OnDemand User Group
Support Forums => MP Server => Topic started by: wawad@sbm.com.sa on April 19, 2017, 04:39:12 AM
-
Hi
if any one have any documentation about how can i configure ondemand v10.1 High availability active active using zookeeper.
-
What does the official IBM CMOD Documentation say?
-JD.
-
the only one white paper and its not clear .
http://www.ibm.com/support/docview.wss?uid=swg27049639
-
the only one white paper and its not clear .
http://www.ibm.com/support/docview.wss?uid=swg27049639
wawad - I recommend you post any questions - and start a new thread - in the MP Server forums.
Ed
-
Moved to the MP Forum.
-
wawad,
I have read the documentation you provided, and it seems pretty clear for me...
You must have the following to be able to have high availability active/active:
First Step)
Pre-requisite:
- DB2 pureScale
- Oracle RAC,
- SQL Server Cluster
If you don't have that, then you cannot use the new functionality. So before doing anything else, setup your database accordingly.
Second Step)
Configuration of zookeeper
If you follow exactly what the documentation is saying, then you should be ok.
I will paraphrase here, since it seems not clear for you:
You must have at least 3 servers for zookeeper, you must have an odd number of servers >=3, so 3, 5, 7, 9 or more... 3 is the minimum which is advised.
Let say you have 3 servers: zoo1, zoo2 and zoo3.
On each server, you must do the following:
- Download zookeeper from https://zookeeper.apache.org/ (https://zookeeper.apache.org/)
- Unzip the file in one directory (ex: /opt/zookeeper)
- copy the file zoo_sample.cfg into zoo.cfg in the directory /opt/zookeeper/conf
- modify the new zoo.cfg to suit you taste and servers
- add in the zoo.cfg the following 3 lines:
server.1=zoo1:2888:3888
server.2=zoo2:2888:3888
server.3=zoo3:2888:3888
- in zoo.cfg you have an entry for the option 'dataDir' in this directory, you will need to create a file called 'myid', and you will need to write an ID inside this file, nothing more.
The ID to write is easy to find... in the previous step you have this line:
server.ID=zoo1:2888:3888
the ID is what you must put in your 'myid' file in the directory pointed by the option 'dataDir' from the zoo.cfg file.
so, to make it crystal clear:
in server zoo1, you need to write the file 'myid' with the content
1
in server zoo2, you need to write the file 'myid' with the content
2
in server zoo3, you need to write the file 'myid' with the content
3
- Go in the directory /opt/zookeeper/conf and write the following command in unix/linux:
$ export ZOOBINDIR=$PWD
$ . "${ZOOBINDIR}"/zkEnv.sh
$ java -cp ${CLASSPATH} org.apache.zookeeper.server.auth.DigestAuthenticationProvider userid:password
You can use any user / password that you want, it is not a user/pwd which is in CMOD or a Unix credential.
BUT you must keep them the same on all your zookeeper servers.
- Start your zoo keeper server, in the directory /opt/zookeeper/bin
$ ./zkServer.sh start
Last Step)
Once you have done that for each zookeeper server, you can do the following now for CMOD:
- For each that you have CMOD Servers, you need to update the CMOD Stash file (the one define in the ars.ini with the option SRVR_OD_STASH) and in each one you need to use the same user / password that you have used for zookeeper credential, by using the following CMOD command:
$ arsstash -a 10 -s <stash_file> -u userid
- Then you can (re)start your arssockd/arsobjd
And if everything is working correctly, you can access any library servers, and access your CMOD to view/arsload/administer it.
I hope what I said is clearer than the official documentation. If not, then please ask !!
-
More Thank for your replay . i already read this link but i have some question about database and storage.
in this situation the oracle RAC must be remote.
and cache storage ,external storage ,temporary file path ,print file path and data load directory must be shared .
and if its share which type of storage is support in this situation (NAS,SAN )
-
in this situation the oracle RAC must be remote.
That is no problem, just read this documentation as reference: http://www-01.ibm.com/support/docview.wss?uid=swg27019582 (http://www-01.ibm.com/support/docview.wss?uid=swg27019582)
and cache storage ,external storage ,temporary file path ,print file path and data load directory must be shared .
and if its share which type of storage is support in this situation (NAS,SAN )
What do you mean shared?
Every customer that I was working with was using mainly SAN.
I know that some (rare) where using NFS.
So if by shared you mean SAN, then yes it is possible.
As long as you can ensure stability, everything is fine.
-
Hi Alessandro
Please could you see the attached diagram for HA Active/Active and my question we can implement this as in Pic. or there is some thing wrong in this diagram
-
wawad,
Can you provide an update on how you implemented HA. And for COB considerations, how it was configured on a remote datacenter?
thanks
SV
-
sorry for the late answer, since I don't work with CMOD anymore (well my old company still asks me for help :-D so I cannot lose my CMOD knowledge too quickly!!) ...
From your diagram it seems ok. I hope you could implement it.
Does it work, as the documentation explains?
-
A bit new to this, does this mean that my database has to be remote using db2 purescale, and the OnDemand servers simply need CMOD installed on them and configured with zookeeper, and the database cataloged of course..