VersusControl/versus-incident v1.4.35

on GitHub

Kubernetes History That Keeps Saving

• Keep saving history on clusters that run for weeks. Saving follows real
  change in the cluster, not how long Versus has been running.
• Keep history across EKS credential renewals.
• Save large rollouts in full. A capture that is too large keeps the newest
  data instead of being dropped.
• Keep history fast as it grows: ten years of history reads about as fast as
  one.
• See "History is temporarily not saving" or "History could not save the latest
  data" when saving fails, with your saved rows still on screen. A database
  problem is never shown as a credential problem.

History Retention

• History is kept for 7 days by default. Set
  kubernetes.evidence.retention to keep it longer (for example 720h for
  30 days), or "0" to keep everything.
• Set the save part size in config.yaml or Helm with
  kubernetes.evidence.save_chunk.

Kubernetes Lists You Can Page Through

• See the newest break first in Issues.
• See the range and page count on every Kubernetes list, for example
  51–100 of 312 · Page 2 / 7.
• Page without missing or repeating rows. If the list changes, Versus starts
  again from page 1 or tells you the list changed.
• Get the same warnings and counts for an issue on every page.

History Everywhere

• Use Timeline, Changes, Crashes and historical Issues without Postgres.
• See the Overview on clusters where Versus may not read some kinds, such as
  Nodes, with the missing kinds listed.

Enterprise

• Get the same history saving, paging and retention under your licensed
  organization, with save_chunk read from the config file.

Upgrade Notes

• Retention now defaults to 7 days. To keep more, set
  kubernetes.evidence.retention before upgrading; "0" keeps everything.
  During a rolling update, keep "0" until every replica runs v1.4.35.
• The first start takes a few seconds longer while history tables are prepared.
  v1.4.34 and v1.4.35 replicas can run together.
• Page links saved before upgrading start again from page 1. API clients get
  400 cursor_expired.
• Security: andybalholm/brotli v1.2.7 (GO-2026-6888) and a patched
  golang.org/x/crypto.
• AWS SDK for Go v2 updated to v1.47.0 (STS v1.51.0, CloudWatch Logs v1.88.0,
  SNS v1.47.1, SSM Incidents v1.47.0). No configuration change.

See the v1.4.35 upgrade guide.

What's Changed

• feat(kubernetes): steady history saving, pageable lists and 7-day history retention by @hoalongnatsu in #400
• docs(src): update agent introduction by @hoalongnatsu in #401
• agent(harness): disable action from web interface by @hoalongnatsu in #405
• docs(finops): aws finops agent in development page by @hoalongnatsu in #406

Full Changelog: v1.4.34...v1.4.35