v0.17.1

A backup hook that fails is tried again every 10 s (2 minutes, 10 for a scheduled run), like the dumps wait for their database: a new pod's first connection can be refused while the cluster's network policy catches up. A hook running as a non-root user can now record its failure; before, it could not write the marker, its container failed and the run stopped there, without the after-hook.

v0.17.0

A tag can move the StatefulSets of its box: switchyard.statefulset.<name>=<repo>:<tag>@sha256:<digest> labels, an opt-in list per box, compared by digest with what the pods run, a backup of the namespace before a move, forward only (Roll back leaves them). The cache command runs up to three times, and a run whose cache command still failed says so.

v0.16.3

Retention prunes only runs with a manifest. Up to 0.16.2, archives from before switchyard in a set's directory (or a run cut off halfway) were pruned like any other run once their date left the daily window; upgrade before that happens to yours. Also in this release (0.16.2): every dump waits for its database (2 minutes, 10 for scheduled runs), and so does a restore before it drops and imports one; bucket copies are tried again in every run; a failed backup run removes its own files; a deploy's pending-SQL check runs once more when it fails.

v0.16.1

A deploy, a backup or a restore opens a run page with only its progress (steps, message, log, Abort, Back); the box and backups pages show no logs any more. Every box keeps its last 10 runs with their logs, opened from a console icon next to each tag; backup runs keep their full logs. Roll back has no number. The pages show their last state at once and put the cluster's in when it comes, and after a quiet spell switchyard answers with its last view while it reads the cluster in the background. Also in this release (0.15.3): Inside a dashboard hides the page headings as well.

v0.15.2

Running shows each namespace as one line until it is opened, then its workloads as a table with aligned columns (a pod row only where there are several). The pages are wider on the desktop, and phones get a layout of their own: a tab bar at the bottom, tables as stacked cards, Running as tiles with a dot per pod. A new setting, Inside a dashboard, drops the logo and the heading and moves the links to the bottom line, for switchyard shown inside a dashboard with a top bar of its own. The header link Login is now called Settings (0.14.1).

v0.14.0

A cluster whose machine is off shows up as offline at once: switchyard dials the API server first (2 s at most, kept 5 s) instead of waiting for kubectl's own retries, and the boxes list gets the offline card too. A power-on command (Redfish, IPMI, Wake-on-LAN or anything else, with an optional user and password in its environment) can be set on the Login page with the password; while the cluster is offline every page then offers Start server.

v0.13.1

A scheduled backup missed while the cluster was off runs when it is back, up to a day late, once; its dumps wait for the databases and its bucket copies are retried, 10 minutes each. Also in this release (0.13.0): the login can be turned off behind an authenticating proxy, and the session cookie stays out of the directories below the top one.

v0.12.0

S3 buckets as backup items: any S3-compatible store, one archive per bucket, restored by sync; path imports of switchyard's own tars fixed.

switchyard 0.11.0

• Running, a page next to the boxes and the backups: one card per namespace with its workloads, pods, containers, addresses, CPU and memory (from metrics-server), live while someone looks.
• Stop and Start per namespace and per workload. Stop brings a namespace down in order (CronJobs, Deployments, StatefulSets) and remembers what ran; Start brings it back the other way round. A box in a stopped namespace refuses deploys, and its backup schedule shows "stopped with the namespace" instead of an error.
• A new ClusterRole switchyard-running: read workloads everywhere, write only the scale subresource. New installs get it from the installer; an existing one shows an Enable button on the Running page that writes it with the admin context, once.

switchyard 0.10.1

• switchyard password sets a new password from the machine it runs on, for a forgotten one: it asks twice, ends every session, then restart the service (systemctl --user restart switchyard).