Migrate CI to Solstice CI #8

New issue

Open

opened 2026-02-07 14:11:08 +00:00 by Toasterson · 2 comments

Toasterson commented

2026-02-07 14:11:08 +00:00

(Migrated from codeberg.org)

Since we are now on Codeberg we need to migrate to a capable testing CI.

I have developed solstice CI which tests inside VM's using cloud-init to provision them. So we can easily buld and test on a large set of operating systems.

Since we are now on Codeberg we need to migrate to a capable testing CI. I have developed solstice CI which tests inside VM's using cloud-init to provision them. So we can easily buld and test on a large set of operating systems.

toasterson referenced this issue

2026-09-19 19:59:07 +00:00

Research: does the rustls/aws-lc-sys crypto stack build on illumos, or should the depot use ring? #35

toasterson commented

2026-09-19 20:01:05 +00:00

Owner

CI now runs on Solstice CI via Forgejo Actions: .forgejo/workflows/ci.yml (jobs linux on the rust image and illumos on the OpenIndiana Rust image, sccache-backed) and .forgejo/workflows/release.yml for v* tags. The GitHub workflows and the legacy .solstice/workflow.kdl were removed in f634ece. Leaving this open until both jobs are green on main; clippy is report-only for now (see the clippy-clean ticket).

CI now runs on Solstice CI via Forgejo Actions: `.forgejo/workflows/ci.yml` (jobs `linux` on the `rust` image and `illumos` on the OpenIndiana Rust image, sccache-backed) and `.forgejo/workflows/release.yml` for `v*` tags. The GitHub workflows and the legacy `.solstice/workflow.kdl` were removed in f634ece. Leaving this open until both jobs are green on `main`; clippy is report-only for now (see the clippy-clean ticket).

toasterson referenced this issue

2026-09-19 20:11:35 +00:00

Research: does the rustls/aws-lc-sys crypto stack build on illumos, or should the depot use ring? #35

toasterson referenced this issue from CloudNebulaProject/solstice-ci

2026-09-19 21:29:34 +00:00

runner-integration: the same Forgejo job is executed by several runners concurrently (duplicate dispatch) #8

toasterson commented

2026-09-19 21:29:38 +00:00

Owner

Status after the first runs on Solstice:

.forgejo/workflows/ci.yml is picked up and both images boot: the rust VM (rustc 1.98.1, sccache 0.8.2 injected by the runner) and the illumos VM (OpenIndiana Hipster 2026.04, rustc 1.96.1, sccache 0.10.0). nextest installs from get.nexte.st on both. cargo fmt --check passes.
illumos (run 3): the whole workspace compiles, including aws-lc-sys through cmake without bindgen (confirms wayfinder #35), but linking pkg6depotd's test binaries fails with a bare collect2: error: ld returned 1 exit status and no linker message. Commit 35d807b turns off incremental builds, uses line-tables-only debuginfo, builds bins before test targets and dumps df/swap/zfs on failure to pin this down.
Infrastructure: every job is currently being executed by several runner instances at once, so jobs sit in "Set up job" for 15-35 min and get marked failed without a failing step. Filed as CloudNebulaProject/solstice-ci#8. Until that is fixed the results here are unreliable; keeping this ticket open.

Status after the first runs on Solstice: - `.forgejo/workflows/ci.yml` is picked up and both images boot: the `rust` VM (rustc 1.98.1, sccache 0.8.2 injected by the runner) and the `illumos` VM (OpenIndiana Hipster 2026.04, rustc 1.96.1, sccache 0.10.0). nextest installs from get.nexte.st on both. `cargo fmt --check` passes. - illumos (run 3): the whole workspace compiles, including aws-lc-sys through cmake without bindgen (confirms wayfinder #35), but linking `pkg6depotd`'s test binaries fails with a bare `collect2: error: ld returned 1 exit status` and no linker message. Commit 35d807b turns off incremental builds, uses line-tables-only debuginfo, builds bins before test targets and dumps df/swap/zfs on failure to pin this down. - Infrastructure: every job is currently being executed by several runner instances at once, so jobs sit in "Set up job" for 15-35 min and get marked failed without a failing step. Filed as CloudNebulaProject/solstice-ci#8. Until that is fixed the results here are unreliable; keeping this ticket open.

No milestone

No project

No assignees

2 participants

Notifications

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

OpenIndiana/ips#8

No description provided.

Rows
Columns