Open fno2010 opened 7 years ago
I think this line is incorrect and conducts the divide-zero exception. Because getSecond
will always return 0. And I believe it also conducts issue #10.
The quick fix cannot work because many node-connector
s cannot update their duration
properties (not sure it's the hardware issue or ODL issue):
"openflow:75096032357597:3401"
{
"second": 0,
"nanosecond": 0
}
"openflow:75096032357597:6201"
{
"second": 0,
"nanosecond": 0
}
"openflow:137432795454528:111"
{
"second": 819660,
"nanosecond": 527000000
}
"openflow:137432795454528:89"
{
"second": 808576,
"nanosecond": 769000000
}
"openflow:365545302427456:137"
{
"second": 0,
"nanosecond": 0
}
"openflow:365545302427456:225"
{
"second": 0,
"nanosecond": 0
}
A trade-off may be taking the timestamp in controller side (system time).
We do not test basic bandwidth monitoring function in either unit tests or integration tests. From my basic code review, it does not work in some special cases. For example, how can this line make sure the initial monitoring work?
We need somebody to check it and make sure it works.