K environment as below. The downtime was exactly explained as above URL, just the time to reload the FEXs. Which was about 1.5 minutes or so...
Or should we NOT reload all the FEXs almost simultaneously by manual command (as mentioned in the upgrade guide) in this topology, in order to minimize the downtime ?
2.However I am curious if there would be any downtime in a single homed FEX topology as below. My guess is there would be just the downtime for the edge server to distribute the traffic to the other N2K.
I guess if the downtime is really a biggy, then the second topology is easier to manage.
Comments area very much appreciated.
Dumlu

I am not able to see the images you attached, could you please try reinserting them?
Also if I am understanding your question correctly, you are wanting to know if there will be downtime, if you upgrade a N5K with a 2K single homed. I assume you meaning something similar to this?
If so, yes the single home fex would be reloaded per the document it states: "Upgrade the first vPC switch (vPC primary switch) - during this upgrade, the switch will be reloaded. When the switch is reloaded, any dual-homed FEX will not be reloaded. Only singled-homed FEXs are reloaded. Servers connected to these dual-homed FEXs retain network connectivity via the second (vPC secondary) switch"

Minimize upgrade downtime in HA environment

Similar Messages

Maybe you are looking for