On Fri, Aug 24, 2001 at 06:04:36AM -0700, Stuart Kendrick wrote: > I'm wondering how other people solve this problem. Would you be willing > to tell me how you do it? When I was doing NMS consulting, we tended to use three solutions: 1) use a monitoring package like Netcool/Impact that has the capability natively (which doesn't address the swatch question, of course); 2) have the exec statement in swatch run a script like the one you wrote; or 3) have the exec statement drop a file into one of a few severity-sorted directories, and have different cron processes periodically scan those directories and send pages based on their contents. When you don't have the $$ for the first solution, I'd tend to go with the latter one, because it lets you do some cool stateful tricks. For example, if you have your execs drop pager messages into directories named "informational", "warning", and "critical", you could have cron only scan the warning and info dirs during business hours, and drop the DHCP messages in one of those directories. If you also name the pager message files by the pager ID of the recipient plus some unique identifier, you could also do things like having the cron script only send one page per recipient per cycle, and only send the subjects of the messages rather than the entire message (after moving the relevant message files to a spool that the user can later look at manually) if the number of queued messages is greater than X. -- Sweth. -- Sweth Chandramouli ; <svcat_private> President, Idiopathic Systems Consulting
This archive was generated by hypermail 2b30 : Fri Aug 24 2001 - 12:03:59 PDT