| 4 Aug 2021 |
 sterni | do dependency logs count towards that if they aren't scheduled separately? | 15:36:12 |
| sterni | I'm a bit confused why gcc would all of a sudden have a lot more logs for the avr target | 15:36:34 |
| sterni | the failure began when avr gcc was upgraded from 9 to 10 though | 15:37:02 |
 Vladimír Čunát | 64 MiB limit sounds high enough to me. | 15:41:01 |
| Vladimír Čunát | * 64 MiB limit sounds high enough to me. (even though logs will typically compress a lot) | 15:41:38 |
| 5 Aug 2021 |
 @grahamc:nixos.org | I think I wouldn't mind increasing the log limit | 14:16:22 |
| @grahamc:nixos.org | 64MiB is a lot though :) | 14:17:07 |
| @grahamc:nixos.org | either here: https://github.com/NixOS/hydra/blob/4e9455160270b9b933582501a47430690d66e5b4/src/hydra-queue-runner/hydra-queue-runner.cc#L45 or as a configuration setting in the nixos-org-configurations repo | 14:18:29 |
| Vladimír Čunát | It would make sense to me to allow increasing it in relatively important derivations where we see no easy way of reducing the logs. That way the overall cost increase should be negligible. | 15:17:47 |
| Vladimír Čunát | * It would make sense to me to allow increasing it just in relatively important derivations where we see no easy way of reducing the logs. That way the overall cost increase should be negligible. | 15:17:58 |
| Vladimír Čunát | * It would make sense to me to allow increasing it just in relatively important derivations where we see no easy way of reducing the logs. (e.g. like we have for the timeouts) That way the overall cost increase should be negligible. | 15:18:16 |
| sterni | tbh I half expected there to be a meta setting for it already :) | 15:53:04 |
| 9 Aug 2021 |
 hexa | https://github.com/NixOS/nixpkgs/pull/133289 | 20:53:51 |
| hexa | author asks for someone with hydra knowledge to check their assumptions | 20:54:06 |
| 12 Aug 2021 |
| hexa | I have no particular love for wendy, but ongoing SIGILL on packages that are costly to build yourself (pytorch, tensorflow, numpy) are less than ideal | 11:25:21 |
| hexa | https://hydra.nixos.org/build/149565546 | 11:25:30 |
| hexa | * I have no particular love for wendy and the ongoing SIGILL problems on packages that are costly to build yourself (pytorch, tensorflow, numpy) are less than ideal | 11:25:55 |
| Vladimír Čunát | Well, there are two sides. Producing packages that are only usable on new CPUs isn't ideal by itself... but that's perhaps something not well supported upstream. Old-style packages always compiled multiple variants where it mattered and chose during startup. | 11:38:36 |
| Vladimír Čunát | * Well, there are two sides. Producing packages that are only usable on new CPUs isn't ideal by itself... but that's maybe something not well supported upstream (for pytorch and some others). Old-style packages always compiled multiple variants where it mattered and chose during startup. | 11:38:59 |
| Vladimír Čunát | Still, wendy's age has many disadvantages. Poor performance (of each single thread) and power efficiency come to mind. | 11:40:52 |
| hexa | surprising people with tensorflow/pytorch builds on stable upgrades sadly isn't a great experience | 11:44:32 |
| hexa | I do acknowledge that we would loose a bunch of compute, that somewhat diversifies our builder situation (eqx metal + X) | 11:46:40 |
| Vladimír Čunát | Diversity: I've been considering to host some better physical machine than the current t4a, perhaps as a replacement of t4a. | 11:48:37 |
| Vladimír Čunát | If there's interest, that is. I have no idea of any existing plans around this. | 11:50:08 |
| Vladimír Čunát | * Diversity: I've been considering to host some better physical machine than the current t4b, perhaps as a replacement of t4b. | 12:58:56 |
 Domen Kožar | hexa: we wouldn't lose that much with wendy gone | 13:47:52 |
| Domen Kožar | it probably just needs to be done | 13:48:06 |
 andi- | It would still be nice to think about diversifying the builders. If packet pulls the plug (for whatever reason) we have a problem. | 13:49:25 |
| sterni | indeed | 13:50:52 |
| Vladimír Čunát | I think ike is also from the old set of machines. | 14:01:31 |
