The poller task uses 10 SNMP sessions in parallel and changes the IP address
when a reply is received or a timeout occurs so all our network is scanned.
If a device does not respond to get requests for 30 seconds then a condition
(we call this an alarm)is raised. When the device responds any condition
(alarm) is cleared.
What we have been seeing for sometime is an alarm for an inactive device being
cleared by the receipt of a false reply (actually a reply from another device).
What is happening in scotty seems to be:
1) A new session is created - the session reqid is set to a random number.
2) A get request is sent - the request id is incremented and sent in the
message.
3) When a reply is received the request id is used to find the requestor.
Unfortunately, when a large proportion of our network is disabled, the request
id of one session can catch up with another and thus a reply is posted to a
wrong requestor.
---------
Malcom Bright |Email work mdjb@jet.uk home malcom@wayzidar.demon.co.uk|
a Bystander of the |Snail 59 Queensway Didcot OX11 8LY |
Internet Super Highway|
-- !! This message is brought to you via the `tkined & scotty' mailing list. !! Please do not reply to this message to unsubscribe. To subscribe or !! unsubscribe, send a mail message to <tkined-request@ibr.cs.tu-bs.de>. !! See http://wwwsnmp.cs.utwente.nl/~schoenw/scotty/ for more information.