Skip to content

v17.4.0 - Restore point reliability & live progress

Choose a tag to compare

View all tags
@KevinJump KevinJump released this 26 Aug 06:49
v17.4.0
4dc169e
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Verified
Learn about vigilant mode.

This release brings the restore point pipeline up to the same live-progress and background-queue improvements from v18, plus a batch of smaller fixes and backports that shipped in the v17.3.9-17.3.11 patch releases along the way.

Added

  • Snapshots: item-level progress for restore point create/restore/report now reaches the backoffice UI. The progress callbacks were never wired into the underlying sync pack request, so they were previously no-ops.
  • Snapshots: uSync:Publisher:Settings:RestorePointIncludeMedia (default false) - restore points created on request from a remote server no longer copy the whole media library before every push unless this is turned on.
  • Snapshots: a cleanup job now removes restore point working folders left behind by runs that never finished. Controlled by uSync:Restore:Settings:KeepWorkingFoldersForHours (default 24, 0 disables it).
  • Publisher: a per-server Hide "Compare" tab UI setting, for target sites that block being embedded in an iframe (backported from v18).

Fixed

  • Publisher: server-to-server requests now honour uSync:Publisher:Settings:RequestTimeout again. The setting had been ignored since v15, so requests were silently capped at 100 seconds instead of the configured value.
  • Snapshots: restore point archives are now built on disk instead of in memory, so a whole-site restore point no longer risks taking the process down on larger sites.
  • Snapshots: the restore point working folder is only deleted once its archive is confirmed stored, so a failure mid-write no longer loses both.
  • Snapshots: uSync:Restore:Settings:RestoreGroups now applies when a restore point is created and restored, not just when it's reported on. Default is now All to match what restore points actually capture.
  • Snapshots/Exporter: media writes no longer block on each other, so one stalled write (a slow blob store, say) can no longer hold up every import behind it.
  • Snapshots: restore point compression and cache saves now run on a dedicated background queue instead of Umbraco's own shared task queue, so a slow compression job can no longer block Umbraco's own background work (and vice versa).
  • #308 - Publisher: pressing Cancel on the "unsaved changes" dialog when pushing to a server no longer opened the push dialog anyway (backported from v18).
  • #306 - Publisher: the scheduled publish date is now normalised to UTC before being written into the pack, fixing schedules landing offset on servers with a non-zero timezone.
  • Publisher: after a Pull from server, an open content editor now refreshes automatically instead of showing the pre-pull data (backported from v18).
  • Startup now checks whether the backoffice is included in the site before loading uSync.Complete.
  • Caching turned off for v17.4/5, using Clear instead of ClearKey.

Changed

  • Updated to uSync 17.3.8 and Jumoo.Processing 17.4.0: restore point progress notifications are now sent asynchronously instead of blocking, and an abandoned restore point request now stops the pipeline cleanly instead of running to completion regardless.
  • Migrated hosted services to RecurringBackgroundJobBase and tidied up JSON options / dependencies.
Assets 2
Loading