[06:02:45] topranks: https://labs.ripe.net/author/suzanne_taylor_muzzin/ripe-ncc-internet-country-report-mediterranean-europe/ I didn't read it yet but looks interesting [08:15:30] 10netops, 10Infrastructure-Foundations, 10SRE, 10Patch-For-Review: Adjust egress buffer allocations on ToR switches - https://phabricator.wikimedia.org/T284592 (10Marostegui) @cmooney I haven't been able to get ahold of you this week, so leaving the comment I left on IRC here: My preferred order for the sw... [10:34:46] 10netops, 10Infrastructure-Foundations, 10SRE, 10Patch-For-Review: Adjust egress buffer allocations on ToR switches - https://phabricator.wikimedia.org/T284592 (10cmooney) @Marostegui Ok thanks for the comments. I've not been feeling so good so hadn't been online. Will review Monday against feedback from... [10:37:18] XioNox: Nice, very topical for us I'll have a look :) [11:36:58] 10netops, 10Infrastructure-Foundations, 10SRE, 10Patch-For-Review: Adjust egress buffer allocations on ToR switches - https://phabricator.wikimedia.org/T284592 (10Marostegui) I hope you get better soon. I am off next week but someone from the team will contact you next week. From my point of view, I think... [13:48:12] TIL: pchar [13:53:03] (http://www.kitchenlab.org/www/bmah/Software/pchar/) [13:54:51] Interesting, just looking at the manpage there... [13:56:08] I'm wondering about the throughput measures.... ICMP TTL exceeded responses missed from routers isn't really a good way to estimate forwarding capacity at those hops I'd have thought. [13:56:14] Assuming that's how it works? [13:56:51] have nto dug into how it works and the talk that mentioned it (which its self was 6 years old), said it was a bit buggy [13:57:04] and needed som love [13:58:04] cool, interesting nonetheless. If you have a link to that talk by all means send it on [13:59:07] talk is more about linux performance debugging/tracing, this tool was just mentioned in passing. https://www.youtube.com/watch?v=FJW8nGV4jxY [13:59:48] i suspect a lot of the stuff covered in the talk may be better servered by eBPF (which is what i was first looking into before going down a rabbit whole) [14:00:44] Yeah eBPF seems to provide loads of scope for various types of tracing I'd wager you're correct on that. [14:00:48] thanks for the link :) [14:01:05] no probs, you should go back to enjoying your vacation ;) [14:02:20] what could be more enjoyable than a talk on Linux performance tooling :D [14:02:27] :D hahaha [15:13:07] BTW we've done some toying with BPF tracing here before -- the bpfcc packages with tools are available to be installed (although I think not installed by default as they need kernel debugging symbols) [15:13:19] check out https://phabricator.wikimedia.org/T265625#6719927 and https://phabricator.wikimedia.org/T260281#6381055 if you're curious [15:16:53] thanks cdanis [15:24:47] I also did some experimentation with my workstation on https://github.com/cloudflare/ebpf_exporter but didn't get as far as producing packages suitable for production [15:24:53] it is a little buggy, but interesting [15:29:11] yes that looks quite nice, in particular the inline C in yaml :P [15:31:59] jbond: I think you'll find it is well in-line with existing practice here https://gerrit.wikimedia.org/r/plugins/gitiles/operations/puppet/+/refs/heads/production/modules/varnish/templates/text-frontend.inc.vcl.erb#34 [15:34:27] lol very true