0
0
plan 0040: decision 80; paused by the owner, the resume order
docs/plans/0040-release-pipeline.md+8−0
| @@ -759,6 +759,14 @@ All answered 2026-11-17. | ||
| ||
| ||
| ||
80. (2026-11-22, 19:55Z) The staging overlay stops depending on the public iroh relays (answer: "both relay now | ||
iroh.socialweb.computer use fqdn. direct in distinct plan doc for later"): our own iroh relay at | ||
`iroh.socialweb.computer` now (plan 0056), direct links between staging VMs in a distinct plan (0057). Measured | ||
that day: all 12 links of a staging network select the public n0 relay, never a direct path; under a bulk | ||
transfer it caps a VM at about 0.04 MB/s and that VM's overlay starves; roll three's `test` went red three times. | ||
The owner, same day: dumbpipe needs code work for a relay of ours ("copy dumbpipe source and modify ... or just | ||
build your own implementation off the iroh docs"): dumbpipe 0.40.1 hard-codes the n0 relays (plan 0056 has the | ||
options); and: finish the relay code work, then pause: no validation cycle until he restarts the lead. | ||
## Lead decisions awaiting the owner | ||
| ||
Taken by the lead while the owner was away; the owner may overrule each. Not owner decisions. | ||
.plans.yaml+1−1
| @@ -35,4 +35,4 @@ plans: | ||
log: "bash scripts/check-docs.sh" | ||
waits_on: "the two rolls" | ||
status: "steps 0-5 done (promote asked-only; promote.sh --dry-run rehearsed on the release image on main de0b91ee; hand builds to local/, rehearsed: publish+verify local/de0b91ee, ci-staging-binary local:); prod on rc174; no stable" | ||
next: "18:20Z 2026-11-22: roll three (main 0019f5f5): release green; test red three times, none traced to the tree: 80 s on the runner container's process limit (173 zombie ssh; runner restarted); 42 of 48 (two bidder.logLines calls over 120 s, then NoSubscriber); cancelled after 3 h at 6 PASS of 20 (the staging overlay over the public iroh relays, throttled: Findings). Promote never ran; prod on stable v0.0.1. No roll until the overlay is understood: a hand network measures it now. Ready, reviewed, not landed: log-filter 2f399ec8, watch-cancel 0e08da46; roll two tree staged (roll2 ee2c30ce). The relay the staging overlay uses is an owner decision. 0 green rolls in a row." | ||
next: "20:40Z 2026-11-22: PAUSED by the owner (he needs the machine; he restarts the lead tonight): no validation cycle, no roll, nothing deployed. Decision 80: our own iroh relay at iroh.socialweb.computer now (plan 0056), direct links later (plan 0057). Ready on branches, nothing landed: iroh-relay 90b4ace3 (plans 0056 and 0057, infra/iroh-relay with build.sh and deploy.sh (dry-run clean; binary built: ~/.tmp/iroh-relay/bin/iroh-relay sha256 26350dad...), the staging stand-up defaulting SOCIALWEB_MESHD_RELAY to our relay); log-filter 2f399ec8 and watch-cancel 0e08da46 (both reviewed safe); roll2 ee2c30ce (cli-help on 0019f5f5). RESUME, in order: 1. review the iroh-relay branch; 2. from its worktree: infra/iroh-relay/deploy.sh --dry-run, then --apply, then --status (rollback: --rollback --apply); read the runner container pids.current first; 3. a hand staging network: every link Relay(https://iroh.socialweb.computer/), bulk transfer at least 1 MB/s per VM, ping loss under 1%, save the meshd log; 4. one tree = main + iroh-relay + log-filter + watch-cancel, a full hand cycle in the default layout, 48 of 48; 5. one pull, the roll, read from inside with knotless run attach; 6. cli-help for the second roll. Known gaps: peer lookup still uses n0 DNS discovery (plan 0056 step 7: a meshd change); dumbpipe 0.40.1 hard-codes the n0 relays (guest ssh, ingress, the ssh gateway dialer: plan 0056 options b and c, owner to choose); the relay is open to anyone with the URL (a restriction is a proposed control). Open questions for the owner are listed in plan 0056. Prod on stable v0.0.1 (324f69a1), untouched; 0 green rolls in a row." | ||