Hmmmm, I'm wondering if a "Hyper-Enterprise" level version
(with matching cost level) that had the full multi-threading, and hosts saved as
a "real" database would be an option. I know that I could easily convince my
supervisors here to spend "real" money if I could get all the data placed on our
high availability SQL servers (nothing saved in the registry, no "hosts"
file, etc...) and could reduce the cycle times. If it was geared towards
those of us who MUST have this higher level of usability and accountability (due
to our own client SLA agreements or contracts) and where there were a large
number of checks, say greater then 200 or so, it should make real business
sense.
I have
to provide externally available reports on the SA system, (who gets what alerts,
escalation trees, uptimes, etc...) and currently this is a real bugger to
do.
Michael D. Shook
Technical Analyst
[EMAIL PROTECTED]
863 668 4477
(work)
863 860 4070 (cell)
863 665 1261 (fax)
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Kevin Stone
Sent: Friday, February 18, 2005 10:02 AM
To: salive@woodstone.nu
Subject: RE: [SA-list] Major product checks missing?
I am also at ~1500 checks and could add 3-500 more.
Every couple of months I ping Dirk about threads and parallel checks.
My understanding is a complete rewrite of the check code with a set of new tools
would be required. I don't know how many of the SA community are or could
be in the 1000+ check range and if it would be worth the development cost.
That, of course, is Dirk's decision.
To mitigate the cycle times I've done many of the same
things. Lower priority checks happen less often, I've lowered the timeout
on pings, reduced the number of web pages generated etc. I also made a
change in the registry to lower the "PerItemCycle" value. This made a huge
difference on my system.
SA works extremely well in our environment despite our
pushing it beyond its normal scope. A comment was made to me the other day
that my monitoring system(based on SA, Kiwi tools, and PHP) is much easier to
use and more informative than the corporate Concord system. That's a
testimony to the usefulness built into SA.
-Kevin
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Peter Hoermann
Sent: Thursday, February 17, 2005 4:18 AM
To: salive@woodstone.nu
Subject: RE: [SA-list] Major product checks missing?
I concur, although let me share my "workaround" to the long amount of
time for a check cycle to complete
I have been using Servers alive since version 2, and over the past 5
years my hostfile has grown to around 2000 checks in total across 10 sites, with
wan speed dictating that some service, disk space and count file checks taking
up to 5 seconds per check. (you do the math)
I've since grown a brain and paired down my entries to around 1450,
removing stupid checks like terminal services, leaving bare essential,
production / workflow affecting checks.
Recently I went through and thought of how I could decrease the cycle
time. All of my disk space checks I have chose now to do every 20
cycles. Service checks against things like server scan service form trend
I run every 20. Service checks which are annoying to us if not running,
although don't affect the user community, eg: exchange sa, I've reduced to
one check every 50 cycles.
Mail queues and information store checks, sure I still run every cycle,
since if these aren't running, my support guys are gonna start getting
phone calls. My scope was if a check fails, and I wouldn't get a call
from users about it straight away, I now only check it every 5
cycles.
By going through all of my entries I got my cycle time down from 20-25
mins to around 5-6 mins. Sure, the first cycle is going to take some time
if the checks are set to start on first cycle, but the return is worth
the bit of time of configuration.
My 2 â cents worth
Peter