Skip to content

release: merge v1.8.1 to main#627

Merged
VijitSingh97 merged 4 commits into
mainfrom
develop
Jul 18, 2026
Merged

release: merge v1.8.1 to main#627
VijitSingh97 merged 4 commits into
mainfrom
develop

Conversation

@VijitSingh97

Copy link
Copy Markdown
Collaborator

Real-merge of develop (v1.8.1, tag v1.8.1 @ ac9b723) into main. Brings the four commits since v1.8.0: doctor check_tor_running refactor (#621), back-merge (#624), the one-click 502 poller fix (#625/#622), and the v1.8.1 prep (#626). develop HEAD is the released commit, so main lands exactly on v1.8.1. Back-merge main→develop follows.

🤖 Generated with Claude Code

VijitSingh97 and others added 4 commits July 17, 2026 19:14
…y outage (#621)

Follows up #619. That fix made BOTH egress checks fail when tor was down
while mining ran — two FAIL lines for one root cause. Replace that with a
single dedicated check_tor_running() that owns the "tor down while the
stack runs" verdict (OK when up, FAIL when down+mining, info-skip on a
clean down), called first in doctor's Network section; the two egress
checks revert to a plain info-skip when tor is down. One loud line, not
two. doctor still exits non-zero on the outage (the #618 fault_tor_down
assertion holds). tor is unconditional (no compose profile gates it), so
this never false-fires on a legitimate tor-absent deployment.

Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
chore: back-merge main into develop after v1.8.0
…625)

The self-upgrade recreates the dashboard container itself. While it restarts,
caddy (the reverse proxy) stays up and answers 502/503/504 — the upstream is
briefly gone, not failed. pollResult already rode out a dropped connection
(proxy down) but treated a gateway 5xx (proxy up, no upstream) as terminal, so
the modal jumped to "failed / HTTP 502" even though the upgrade landed and the
durable control result said "upgraded".

Ride out 502/503/504 like a dropped connection; the durable control result is
the real outcome and the poll cap is the backstop. A genuine backend 500
(upstream up, erroring) still fast-fails. Regression test: a 502 for the first
N polls then the terminal "upgraded" result must resolve to done, not failed.

Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Patch release over v1.8.0: the one-click upgrade poller now rides out a
502/503/504 from the reverse proxy during the dashboard's own restart, so a
successful upgrade no longer surfaces a false "HTTP 502 — did not complete"
modal (#622). VERSION + dashboard pyproject/uv.lock bumped in lockstep.

Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
@VijitSingh97
VijitSingh97 merged commit f781310 into main Jul 18, 2026
32 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant