MIMIX Recommendations for 8.12

It's because your AS400 host names are different.

You can either (1) Rename the DR box as part of your recovery process or (2) Create a script to go change the host name in about 500 places on the DR box as part of the recovery process.


There are a few other tricky things you can do but I can't recall if we can do them all the way back to 8.97

Colin
 
I'm going to disagree with you Colin. The server names don't need to be the same, only the *LOCAL relational database name needs to be the same on each. We have done a failover test with this config and it worked perfectly. As part of our fail over process the fail over script that Mimix runs activates an ethernet line on the fail over box that has the same IP as the Prod box and the prod box ethernet line gets disabled. The ethernet lines on each box reserved for Mimix don't change. Our servers are in the same subnet so that makes it easy. If they are in different subnets you would have to change the IPs of the servers in DNS and wait for the changes to replicate. Be sure you don't have any host table entries that would override DNS.
 
Actually my previous post is correct.

Mimix is not the issue and is easy to get up and running.

What the post wants is for JDE to work.

The Enterprise Server is very host name aware in a default configuration and the easy way around this is to rename the DR box as part of the "fail over test".

The security server won't come up without a few changes on a machine that has a different host name without a few changes.

Colin
 
I made the host names the same and it didn't help any.
Still can't log into security server.

Would it matter if we are leaving up our production system when we bring up our DR box services? DNS still points to production so users are not effected and using IP address to connect to JAS on DR box.

Thanks for your input.
 
Cody

I've gone through this drill a few times with one of my AS400 customers. To tack on to Colin's post (we work together), in our strategy, we simulate like a true DR scenario. The regular enterprise server is backed up and ipl'ed. We put it in a restricted state, leaving it down.

We fail over the customer's network link from our primary data center to our DR data center.

We bring up a copy of their domain controller and validate DNS.

Then we run a script to change the host name and ip address of DR AS400. It assumes the identity of the regular 400.

Then we bring up that server and bring up JDE. I validate that JDE comes up using a fat client. When I'm happy with that, we bring up Websphere and I validate that.

We let the customer do their testing, then reverse the process to put the DR boxes back to sleep and bring the primary boxes back on line.

That's a 10,000 ft view of our process. Our actual procedure is a 33 page doc with tons of scripts, tests, and checkpoints. Quite a bit of work to put this all together. And before someone asks, no I can't share the document, that is proprietary information.

- Gregg
 
Thanks for your reply Greg.

I guess what I'm wanting is a little more detail on the iSeries configuration between the two boxes. I gather that your primary & dr boxes are identical where JDE is concerned? I need to test JDE on the DR without mimix involved to validate the restore from primary to DR.

We have a host table entry on the DR box that points the host name to the DR boxes IP address. Would this be enough to test the DR box if it is an all-in-one configuration?

Cody-
 
Please post screen shots of each of the following from the PD and DR boxes:

1. CFGTCP Option 10
2. CFGTCP Option 12
3. RDBDIRE
4. DSPNETA
5. A sanitized version of your JDE.INI file


Colin
 
Colin,
Here are the screenshots and files you requested attached.

Cody-
 

Attachments

  • 187246-Screenshots.txt
    18.7 KB · Views: 22
[ QUOTE ]
Actually my previous post is correct.

Mimix is not the issue and is easy to get up and running.

What the post wants is for JDE to work.


[/ QUOTE ]

To clarify my previous statement a little more then is we run a failover test and bring up JDE, web and fat clients included, and everthing works fine and dandy. It doesn't matter if your primary server is actually shutdown or not as long as the primary ethernet line that uses the IP in DNS is disabled it's the same as unplugging the cable from the server or shutting it down. As long as the *LOCAL RDB entries on both servers are set to whatever the primary server name is supposed to be and always kept that way then you should be good to go. We worked with Mimix Professional Services devision to build the failover scripts, which they did in about a day, and then did a full test with them over the phone over a weekend in case of issues but we didn't have any that they needed to address when we actually did it.

Maybe the way you have your iSeries configured you have to do the host name rename but I know that you can set it up so it's not required and I know of other JDE customers using Mimix that have it setup and working the same way we do. I'm not saying that the way you are doing it is wrong (not trying to undermine anyone or step on anyones grand intelligence!) but I'm just saying it doesn't have to be done the way you're describing either.

Taking Greg's line, our 10,000 ft view of the process is to run the necessary Mimix commands to switch the journal replication(all scripted as well) and then run the scripts to disable the ethernet with the primary IP on the Prod box so no clients can connect to the real Prod box and then enable the ethernet line on the DR box that has that same IP so clients are really sent to the DR box then. As I said before our Prod and DR boxes are on the same subnet so it's easy to do just that. If they were on different subnets we would make some additional changes in DNS.
 
I have to agree with Junior. We are set up the same way. The actual machine name doesn't matter, as long as the *LOCAL RDBE is correct. As long as the DNS name goes to an IBM i that has the correct RDB name JDE is happy.

I have attached the WRKRDBDIRE screen (sorry about supressing the IP's but corp doesn't allow me to give them our) for our JDEHA machine.

In our case we have to change the IP address in DNS because our primary is in the US and the HA is in Europe. We are brining up a second one in the states so we can failover to it without the DNS stuff to do upgrades, PTF's, etc. It will be working as Junior says, just bring down the line with the JDEPROD IP on PROD, and bring the line with the IP up on HA.

Hope this helps.

Tom
 

Attachments

  • 187296-JDEDHA WRKRDBDIRE.png
    187296-JDEDHA WRKRDBDIRE.png
    11.2 KB · Views: 15
Upon more testing of our DR box I found that there were several directories and files on the IFS that have the wrong permissions. I suspect it was from the restore from tape. I corrected the permissions issue with a compare to our primary box.

I was unable to log into the JAS, unable to connect to security server error. I found a JAS log that said it couldn't connect to port 6014. DR & Primary were both set to use the same port-6014. I changed the port for our DR box to 6015 and now I can log in.

Did anyone else have to change the JDE port number?

CORRECTION:
Changing the port number did not fix anything!
confused.gif
 
Back
Top