Norman Yamada <nyamada at millburncorp.com> writes:
> That's why I thought it was safer to update the tables by hand on each
> node. But if you think it'll work, I'll do it via store path , because
> it's faster...

There will be a brief race condition where the servers are a bit
confused as to which server is which, however it shouldn't cause any
problems, because the slons drop connections whenever they realize
that the node number they're connected to doesn't match the node
number they expect.

You can see this in src/slon/remote_listen.c; it disconnects if
"db_getLocalNodeId() returned % - wrong database?"

I don't think "safe" has a unique meaning, here!

 - In my mind, it's somewhat unsafe to do the direct updates as 
   there's risk of divergence of configuration introduced.

   If you "update" then "select" to check the results, then that
   mitigates that risk.

 - Yes, there's something a bit unsafe about "just hoping the
   configuration stabilizes" :-).

On balance, given that you'll be checking your config afterwards to
make sure the paths are right, then it's fine to update them by hand.
-- 
output = reverse("ofni.sailifa.ac" "@" "enworbbc")
<http://dba2.int.libertyrms.com/>
Christopher Browne
(416) 673-4124 (land)
"Bother,"  said Pooh,  "Eeyore, ready  two photon  torpedoes  and lock
phasers on the Heffalump, Piglet, meet me in transporter room three"