Prev: Transaction log replication
Next: WITH FILLFACTOR is missing when snapshot recreates indexes
From: Ben on 15 Jan 2010 10:18
I know that I can alter the location of a snapshot both by setting the
default snapshot location, or by specifying the alt_snapshot_location
parameter. However, both of these only specify the beginning of the
path (e.g., E:\MSSQL\REPLDATA). What I'd really like to do is
determine where the snapshot was actually stored (full path), e.g. E:
\MSSQL\REPLDATA\unc\MACHINE_DBNAME_PUBLICATION\.

I can tell generally how Microsoft puts this path together with
<MACHINE>_<DBNAME>_<PUBLICATION>, and that's what I've been doing for
several months. However, I recently found that if this name becomes
too long (about 64 characters), Microsoft mangles the name by trimming
it down to about 55 characters and appends a short hex string.

Now, obviously the subscribers can find this directory when they need
to apply the snapshot, so Microsoft is either able to generate it on
the fly, or they store it somewhere. I would like to be able to know
the algorithm for generating it, or where they store it at.

My need is this: I'd like to use tar and gzip to compress the
snapshot (and specifically excluding the .bcp files), and since we
have multiple snapshots in the REPLDATA directory, I don't want to
compress and move all of them all the time.

Does anyone know where I can find this information?
 | 
Pages: 1
Prev: Transaction log replication
Next: WITH FILLFACTOR is missing when snapshot recreates indexes