|
I'm not completely sure, but it looks very much as you say - your data collection scripts are so slow that they are taking more than your (5 min?) polling window to run.
If MRTG does not complete its polling within the window, then the poll is missed, and 'unknown' is implicitly stored in the RRD for that window (if the poll subsequently completes, the data goes into the NEXT window in the RRD instead)
In the RRD is the 'XFF' setting which says how much of the data need to be there. This is default at 0.5 which means you can miss half the polls required for the RRA and get away with it, but miss more and it goes unknown.
The 'daily' graph is 1 poll per bucket, so miss one and you get a grey unknown.
The weekly is 6 polls per bucket, so as long as you have 4 of the 6 then it will still display.
And so on for monthly (24) and yearly (264).
Therefore, if you are missing half your polls (because a poll takes >5min to run), you'll get nothing.
You need to try to run more MRTG instances at once; under UNIX/Linux you can use Forks: to do this (doesnt work under Windows though) or split your cfg files into groups and run multiple instances of MRTG.
_________________ Steve Shipway UNIX Systems, ITSS, University of Auckland, NZ Woe unto them that rise up early in the morning... -- Isaiah 5:11
|