Our ID server crashed without a backup. It's gone.

At the same time we have a number of 24/7 live rep.servers using that
server as an ID, and they have been running live for some time without
an ID server.
Replication is used for warm standby only. We replicate a number of
databases in such fashion.

What is the proper procedure to reinstate the lost ID server on such
systems.
What is the negative impact?

Any hints will be highly appreciated.

Office Drone wrote:
I assume you mean you lost the RSSD?

And you've exhausted all possible solutions for recovering said database?

And no (backup of) scripts used to build the ID server in the first place?

Well, all the recovery scenarios I can find assume a copy of the RSSD
exists. A couple of the recovery procedures go onto to require you to run
rs_subcmp between the various RSSD's in your system so as to re-sync the
newly recovered RSSD with any activity (repdefs/subs) that may have
occurred since the last dump of the RSSD.

I'm wondering if you could build a new ID server (hey, this would be like
loading the RSSD from a really, really old dump, right? ;-) ... then follow
the procedures for syncing upstream and downstream RSSD's with the new (ID)
RSSD?

Assuming RS 12.6 ... Chapter 19 (Replication System Recovery) has a
sub-section titled "Recovering from RSSD failure" ... check out the
sub-sub-sections titled 'Subscription comparison/re-creation procedure" ...
to get an idea of what I'm talking about (re: rs_subcmp).

Step 6 under the 'Subscription comparison procedure' has some additional
steps to take if you're rebuilding the RSSD, eg, manually updating
rs_idnames (the id name/# pairs).

You won't be able to add new RS's, ds/db's or routes.

Contact Sybase TS to see if there's anything else they can suggest?

--
Mark A. Parsons

Iron Horse, Inc.
iron_horse (AT) NOSPAM (DOT) compuserve.com

fulbrighter2 (AT) yahoo (DOT) com (Office Drone) wrote in message news:<68497021.0405111426.d4cc5ba (AT) posting (DOT) google.com>...
Don't bother about the lost ID server. In your case simply create a
blank ID server and add a rep.server each time you do the new switch
or stop/restart the new replication (I mean databases that still point
to the old ID). It takes little time, but then you don't need to worry
about the ID server. Even better solution would be to make replication
servers that are combined Rep/ID servers. You can later define an ID
server domain, should you decide to start using routing.

fulbrighter2 (AT) yahoo (DOT) com (Office Drone) wrote in message news:<68497021.0405111426.d4cc5ba (AT) posting (DOT) google.com>...

Thanks for the info everyone. I also been in touch with Sybase. Their
response was that reconstructing the lost ID server is the task they
would recommend to torture enemy DBAs in detention centers.

And basically they recommended to recreate a blank ID server, and add
rep.servers & databases to it, which is not a very difficult task. So
we're back.