Connected environments and Fail/Over

Vincent,

This is only a guess:

The search paths on nodes are for Name Services and connecting environments

merges Name Services.

When your scenario works:

Name Service Backup Env (NSBE) contains the entries BE.P1 then ME.P1.

Name Service Main Env (NSME) is not available so the node looks at NSBE.

It requests P1 and gets BE.P1.

When it doesn't work:

NSBE contains the entries ME.P1 then BE.P1. When the node requests P1 it

gets the reference ME.P1. ME.P1 as part of a connected environment will be

something along the lines of a reference to NSME.P1 rather than an actual

lookup on the NS and thus touching the reference will fail.

The clients on the other hand do not use the reference directly but use the

NS (NSBE) to dereference the reference and thus resolve the reference to

BE.P1.

To shed some more light try using escript and

findsub nameservice

nsdebug

on all the configurations.

Regards,

Richard Stobart

Technical Consultant for Forté

E-mailRichardStobart@mail.com

Quick-mail:0832691942@sms.mtn.co.za

Voice:(+ 27 83) 269 1942

(+27 11) 456 2238

Fax:(+ 27 83) 8269 1942

--Original Message--

From:Vincent R Figari [SMTP:v.figari@juno.com]

Sent:Tuesday, February 03, 1998 4:23 PM

To: RichardStobart@mail.com

Cc: James_Hutchinson@chordiant.com; kamranamin@yahoo.com

Subject:Connected environments and Fail/Over

Hello,

About connected environments, does anyone know the reason for this

strange behavior :

- Forte=2.0H1, Server=AIX, Client=NT 4.0 Ws

- 2 Connected environments, MainEnv and BackupEnv on 2 separate servers,

with Env. search path = @MainEnv:@BackupEnv

- Partition P1 replicated for Fail/Over in MainEnv and BackupEnv

- Partition P2 running on a 3rd server, in MainEnv

To simulate a Hardware Failure, we kill both MainEnv (ie its EnvMgr) and

partition P1 on the 1st server, and we expect both the clients and the P2

partition

to switch to the replicate of P1 in BackupEnv on the 2nd server.

This scenario runs great if :

1. We connect BackupEnv to MainEnv

2. We start P1 in BackupEnv

3. We start P1 in MainEnv

BUT, if from this point, we stop and restart P1 in BackupEnv, and repeat

the

crash scenario, then the clients still switch to the P1 replicate in

BackupEnv,

but partition P2 raises a DistributedAccessException telling that it can

not find

the requested SO and that it lost contact with MainEnv Name Server...

In other words, if we start P1 in MainEnv BEFORE starting P1 in

BackupEnv,

then other server Partitions in MainEnv never "see" the P1 replicate

available

in BackupEnv.

Any clue ?

Vincent Figari

You don't need to buy Internet access to use free Internet e-mail.

Get completely free e-mail from Juno at <a href=

"http://www.juno.com">http://www.juno.com</a>

Or call Juno at (800) 654-JUNO [654-5866]

at 2007-6-29 9:22:25 >