| 25 Oct 2021 |
 Ryan Burns | alacritty succeeds for me on x86_64-darwin 🤷 | 09:19:02 |
| Ryan Burns | id be willing to bisect jx down to the individual coreutils commit | 09:23:31 |
| Ryan Burns | next time i feel like heating up my room | 09:23:39 |
 das_j | I can borrow you some build resources if you are interested | 09:24:33 |
| das_j | Ah that will probably not work because it has to be built on ZFS… | 09:24:49 |
 trofi | master->staging merge fails again: https://dpaste.com/DQMXWX5K7.txt. Looks like due to sha256 format difference. | 18:31:33 |
 Sandro | In reply to @trofi:matrix.org
master->staging merge fails again: https://dpaste.com/DQMXWX5K7.txt. Looks like due to sha256 format difference. and an update | 18:52:04 |
| Sandro | In reply to @trofi:matrix.org
master->staging merge fails again: https://dpaste.com/DQMXWX5K7.txt. Looks like due to sha256 format difference. * and an upgrade for a python package | 18:52:18 |
 hexa | happens, did the merge a while ago | 19:42:39 |
| das_j | Vladimír Čunát so is jx a problem or is it not? I really cannot tell anymore | 20:34:40 |
 Vladimír Čunát | I wasn't looking into jx. | 20:35:01 |
| das_j | I also cannot reproduce it | 20:35:01 |
| das_j | Oh | 20:35:09 |
| das_j |
It truly seems that the issue disappears with reverting the coreutils update #139541. (EDIT: 2/2 attempts succeeded)
| 20:36:09 |
| das_j | Does this mean that 2 attempts succeeded while the PR was reverted? | 20:36:28 |
| Vladimír Čunát | That's peertube, not jx. | 20:37:35 |
| das_j | In reply to @vcunat:matrix.org
That's peertube, not jx. Yeah, sorry. That was an unrelated question | 20:37:51 |
| Vladimír Čunát | Yes, I tested staging-next build (2/2 failures) and staging-next with reverted coreutils (4/4 successes now). | 20:38:18 |
| Vladimír Čunát | The retries were just for the final builds, naturally. | 20:38:32 |
| Vladimír Čunát | * The retries were just for the final build, naturally. | 20:38:38 |
| das_j | What fs do you use on /nix and /tmp? | 20:38:47 |
| Vladimír Čunát | ext4 both, in this case. | 20:40:15 |
| das_j | 🤔 I was hoping this is just a zfs issue | 20:40:33 |
| das_j | It's weird because I cannot reproduce it on xfs with a 5.10 kernel and nix 2.3 | 20:41:14 |
| Vladimír Čunát | This is Ubuntu 18.04, kernel 5.4... in case it matters. | 20:44:10 |
| Vladimír Čunát | Well, I could nix-copy-closure and retry the final build on a different system... e.g. NixOS with 5.10 kernel. | 20:44:53 |
| das_j | In reply to @vcunat:matrix.org
This is Ubuntu 18.04, kernel 5.4... in case it matters. I have no idea what matters 😢 | 20:45:05 |
| Vladimír Čunát | And I don't know in which step these parameters matter. | 20:45:13 |
| Vladimír Čunát | If it's just the last step, perhaps bisecting inside coreutils could be viable. | 20:45:50 |
| das_j | I'd love to do so, I could easily do that on our build machines but I haven't been able to reproduce it. That's probably the downside of having the same environment everywhere | 20:47:52 |
