The change we need has been released and propagated to nixos channels.
Change-Id: Ib10a1d42d7ef6deaf5665a13b72ece345e83d7dc
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11457
Reviewed-by: tazjin <tazjin@tvl.su>
Autosubmit: sterni <sternenseemann@systemli.org>
Tested-by: BuildkiteCI
This relates to nixery#167.
Using our GC module is much more reliable than what we were doing previously.
Change-Id: I1956457812a3a847a7c8a1f4e7e91e50fad08ac0
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11453
Tested-by: BuildkiteCI
Autosubmit: tazjin <tazjin@tvl.su>
Reviewed-by: flokli <flokli@flokli.de>
I accidentally deallocated the previous public IP, and had to make a
new one :(
Change-Id: Ie30305bdfdb8443e058270e5324baf555343441c
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11452
Tested-by: BuildkiteCI
Autosubmit: tazjin <tazjin@tvl.su>
Reviewed-by: flokli <flokli@flokli.de>
We just had a minor incident where apparently our build cache for the
critical security fix was deleted by automatic-gc (which I had stopped
manually) being reenabled by an unrelated whitby deploy.
This adds a new mechanism where by touching a file called
`/run/stop-automatic-gc` the GC can be prevented from running.
We might want to configure an occasional alert or something if this
file exists, so we don't forget about it when we are using it.
Change-Id: I041e57e24b2b684696164a2d516581d7f5696ef0
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11326
Tested-by: BuildkiteCI
Autosubmit: tazjin <tazjin@tvl.su>
Reviewed-by: flokli <flokli@flokli.de>
At least terraform wants all parameters passed via JSON to be strings.
It can't accept maps.
This means, allowing to pass a `argstr` JSON dict isn't really possible.
However, terraform is perfectly able to JSON-encode a map. So accept a
`argstr_json` argument instead, which `jq` will JSON-decode before
further processing it.
I dropped `argstr` support again so the jq expression still fits on my
screen, if anyone else (started) using this, I'm happy to review a CL
adding this. We should probably move the jq expression to some multiline
format then, though.
Change-Id: I5ab7a1169ab7305d3ab02db31c27732d9d1ab4e8
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11228
Reviewed-by: tazjin <tazjin@tvl.su>
Tested-by: BuildkiteCI
If this is set to true (and only then), also invoke `nix-build` on the
previously-instantiated .drv to cause builds/substitutions on the local
machine.
There's no terraform example for this in here, but this is useful if you
want to perform builds locally, for example to upload nix-built blobs
elsewhere through terraform.
Change-Id: Idcf7b8527aa9c27f6f9ca60ca607c29d82e1cce9
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11215
Reviewed-by: tazjin <tazjin@tvl.su>
Autosubmit: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
We already did all the instantiation, grabbing the calculated output
path too is cheap.
Change-Id: Id591865c65159409da739f706a9de29a9f50456a
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11214
Reviewed-by: tazjin <tazjin@tvl.su>
Tested-by: BuildkiteCI
Autosubmit: flokli <flokli@flokli.de>
This documents the input and output format, and also removes some
references to Terraform and evaluating NixOS system configurations.
It can be used to evaluate anything.
Change-Id: I8492cc3e386f89b299469c78e586644ee82a708f
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11213
Reviewed-by: tazjin <tazjin@tvl.su>
Autosubmit: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
In hope that iwlwifi works again on this commit, and I don't actually
have to debug it.
Includes following changes:
* users/aspen: home-manager is shuffling around pinentry options again
* users/flokli: rebase ipu6-softisp patches to Linux 6.8
make cl/11097 a separate patch
* ops/modules: remove unused (and now broken) v4l2loopback module
Co-Authored-By: Florian Klink <flokli@flokli.de>
Change-Id: I763f1f075778f2ed8db7803f87248c9dabde4213
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11174
Reviewed-by: tazjin <tazjin@tvl.su>
Reviewed-by: aspen <root@gws.fyi>
Reviewed-by: flokli <flokli@flokli.de>
Autosubmit: tazjin <tazjin@tvl.su>
Tested-by: BuildkiteCI
Resolves b/385. I have a feeling bufCheck should be moved to
//tools as well.
