| 20 Aug 2025 |
 Winter | why do only some Hydra builds show the "build dependencies" tab? | 17:13:11 |
 dgrig | I'm following the breadcrumbs here, might be completely off, but:
[% IF drvAvailable %]
[% INCLUDE makeLazyTab tabName="tabs-build-deps" uri=c.uri_for('/build' build.id 'build-deps') callback="makeTreeCollapsible" %]
[% END %]
&&
[% IF drvAvailable %]<li class="nav-item"><a class="nav-link" href="#tabs-build-deps" data-toggle="tab">Build Dependencies</a></li>[% END %]
in https://github.com/NixOS/hydra/blob/0d2a030661fb1a6ba3f5cb83c627a72b562ebe74/src/root/build.tt#L469-L471
While drvAvailable is defined as:
$c->stash->{drvAvailable} = $MACHINE_LOCAL_STORE->isValidPath($build->drvpath);
So, since MACHINE_LOCAL_STORE seems to be local indeed, this tab will only exist if the derivation is available on mimas?
| 17:42:27 |
| 21 Aug 2025 |
|  sinan changed their profile picture. | 02:27:02 |
| sinan | 02:27:05 |
 Tristan Ross | HW team is likely going to be able to get remote hardware which could be usable for RISC-V support. What sort of requirements are necessary in terms of access, performance, storage, etc? | 15:37:34 |
 K900 | How much hardware? | 15:40:50 |
| K900 | If we're talking Hydra support, the answer is most likely just "no" | 15:41:05 |
| Tristan Ross | We're figuring that out, we're needing to know what sort of requirements for infra there is for builders. | 15:43:28 |
| K900 | I am going to be real, it'll be a hard sell when there's no practical usage for RISC-V and no cores that are even Ampere ast | 15:44:08 |
| K900 | * I am going to be real, it'll be a hard sell when there's no practical usage for RISC-V and no cores that are even Ampere fast | 15:44:09 |
| K900 | And we can't fan out early bootstrap | 15:44:32 |
| K900 | I'd expect like, 2-3 days overhead per staging cycle from this, which is BAD | 15:44:47 |
| K900 | If my estimates of single machine RISC-V perf are anywhere close to reality | 15:45:06 |
 Alyssa Ross | Couldn't we just not block staging on RISC-V? | 15:45:28 |
| K900 | We could run a separate jobset, but then the same thing could be done outside of official ifnra | 15:45:57 |
| K900 | * We could run a separate jobset, but then the same thing could be done outside of official infra | 15:45:58 |
| K900 | (and probably should be done outside official infra, at least initially) | 15:46:06 |
 vcunat | Trust is needed. People who have physical access to the HW can more-or-less put anything into the cache. | 15:53:05 |
| vcunat | (though not trivially at least) | 15:53:34 |
| Tristan Ross | If the infra team doesn't have physical access but the hardware team does, is that acceptable for trust? | 15:53:55 |
| K900 | Would not be great | 15:54:38 |
| K900 | But also, again, I want to first see proof that there's enough hardware | 15:54:38 |
| K900 | And that there's enough demand to justify the Hydra costs | 15:54:43 |
| Tristan Ross | What sort of costs does it make on Hydra? Is it just eval and storage? | 15:55:56 |
| K900 | Queue runner is slow | 15:56:35 |
 emily | cache for the entire jobset on a new platform is substantial (since we pay the whole cost every staging cycle) | 15:56:40 |
| K900 | Queue runner slowness scales with number of jobs | 15:56:44 |
| emily | and Hydra scheduling is a bottleneck | 15:56:53 |
| K900 | Storage is also still questionable | 15:56:59 |
| K900 | Eval is largely not an issue | 15:57:06 |
