Michael Peppler
Sybase Consulting
Sybase on Linux
Install Guide for Sybase on Linux
General Sybase Resources
General Perl Resources
BCP Tool
Bug Tracker
Mailing List Archive
Downloads Directory
Sybase on Linux FAQ
Sybperl FAQ
Michael Peppler's resume

sybperl-l Archive

Up    Prev    Next    

From: Liz <ainsleyg at optonline dot net>
Subject: Re: Replication Server
Date: Jul 24 2001 3:26AM

we use a perl / sybperl script to do this....
a) poll on whatever interval is appropriate for you using :
while {
poll the target databases;
run your queries;
sleep for some amount of time;
b) push a heartbeat thru from each source database ... you decide the
interval. Ours is 5 minutes.
c) use the perl script to :
1) check sq space (simple rssd query
2) check the rs_lastcommit at the replicate.
calculate the latency and time since the hearbeat was last replicated.
if the amount of time 'since' is greater than some interval (certainly more
than the hearbeat interval , obviously) then page or email.
Note: recent rep server p&t doc available at sybase (one of the best docs
I've ever seen imho - from sybase or anyone) advises NOT to do this.
We've been doing it for 3 years and not had a always one's
mileage varies.
We do this because we've had ltm's stop replicating as a result of network
hits BUT admin who_is_down shows that every is happy.
Besides monitoring latency is an absolute must .....
honestly ... the script is about 1 hours work if you crib from Michaels
sybperl docs (hell , that's how I did it....)
start out simple (I did) and add complexity (like more replicate database
servers , rep servers etc)
other alternatives ... checkout those fellows know
their stuff and you'll probably find something there. You might also want to
look at John Kirkwoods recent Rep Server book ... you'll get some useful
info from that...


----- Original Message -----
To: "SybPerl Discussion List" 
Sent: Monday, July 23, 2001 12:13
Subject: Re: Replication Server

> > (a) Use the heartbeat facility built into repserver
> > as well as the replication server monitor (RSM), if all you want to do
> > make
> > sure all of your processes are alive
> This tells you if the processes are alive and that replication is backed
up, but
> not how full the queue's are.  At least, not all in one screen and not in
> 'at-a-glance' format.  I agree that the tools are there, but you do have
> dig a bit for all of the data.  The other thing that is missing is
> of failure from the standard toolset.  It would nice to have a facility to
page or
> email people when certain errors are detected.
> > (b) Write a script that will check the rowcount for all replicated
> > tables on a given DBSERVER/DATABASE/RepServer --- I can't post my script
> > (sigh)
> > but the logic is:
> Another alternative for this part is the rs_subcmp, which need not be too
> on the server, provided you choose you where clauses carefully, but it
does require
> a certain amount of free processing power on the machine that it is
running on.
> dowen