Change-Id: I2a2b63d135a2f8bcc982aa1138ff3213c6012f20
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11152
Autosubmit: sterni <sternenseemann@systemli.org>
Tested-by: BuildkiteCI
Reviewed-by: flokli <flokli@flokli.de>
This allows setting argstr to a map of keys and values.
We use jq to construct "--argstr k v" arguments, which are passed to
nix-instantiate.
Change-Id: I720a597ca2276364bc7005c156064d938f143041
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11141
Tested-by: BuildkiteCI
Autosubmit: flokli <flokli@flokli.de>
Reviewed-by: tazjin <tazjin@tvl.su>
Some of the docs are still outdated (like architecture and drv path
inconsistencies).
Change-Id: I7a6afceb008ef4cd19a764dd6c637b39fa842a2e
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11072
Autosubmit: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
Reviewed-by: edef <edef@edef.eu>
The other option is deprecated and prints scary eval warnings.
This probably has no effect, as the database is already initialised.
Change-Id: Ida4e79517436fa4572e69317b28f3712afc17159
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11029
Tested-by: BuildkiteCI
Reviewed-by: sterni <sternenseemann@systemli.org>
A recent change in nixpkgs introduced evaluation warnings if a systemd
service is configured to start after network-online.target, but does
not directly depend on it.
This is done because the existing dependency from multi-user.target to
network-online.target is being removed, leaving these services without
an actual dependency on the service.
This affected autosubmit (I added a weak dependency here, for now the
service is actually on the same host as Gerrit), and sterni's mirror
setup (I added a strong dependency here).
Change-Id: I88a4aa69f6788c489f59533d34be3c9cea681326
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11026
Autosubmit: tazjin <tazjin@tvl.su>
Reviewed-by: sterni <sternenseemann@systemli.org>
Tested-by: BuildkiteCI
The .dev TLS is on the HSTS preload list, so there's no need to set this
header here at all.
Change-Id: I253fa2427e75bd0808945cd5d53159cac74e7f8b
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11018
Autosubmit: flokli <flokli@flokli.de>
Reviewed-by: tazjin <tazjin@tvl.su>
Tested-by: BuildkiteCI
Make tvixbolt.tvl.su just serve a redirect to the new domain, and fold
everything into the tvix.dev.nix module.
Change-Id: I3a9ccf37d2ceee8886208d6f662e7598ce395b1a
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11015
Reviewed-by: lukegb <lukegb@tvl.fyi>
Reviewed-by: tazjin <tazjin@tvl.su>
Autosubmit: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
Assigning copyright to the TVL community (whatever that is), and
adding AGPL-3.0-or-later license.
I also cleaned up some of the stuff on the landing page.
Change-Id: I4dbca19406e00e5105fed50e8fb64e0fcca23e3a
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11013
Autosubmit: tazjin <tazjin@tvl.su>
Reviewed-by: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
Using an actual store path here means we have to copy all of depot into
the Nix store just to rebuild NixOS for a system - this is especially
painful if, like me, you have a rust target/ directory in depot which is
hundreds of GB - nix-build just OOMed on my system with 128GB of RAM!
There's no reason to use a store path here - we can just point to
wherever depot happened to be cloned.
Change-Id: Ibfd7181f0f75de077561db70ac8636389836980f
Reviewed-on: https://cl.tvl.fyi/c/depot/+/10997
Reviewed-by: sterni <sternenseemann@systemli.org>
Reviewed-by: tazjin <tazjin@tvl.su>
Autosubmit: aspen <root@gws.fyi>
Tested-by: BuildkiteCI
Keycloak has a new thing going on: Wildfly (whatever that is) is out,
and Quarkus (whatever that is) is in.
https://www.keycloak.org/migration/migrating-to-quarkus
This breaks our stuff, however, so we're using the Gerrit OAuth
plugins recommendations for how to work around that:
8d467e9945%5E%21/#F1
Change-Id: I2391a89c6791015e66c5e480b905b6ee56663020
Reviewed-on: https://cl.tvl.fyi/c/depot/+/10905
Tested-by: BuildkiteCI
Reviewed-by: flokli <flokli@flokli.de>
Midnight is kind of when you're still up and may want read performance
that is not affected by a btrfs scrub.
