Very good Duane!! It seems you've figured us out :-)
It does attempt to limit how many outstanding ICMP
requests are simultaneous, but the real number is
based on number of active maps and number of devices
on each map. It is possible with many maps that each
have >20 devices, to "flood" the network with ICMP
requests but its unusual.
John
In reply to 28 Aug message from [EMAIL PROTECTED]:
>I'm no expert here, but I can describe what I (think) I've seen
>so far with WUG Ver 6.02 -- YMMV on others. When I was writing
>custom services plugins, the docs made sure to notify you that
>your plugin needed to be thread-safe. I never figured out exactly
>why until playing around with it a bit more... I believe that a
>(very very rough) description of how the polling engine works
>would be:
>1. Wait for the poll timer on a map to make it to 0. When it
>gets there, start a poll and reset it
>2. Go through the devices on a map, one by one adding them to a
>'ping request queue'
>3. As you go through the request queue, send out an echo
>request. This causes a flurry of ping requests to go out at the
>beginning of a poll cycle.
>3. When a device returns a ping, kick off a new thread to do
>all of the 'higher' service polls defined. (Like HTTP, FTP etc)
>4. If a device's timeout happens before you get the ping
>response back, mark it as missed. If it's the 1st missed one,
>send out another ping request as almost a 'mercy poll'
>Of course, this is all theoretical - based on my attempts at
>plugin development and watching a protocol analyzer. But, if
>it's at least mostly true there is some good to be gained from
>it.
>1. If all of your maps are starting their poll cycle at the same
>time [ie all on 60 seconds] then you're gonna dump a boatload of
>ICMP traffic all onto the net at once. I don't think it matter
>if you have 5 maps at 20 nodes each or one 100 node map (from a
>polling standpoint). If you can stand to stagger some maps (poll
>important things every minute, things less important every 2.5
>minutes) that could distribute the ICMP load a little.
>2. Use the up and down dependencies to make sure that you don't
>poll unnecessarily.
>The big question in my mind is how close did I get to how it
>_really_ works?
>--D
>Duane Waddle
>[EMAIL PROTECTED]
>"With sufficient thrust, pigs fly just fine..." -- RFC1925
Please visit http://www.ipswitch.com/support/mailing-lists.html
to be removed from this list.
An Archive of this list is available at:
http://www.mail-archive.com/whatsup_forum%40list.ipswitch.com/
