Re: memory corruption on reload

Morgan Aldridge wrote:
> I definitely agree with you on this one, but I bet the logistics can  
> be pretty complicated. For example, I use rely heavily on Sysmon's  
> ability to handle fail-over monitoring on another server, but would  
> one want to sync that history between the various monitoring servers?  

Two fairly simple solutions to this that I can think of:
a) let sysmon store the data in an SQL database, for example MySQL, and
use MySQL's clustering or replication to provide hot-backup. Replication
in MySQL works well even over a WAN.
b) let each sysmond store the data in files on the filesystem, and
create a tool to merge the info from multiple servers.

> Also, how do you handle when the Sysmon daemon is down during a  
> reboot or other maintenance? Assume the services were up the entire  
> time?

If you want figures accurate to five nines, run multiple servers. If
not, then the state of the service at the point when sysmon is killed
should be preserved. Once sysmon is restarted it should check all
services and update as necessary. So if your service was down when
sysmon was shut down, it will be counted as down for the duration that
sysmon is shut down. At least that seems to be the logical answer for
me, but maybe you have a different idea!

-- 
Chris Wik
ANU Internet Services Ltd
W: http://www.anu.net/
E: [email protected]
T: +44 (0)117 9118820