On 2014.03.20. 9:49, Robert Kisteleki wrote:
On 2014.03.19. 12:02, Brett Carr wrote:
2. An ability to provide sms/e-mail alerting depending on certain (user definable) conditions.
To be slightly more specific,
https://labs.ripe.net/Members/suzanne_taylor_muzzin/introducing-ripe-atlas-s...
Something more DNS specific but similar to the above would be very useful.
This is entirely possible. In order to implement it, we'd like to keep the mechanism described in the article, but extend it to support DNS. Still, because determining success/failure of DNS measurements is a bit more tricky than pings, the details need some discussion. I invite all to have that discussion on the MAT-WG mailing list (also on cc now).
As a conversation starter: what would you expect from these DNS checks? Some ideas / questions: * the generic framework will be the same (pull model, options specified in the URL, answer method still applies, ...) * checking RTT probably does not make that much sense for DNS, can we ignore that and do some combination of the ones below instead? * always treat "unable to connect" or "no answer" as an error condition (per probe)? * option for treating any response except "NOERROR" as an error condition (per probe)? * is there a need for comparing the results to an expected "known good" value? note that this is somewhat difficult in cases of random A/AAAA order. Maybe do this for SOA (serial) only? Please chime in if you're a DNS guy interested in how we could help you. (And I'd like to recommend reading the article and the status check API documentation first.) Regards, Robert