Change-Id: I0609269d3ee9853f7c7fe08cae18efe1d9259e7e
Reviewed-on: https://cl.tvl.fyi/c/depot/+/10864
Reviewed-by: sterni <sternenseemann@systemli.org>
Autosubmit: sterni <sternenseemann@systemli.org>
Tested-by: BuildkiteCI
Errors can only be wrapped in fmt.Errorf, in these two cases, we want to
print their string representation to stderr.
Change-Id: I65d345daacdd3960428ce82b5fdafceae61c6cc7
Reviewed-on: https://cl.tvl.fyi/c/depot/+/10527
Autosubmit: flokli <flokli@flokli.de>
Reviewed-by: tazjin <tazjin@tvl.su>
Tested-by: BuildkiteCI
This is the result of a `"reuse annotate --copyright "The TVL Authors"
--license MIT"` in that directory, making it conformant with the REUSE
Specification:
https://reuse.software/spec
Change-Id: I13e069b4621e8d5ccb7a09c12f772d70dea40a11
Reviewed-on: https://cl.tvl.fyi/c/depot/+/10170
Autosubmit: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
Reviewed-by: tazjin <tazjin@tvl.su>
Enables the new autosubmit bot, albeit without rebase
functionality (this will be a separate change).
Change-Id: Ia42a4f08c0edca5e6cc8bf4770ec24dbf16a5db7
Reviewed-on: https://cl.tvl.fyi/c/depot/+/10132
Reviewed-by: sterni <sternenseemann@systemli.org>
Tested-by: BuildkiteCI
Autosubmit: tazjin <tazjin@tvl.su>
Adds a small Rust program that connects to the Gerrit API and uses a
simple algorithm to figure out which changes should be submitted, and
submits them:
* it fetches all changes the Gerrit query API considers
submittable (i.e. all requirements fulfilled), and that have the
`Autosubmit` label set
* it filters these changes down to those that are _actually_
submittable (in Gerrit API terms: that have an active Submit button)
* it filters out those that would submit ancestors that are *not*
marked with the `Autosubmit` label
* it submits the longest chain
After that it just loops.
There is no rebasing logic yet for when it "runs out" of submittable
changes, but it will not be difficult to add.
Relates to b/333.
Change-Id: Ib91ecf2c45b178e8c64ff7b2174d617d4c45efe2
Reviewed-on: https://cl.tvl.fyi/c/depot/+/10131
Tested-by: BuildkiteCI
Reviewed-by: sterni <sternenseemann@systemli.org>
Reviewed-by: ezemtsov <eugene.zemtsov@gmail.com>
Autosubmit: tazjin <tazjin@tvl.su>
`lib.types.attrs` is deprecated in favor of `lib.types.attrsOf
lib.types.anything` because it doesn't merge attribute sets
/recursively/. `attrsOf` and `anything` do, the former is used to ensure
that the top value is an attribute set as expected by irccat.
Change-Id: I2a9d943a06c8f99f7d6d20c9944288e854924bff
Reviewed-on: https://cl.tvl.fyi/c/depot/+/10129
Reviewed-by: tazjin <tazjin@tvl.su>
Autosubmit: sterni <sternenseemann@systemli.org>
Tested-by: BuildkiteCI
While we are at it, rename disk-checkup.nix to btrfs-auto-scrub.nix and
move it into //ops/modules. I originally wanted to have additionally
disk health related services in that module, but the btrfs scrub
functionality is nicely self-contained and reusable, so I think it makes
sense to have this in a more central location.
Change-Id: Iabdd62838eef009540ca71abafd921afda2a9b47
Reviewed-on: https://cl.tvl.fyi/c/depot/+/10128
Reviewed-by: sterni <sternenseemann@systemli.org>
Autosubmit: sterni <sternenseemann@systemli.org>
Tested-by: BuildkiteCI
In some Gerrit version upgrade the syntax of this config element seems
to have changed. There's now one less level of escaping, and it no
longer produces raw HTML but rather a link.
Fixes b/319.
Change-Id: I8d86d23e91cb003e950d9a6723bb0a5ee5d80bb0
Reviewed-on: https://cl.tvl.fyi/c/depot/+/9952
Reviewed-by: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
