LuxorLabs/stickydisk@v1 mounts a workspace-scoped cache path on a compatible Tenki runner.
Your workspace must have runner sticky disks enabled by a Tenki operator.
The runner supplies a session capability through its environment.
The action does not accept storage names, pool identifiers, or other storage access details.
- uses: LuxorLabs/stickydisk@v1
with:
key: npm-cache
path: ~/.npmcommit defaults to auto and asks the control plane to publish after a successful job.
Set it to never when a job should read the published cache without publishing its own writes.
Failed requests to the cache service are retried with exponential backoff.
retries sets how many retries follow the first attempt (default 2), retry-delay the wait before the first retry (default 150ms, doubling each time), retry-max-delay the longest wait (default 5s), and attempt-timeout the longest a single request may take (default 35s for the mount, 5s for finalization).
When the retries run out, or the service reports degraded storage, the job continues uncached with source=fallback unless fail-on-error is true.
An unconfirmed mount is never finalized, so the service cannot publish it.
- uses: LuxorLabs/stickydisk@v1
with:
key: npm-cache
path: ~/.npm
retries: 1
attempt-timeout: 4sStorage loss after attachment can cancel the owning job to prevent publication of damaged cache data.
The action writes mounted, source, and generation outputs.
Its job summary includes the mount mode, generation, fallback reason, previous candidate outcome, and publication state.
Use LuxorLabs/stickydisk@v1 in workflows to follow version 1 releases.
Release tags such as v1.0.1 are immutable; the v1 tag advances with each version 1 release.
Pin an immutable release tag or commit SHA when you need a fixed version.