Closed fendemann closed 4 years ago
It turns out that there are duplicate VRPs. That’s not really surprising in hindsight, but I didn’t consider that. So now the metrics counts all VRPs while duplicates are filtered out in all output formats.
The difference is actually kind of interesting, so I think I will keep the current vrps_total
as it is and add a new unique_vrps
metric that only counts unique VRPs.
Sounds good. Will it be provided with 0.7.2 ??
I would have waited to 0.8 but that may still be a bit away. If you would like a fix now, I am happy to make an intermediary 0.7.2 pretty much now.
I would very much like to start monitoring the unique_vrps metrics, but only have the next 5 days. After that, I'm not in the office for 3 weeks. If the 0.8 is planned for the end of September, I can wait for it too.
Then let’s fix it in 0.8. I’d rather no rush a release.
I just checked "/metrics" on a routinator 0.8.3, and can't see "unique_vrps". Was this really added? Or should I lokk for something else?
0.8 only has the overall number of unique VRPs in routinator_vrps_final.
But that doesn't provide stats separated by TA, right?
Correct. You could determine them by subtracting routinator_vrps_duplicate
and routinator_vrps_filtered_locally
from routinator_vrps_total
. Unless you have changed the unsafe_vrps
policy to reject
, in which case you would also need to subtract routinator_vrps_unsafe
.
Current main doesn’t have routinator_vrps_final
per TAL either, that should be part of #494.
Hi, can u please explain the following differences:
While monitoring the Routinator metrics I found differences between the vrps values in the metrics file and the line count [grep on tals] in the csv file.
As far a I understand a vrp (validated roas payload) value is a one-to-one mapping between an ASN and an ip address and the csv file is an export of these mappings. So counting the lines in the csv file should bring up the same values, or not !?
Frank