| 16 Feb 2026 |
Picnoir | * Hey.
Substitution of big files is pretty slow today. The connection starts full speed but gets throttled to ~200kbps. It's not noticible on small files but very noticeable on big ones.
Is it a transit doing unhinged QoS between me and fastly (zayo -> LONAP IX) or did we change something in the fastly config?
| 17:43:04 |
hexa | nothing in the fastly config on our end changes for a long while | 17:44:07 |
hexa | well, that's a lie | 17:44:21 |
Picnoir | wget https://cache.nixos.org/nar/1sjixbc2hfa2h8q5phbalqcrf648k7c46sv4gb2nii46hpg1w23n.nar.xz <= is a good test | 17:44:28 |
hexa | 1sjixbc2hfa2h8q5phbalqcrf648k7c46sv4gb2nii46hpg1w23n.nar.xz 100%[====================================================================================================================================================================================================================================================>] 9.85M 29.9MB/s in 0.3s
| 17:44:46 |
Picnoir | Right, it's definitely on my side. Sorry for the false alarm. | 17:45:00 |
hexa | 1sjixbc2hfa2h8q5phbalqcrf648k7c46sv4gb2nii46hpg 100%[====================================================================================================>] 9.85M 46.4MB/s in 0.2s
| 17:45:17 |
hexa | we enabled h3 and tls1.3 recently | 17:46:40 |
hexa | and switched to LE certificates | 17:46:47 |
Picnoir | Today? | 17:46:56 |
hexa | but nix isn't using h3 yet iirc | 17:46:57 |
hexa | no, weeks ago | 17:47:01 |
hexa | early december | 17:47:12 |
Picnoir | Right, likely unrelated. It was working fine yesterday. | 17:47:18 |
Vladimír Čunát |
Right, it's definitely on my side. Sorry for the false alarm.
It might be a problem specific to your region. If a prefix is fast and then it slows down a lot, it might be just filling from S3 to Fastly being slow for some reason (at that location).
| 18:18:41 |
| 17 Feb 2026 |
| Sergei Zimmerman (xokdvium) joined the room. | 14:45:00 |
Sergei Zimmerman (xokdvium) | Something ate up all the space on builders? https://hydra.nixos.org/build/321929731/nixlog/1 | 14:46:09 |
hexa | plausible | 15:33:59 |
hexa | [https://grafana.nixos.org/d/rYdddlPWk/node-exporter-full?orgId=1&from=now-7d&to=now&timezone=browser&var-datasource=default&var-job=node&var-node=elated-minsky.builder.nixos.org:9100&var-diskdevices=%5Ba-z%5D%2B%7Cnvme%5B0-9%5D%2Bn%5B0-9%5D%2B%7Cmmcblk%5B0-9%5D%2B&refresh=1m&viewPanel=panel-152](disk usage) sometimes jumps and then boom | 15:35:26 |
hexa | disk usage sometimes jumps and then boom | 15:35:39 |
hexa | restarted | 15:37:36 |
Vladimír Čunát | The tmpfs space isn't that huge. | 17:51:54 |
Julien | Hello hello | 20:27:24 |
Julien | Would it cause any issue to the infra team if the nixos project took ownership of nixpkgs-update ? | 20:27:55 |
Julien | If I understand correctly, short term that would involve just transfering the server(s) to our hetzner account | 20:28:44 |
Julien | Maybe some context here: with potential unstability in the short-term future of nix-community, SC had been contacted to recommend nixpkgs-update migrates to the NixOS org. | 20:39:02 |
hexa | https://github.com/orgs/nix-community/discussions/2132 is the related discussion at nix-community | 23:16:44 |
hexa | We certainly need to make sure nix-update keeps working | 23:17:10 |
hexa | There is no Haskell knowledge in the infra team from what I know. | 23:17:34 |
hexa | But maintaining another machine should not be a problem | 23:18:28 |