You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
fix(up): start shared-only (hub) workspaces and self-heal the network (#91)
Two drift bugs made `devstack up` a silent no-op that started nothing —
the exact symptom hit on a `devstack init` starter workspace (shared
pg/redis/minio, no projects):
1. Selective-up derived the shared set only from projects' `uses`
(profile.sharedUsedBy). A hub / shared-only workspace has no project
referencing its shared services, so `active.Shared` was empty and the
shared phase brought up nothing. Now a project-less workspace that
declares `shared:` brings up its full declared shared stack.
2. The network phase had a constant fingerprint, so once satisfied it was
skipped forever — even after the external `devstack_shared` network was
removed out-of-band (docker network prune, a Docker Desktop / WSL
restart). Every subsequent compose-up then failed with "network
devstack_shared declared as external, but could not be found". Marked
the phase AlwaysRun: EnsureNetwork is a cheap create-if-missing under
the lock, so re-running every time is correct and self-healing.
Tests: a hub workspace brings up all declared shared services (and no
compose-up phase); the network re-ensures after out-of-band removal; the
happy-path re-run now expects network to re-run (AlwaysRun) rather than skip.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
0 commit comments