Discussion:
How to create a test server identical to production?
(too old to reply)
Bret Halford [Sybase]
2008-10-22 15:46:36 UTC
Permalink
Raw Message
In the past I have installed ASE & ebf to match production
server and then ran disk inits, create & alter database
scripts so that new test dbs match production and then
loaded from dump to create a test server. Is it possible to
load a master db dump into the new test server, rename it to
'test' and load user dbs?
No, not on a new one. The problem is that the master
database is pointing to a bunch of devices which may
or may not exist at the same path name. Lets say thay
do. When you boot up on that master database, ASE will
activate all those devices and then try to recover
those databases. Even if the devices exist, they don't
contain the expected data (system tables, etc) to
be a recoverable database, so they will be marked
suspect.

It does work if you can make copies of all the production
devices either while the production server is shutdown,
or if the copies are made while the "quiesce database"
command is active. But then you will have a copy
of the master device as well, and so no need to
have/load a dump of master db - it is already on the device!

Cheers,
-bret
John Flynn
2008-10-22 22:43:20 UTC
Permalink
Raw Message
Post by Bret Halford [Sybase]
The problem is that the master
database is pointing to a bunch of devices which may
or may not exist at the same path name. Lets say thay
do. When you boot up on that master database, ASE will
activate all those devices and then try to recover
those databases. Even if the devices exist, they don't
contain the expected data (system tables, etc) to
be a recoverable database, so they will be marked
suspect.
I once solved a similar problem through use of mirroring. First I arranged
it so my devices were mirrored to other devices whose names didn't exist on
that server, but did exist on the second server. When I started the first
server, it would report errors due to the invalid secondary device names,
but the database would still start up with just the valid primary halves of
the mirrors. Later, I copied the database to the second server. When I
started that 2nd server, with an exactly duplicated master, it would get
errors due to the primary device names being invalid, but then it would
failover to the secondary halves of the mirrors.

This was convenient since I frequently needed to copy this database from one
server to the other, but the physical device names were not the same between
the two. This isn't the exact situation the OP described, but maybe it can
be adapted.

- John.

Loading...