| @jim:acmegating.com | i just returned from attending the "Gerrit Meets" event in person; there were two talks, both of which i think are of general interest here: 1) an overview of the cached-refdb plugin and the performance issues it addresses; and 2) a preview of gerrit 3.14 changes | 00:25 |
|---|---|---|
| here's the current url: https://www.youtube.com/live/qmtpjV7X9Wk?t=6732 | ||
| that's the livestream url; i imagine it'll get a permanent home once they edit the start/end, so that might change. shouldn't be hard to find if it does. | ||
| @fungicide:matrix.org | i'm taking advantage of a quiet friday to venture a bit further up the island for lunch, but i should be back in a couple of hours | 15:53 |
| @jim:acmegating.com | bon appetit | 16:02 |
| @fungicide:matrix.org | thanks! it was clearly a good time to get away for a bit, but back now | 18:11 |
| @fungicide:matrix.org | one thing i wonder if the gerrit community has a solution for is curbing notifications to dead addresses. the sender configured in our gerrit is inundated with ndrs all day long, and aside from implementing a manually-updated discard list in the outgoing mta i'm not sure what our options are | 18:23 |
| @fungicide:matrix.org | i suppose we could start just discarding any incoming messages to the sender address instead, but that seems less respectful of others' systems | 18:24 |
| @jim:acmegating.com | i haven't heard of such a think. but maybe we could write a script to collect those and disable accounts. basically a VERP script. | 18:28 |
| @jim:acmegating.com | * i haven't heard of such a thing. but maybe we could write a script to collect those and disable accounts. basically a VERP script. | 18:28 |
| @jim:acmegating.com | i suspect a lot of corp environments don't have this problem, but maybe other floss projects do. could be worth an email to the list. | 18:28 |
| @fungicide:matrix.org | yeah, an external bounce analyzer is a possibility (and maybe such a thing already exists, or we could make a wrapper around the flufl.bounce library mailman uses), just didn't know if there were any hooks or plugin points to more tightly integrate a solution | 18:29 |
| @fungicide:matrix.org | the biggest part of the problem is that, without a proper verp type implementation, identifying the bouncing address is very challenging | 18:31 |
| @jim:acmegating.com | oh, so the tech level here is just extracting an addr from an ndr :( | 18:37 |
| @fungicide:matrix.org | yeah, verp solves that for well-behaved mtas, but there are also a lot that bounce back to something other than the sender | 18:40 |
| @fungicide:matrix.org | or rather, a verp implementation makes that determination trivial for well-behaved mtas | 18:41 |
| @fungicide:matrix.org | it doesn't appear that gerrit inherently supports verp, or if it can then we don't have it turned on and i'm not locating the documentation. but in theory we could use an outbound filter in exim to change the sender based on the recipient address and then a corresponding inbound filter to make a database of things that arrive destined for verp addresses | 18:50 |
| @fungicide:matrix.org | mailman has some heuristics like requiring a certain number of messages over a set timespan in order to consider the address defunct, and a big part of what flufl.bounces does is try to differentiate temporary ndrs from permanent ones | 18:51 |
| @fungicide:matrix.org | discussions i'm finding about exim and verp are all specific to setting up verp handling in exim for mailman though, not fully independent very configuration with bounce tracking | 18:55 |
| @fungicide:matrix.org | s/very/verp/ | 18:55 |
| @fungicide:matrix.org | gerrit also triggers a lot of vacation-type autoresponders, where the problem is less easily solved because those are even less standardized than bounces | 18:57 |
Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!