I am using PRTG to log SNMP bandwidth data from a Juniper SSG device. I cleared counters on the Juniper and the data logged during the corresponding 5 minute PRTG interval is crazy (e.g., a bandwidth spike on the serial interface of 700,000 kbps! As you'd guess, a spike of this size makes all the graphs rather useless).
I tried stopping the PRTG service, deleting the data folder for the date, then restarting, but graphs and data table (in the win32 app) still show the data.
I'm not sure if this is a data issue, or a graph/table issue.