Troubleshooting

organism: not running — reading the Journal in status. No host is up here. status, review <id>, history, board, fleet and ui work from the record; ask and a verdict go into the record when the repository has a remote (an organization elsewhere answers), and need hale dna run or dev in another terminal when it doesn’t.

the organism is not running here (no membrane) and the repository has no remote to reach one through. Same thing, with nowhere to send it. Start a host, or add a remote and start one somewhere.

membrane client failed: … File name too long. The sockets under .hale/dna/ are unix sockets, and a path over about a hundred bytes cannot be bound. Move the checkout somewhere shorter.

task t1 born … [failed], and history t1 says credential not present. The editor’s hosted model has no key. Set OPENAI_API_KEY, or point the router at a local or scripted model — Shaping it.

needs leader, and nothing happens. The Leader decides with a model; with no key it cannot, and the review waits. Answer it as the Board (hale dna review m1 approve --as you), or give the Leader a model.

needs board on every review. Requests through ask are application changes, and the default grant is refactor docs. It’s the expected posture for a new organization: you are asked, and told why. Widen classes in dna/org/main.hl when you want the Leader’s judgement on them.

disposition stage. Within the grant, but the evidence doesn’t reach the bar — usually a first change in a fresh history, or a program with no tests. Still a normal review; the bar rises and falls with the evidence.

review m1 refused the verdict: digest mismatch. The verdict named a candidate other than the one under review — the sandbox changed, or a --digest was stale. Re-render with hale dna review m1 and answer again; the digest in the decide: line is the one to name.

authority reviewer does not satisfy board. The verdict’s --authority doesn’t meet what the review requires. Board-class changes, and anything touching law, effects or the fleet’s shape, need the Board.

reviewer editor authored the candidate. The name on the verdict is the author’s. Use your own.

mutation.refused: candidate moved after review in the history. The sandbox’s commit changed between the review and the apply. Nothing was applied. Ask again; the new candidate gets its own review.

m1 [rolled_back]. Either an expression exited inside the observation window (history m1 shows expression.crashed with the exit code — and, on a fleet, the instance and node), or you rejected a post-review change. The repository is back at the commit the change started from, and the old expression is running.

never expressed by api-1. A touched instance never reported instance.up at the revision within the settle time. Look at that node’s terminal: the revision may not have fetched, or the seed may not build there.

no [dna] fleet = "<name>" in hale.toml. hale dna fleet, deploy and rollback need to know which declared fleet the DNA expresses. Add the section — Operating the fleet.

candidate breaks the fleet. The candidate’s services compose, but a claim the plan makes over them fails. The witness is in the fleet receipt of the denied mutation.

journal: … chain BROKEN. refs/dna/journal has more or fewer commits than rows. Someone rewrote the branch. git reflog refs/dna/journal finds the old head; hale dna sync from a clone that has it restores the rest.

Two organizations on one record. One hale dna run per repository. Two hosts syncing one remote will both relay and both answer.

Where things are. refs/dna/journal (the record), refs/dna/receipts/<sha256> (receipts), refs/dna/lease/*, refs/dna/revisions/<rev> (what nodes fetch); .hale/dna/ (sockets, sandboxes, scratch, status.json); .hale/node/<name>/ on a node (pid files, artifacts).

Starting over. Delete the refs and the record is gone; git log keeps every change it applied:

git for-each-ref --format='%(refname)' refs/dna | xargs -n1 git update-ref -d
hale dna init .        # seeds a fresh record from the program as it is now