| 14 Jan 2025 |
 hexa | enhaaance! | 20:56:09 |
| 15 Jan 2025 |
 Jeremy Fleischman (jfly) | In reply to @joerg:thalheim.io
hexa (signing key rotation when): Jeremy Fleischman (jfly) We should probably do another Infra meeting tomorrow. DavHau was also interested in joining. Both Jeremy Fleischman (jfly) and DavHau are in an Asian timezone just now, so maybe we can meet earlier in the day? For me, it would probably be best if we could do a meeting at the normal time (6 pm CET) minus 3 or more hours, or plus 3 or more hours. | 01:27:00 |
| hexa | I'll be on a train most probably | 01:29:53 |
 Vladimír Čunát | (this Thursday is also bad for me) | 05:50:08 |
| Jeremy Fleischman (jfly) | FWIW, I am pretty flexible on the exact day of week | 08:16:52 |
 @hsngrmpf:matrix.org | Everyone can put their availability here: https://schej.it/e/c56BA | 09:35:33 |
 Mic92 | Ok. Added my availability to https://schej.it/e/c56BA | 12:45:33 |
|  SomeoneSerge (back on matrix) changed their display name from SomeoneSerge (utc+3) to SomeoneSerge. | 18:59:15 |
| 16 Jan 2025 |
|  j4m3s set their display name to j4m3s. | 10:28:02 |
| 17 Jan 2025 |
 Arian | GitHub just launched ARM runners | 10:56:08 |
| Arian | This means we can start looking moving image builds out of hydra | 10:56:21 |
| Jeremy Fleischman (jfly) | I'm having a lot of trouble reading this calendar, unfortunately. | 11:34:17 |
| Jeremy Fleischman (jfly) | 
Download image.png | 11:34:25 |
| Jeremy Fleischman (jfly) | There might be a good window of time tomorrow? Or perhaps it was earlier today? | 11:34:38 |
| hexa | -rw-r----- 1 nginx nginx 152M Jan 17 17:17 abuse.log
-rw-r----- 1 nginx nginx 116M Jan 17 17:17 access.log
| 17:17:33 |
| hexa | so we're filtering like two thirds of requests to h.n.o | 17:18:02 |
 Rick (Mindavi) | Not bad | 17:25:22 |
| Arian | Nevermind. no virtualization on arm runners :( | 17:44:04 |
| Arian | So unless we change our image builds to not rely on nested virt we're out of luck | 17:44:23 |
| 18 Jan 2025 |
| Vladimír Čunát | sleepy-brown: now I simply changed its --jobs from 8 to 4. I needed to restart at least a dozen kernel builds in the past few days, always out-of-space on /tmp. | 06:44:44 |
| Vladimír Čunát | For a longterm solution, I'm not convinced that /tmp on tmpfs is worth the trouble. | 06:45:35 |
| hexa | increased tmpfs to 160GB 14 hours ago | 13:47:49 |
 @emma:rory.gay | what even is the point of /tmp on tmpfs in this case? reserving disk space for built derivations? | 23:34:32 |
| hexa | higher throughput because we are building in memory | 23:35:31 |
| @emma:rory.gay | ah, i guess that makes sense | 23:35:52 |
| hexa | less wear on the disks | 23:35:53 |
| @emma:rory.gay | im guessing that comes with the downside of being more expensive, due to memory usage? | 23:36:32 |
| hexa | increased to 6, because 40G per job on average is hopefully fine | 23:36:50 |
| @emma:rory.gay | * im guessing that comes with the downside of being more expensive, due to memory usage?
aka needing nodes with more ram | 23:36:54 |
| hexa | big compute generally comes with sufficient ram | 23:37:13 |
