Both container images published manually (CI on indri couldn't complete
due to host RAM pressure — 7.2 GB in macOS memory compressor causing
~10–100x slowdown on memory-touching work, see separate ticket thread):
- registry.ops.eblu.me/blumeops/alloy:v1.16.0-26a3ab5
(dagger build, run locally on gilbert via container.py)
- registry.ops.eblu.me/blumeops/alloy:v1.16.0-26a3ab5-nix
(nix build, run on ringtail's nix-container-builder)
Both images built from the alloy-v1.16.0 branch tip 26a3ab5b. After
PR squash-merge, follow up with a C0 to rebuild from the main SHA and
bump tags to a [main]-marked variant per the squash-merge convention.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Local nix-build on ringtail (after the go_1_26 + GOTOOLCHAIN=local fix)
surfaced the new module-cache hash:
specified: sha256-rD7zqomSVv4d8NaC7jXXgihuQvK8guaAN0KrsBRWMVQ=
got: sha256-9/v85HyDInJB+9qHauKVuDol6Yf5mkXfMWgCr7RdRTk=
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
CI run 522 dagger job failed with `go.mod requires go >= 1.26.2 (running
go 1.25.9; GOTOOLCHAIN=local)`. golang:1.25-alpine3.22 was a holdover
from the original Dockerfile; alloy v1.16.0's go.mod requires Go 1.26.
Switching to golang:alpine3.23 (currently Go 1.26.2) — matches both
alloy's toolchain requirement and the go_build helper's image choice.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Bump alloy v1.14.0 → v1.16.0 across both build paths:
- containers/alloy/Dockerfile replaced with native container.py
(3-stage Node UI → make alloy Go build → Alpine runtime).
- containers/alloy/default.nix bumped: src hash + npmDepsHash refreshed.
goModules outputHash still TODO — will be updated after a build attempt
on the ringtail nix runner reveals the new hash.
The container.py uses the upstream Makefile via `make alloy` (not plain
`go build`) so version stamping, release flags, and the netgo+embedalloyui
tags match upstream releases. promtail_journal_enabled is omitted as
before — k8s deployments read pod logs from the filesystem, not journald.
Per [[review-services]] for the four stale Alloy entries (alloy-k8s,
alloy-ringtail, alloy-tracing-ringtail, alloy ansible — all reviewed
2026-03-13).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
