v1.26.0

Added

Per-container update cooldown (advanced mode) — opt-in pause after recreating a container, before the next one in a batch update. From @famewolf's GPU case in #2: two memory-heavy containers updating in the same run contend for memory because the freshly-recreated one is still loading when the next begins. The manual "Update all" path had no inter-container wait (only the scheduled path did, between Container Group members) — this fills that gap in both paths, no group required.

Set /cooldown <name> <seconds> (Telegram) or the field on the container detail page (Web UI, advanced mode); 0–600s, default 0.

Scope, honestly: this spaces out load-peak overlap — it does not create memory. If your GPU/RAM genuinely can't hold all recreated containers at once, ordering matters more than spacing: put them in a Container Group heaviest-first (both update paths already honour group order). The crash-loop guard (v1.23.5) stays the backstop that rolls back an update that OOMs anyway.