Pin PTY sessions to the peer they were opened on so shells survive
active-node switches until the window is closed. Electron host owns
the connection; pop-out windows relay I/O via IPC (BroadcastChannel
fallback). Explicit terminal sessionIds no longer wipe sibling PTYs
on the same container.
Disable select/delete for images with container usage, and broadcast image lists with usage attached so the Unused tab no longer flashes the full inventory mid-removal.
Leave kills are fire-and-forget (no longer block the next start)
startTerminal still clears leftover PTYs; unique session ids still protect against late kills
Shorter layout wait (12 frames) with an 80×24 fallback so fit doesn’t stall the open
Terminal
Invalidate local UI immediately on leave (bump gen, dispose xterm, clear DOM)
Show a “Connecting…” placeholder on re-entry
Skip stale queued inits via a desire-generation token
Unique PTY sessionIds so a late kill can’t destroy a new session
Server killTerminal: if a specific sessionId is missing, don’t fall through and kill the active PTY
All live tabs
Enter handlers only run when newly entering (no double-init)
Leave handlers for logs / terminal / stats / processes
Processes: ignore late replies; reset on container switch
Stats: don’t create charts on a hidden 0×0 canvas
Inspect: ignore stale replies; don’t wipe a newer container’s callback
Processes shortcuts use showContainerDetails(container, { tab: 'processes-tab' }) instead of click + double-fetch
Switching Terminal ↔ Logs ↔ Stats ↔ Processes repeatedly should stay stable; Terminal should recover cleanly instead of hanging.
Squash a container into a single-layer image from Manage → Flatten, with
repo/tag, pause, config re-apply, vault credentials, and optional registry push.
Server-side alerting for Docker/container/stack health with Discord,
Slack, Teams, ntfy, Gotify, Telegram, and generic webhooks. Settings
tab, client cache, and backup/restore coverage; marked experimental.
Store search/status/sort under settings.containerFilters in
~/.config/peardock/cache/settings.json so preferences survive restarts,
and migrate the old localStorage key once.
When duplicating with the same host ports, stop+rename the origin, deploy
and verify the new container, then remove the parked source. Client
precheck ignores ports owned only by the origin to avoid false blocks.
Stop overwriting the header on every parallel restore connect; wait for
the last-used peer before falling back; avoid persisting temporary
active selection so the remembered server stays correct across restarts.
Fill the details terminal host for FitAddon, wait for a measurable box
before starting the PTY, re-apply xterm palettes after open, and re-fit
on UI color-mode changes so light sessions paint without a dark/light flip.
Store theme, density, and other prefs in a versioned JSON cache file
(matching peers.json) instead of localStorage-only, with migration and
a localStorage mirror for FOUC and legacy readers.
Expand light-mode CSS for dashboard, details, terminals, job/pull progress, restoring boot card, and settings chrome. Add a Jobs tray settings tab for dismiss timing and reconnect feedback volume; auto terminal themes follow UI color mode when saved.
Launch finishes as soon as any peer comes online; offline dials notify the tray and keep retrying in the background. Fleet cards that hit max reconnect tries show a clear Reconnect button that resets the retry budget.
Fleet builds a roster from configured peers so hosts stay visible when
they drop (online / reconnecting / failed / offline). Settings → Behavior
adds max reconnect tries (1–8 or unlimited, default 8) wired into
ConnectionManager, with Retry connect to redial after the budget is spent.
Dead links now force disconnect after failed pings, failed dials keep
retrying every 5s, and the UI rebinds peers after a successful redial
instead of leaving the workspace stuck offline.
Grey out the Swarm sidebar item with a help popover explaining how to
enable swarm mode. Recheck the peer on each hover and re-enable the
nav when LocalNodeState becomes active.
Reusable column visibility control after Clear on containers and on
other resource tables, with preferences persisted. Suppress horizontal
scrollbars so columns ellipsize instead of overflowing.
Validate port mappings for empty host ports, in-form duplicates,
privileged low ports, and peer-side host port conflicts. Runs on
template deploy, deploy view, and add-container before create RPC.
Replace em-dash and ad-hoc loaders with the shared job-tray spinner
for deploy templates, host/swarm/tunnels/events, health latency,
container stats/logs/processes, and registry/volume browsers.
Peers live under Settings; Images is local-only. Registry is a top-level view with vault credentials, Hub search, and Registry API V2 catalog/tags/manifest/delete over the wire.
Registry HTTP now sets host/hostname, disables keep-alive, and times out cleanly; resolve container image tags/digests more reliably and surface check errors in the UI.
Compare local RepoDigests to remote registry manifest digests, show status in the containers table Updates column, and support force recheck with vault credentials for private registries.
Adds a dedicated blank create form (name, image, always pull, ports,
auto-remove, advanced settings) separate from template Deploy, reuses
deployContainer with pull-if-missing and publish-all-ports support.
Lifecycle RPCs now use a 120s operation timeout, remove cleans up
stats/terminal/log streams then SIGKILLs before force-remove, and the
UI awaits the request instead of a fragile 30s wait race.
Raven Scott