On 23 February 2016 at 18:15, Lunar lunar@torproject.org wrote:
David Goulet:
On 23 Feb (14:13:15), Tom Ritter wrote:
See https://www.marc.info/?l=tor-dev&m=145440925917524&w=4 (Don't count this as a commitment to mentor, but we can talk about it.
Custom Alerting Effort Level: Low-Medium Skill Level: Low Likely Mentors: Karsten, Damian
Tor has a lot of services that run in complicated ways. Bandwidth Authorities and Metrics archives are two large examples. While DocTor and DepicTor (consensus-health.torproject.org) provide alerting for certain DirAuth related concerns - these other parts of the Tor Project lack them. Improving the alerting around these systems would give greater confidence in their operation and help resolve problems faster.
Network monitoring GSOC! Very very very good idea! :)
In some ways, that's the case. But if an intern works on monitoring software, then we need to think from the beginning about who's going to run it and who's going to pay attention to its results.
Otherwise, even if it's the best code ever, it's likely to be a waste at the end of the Summer. (Please remember TorBEL.)
Truth. Since I run a bwauth and a collector instance, I'm incentived to learn, run, and maintain it (and pay attention to the results.) Especially if I can add my own tests into it easily enough (for making sure my bridge node/mixminion nodes are still running, etc)
I have crappy tests running now, and while I get more emails than I like, I know how to tell when things are really breaking. In my ideal world they'd improve the skeleton framework I have. If they wind up picking a language I'm not familiar with, or something that requires dependencies I'm unhappy with - I'm not gonna lie it's unlikely I'll maintain it. Just too little time to go around.
-tom