Failed work under Settings → System → Queue now says why each item failed: a platform fault (disk, model, encoder, timeout) that a Retry fixes after the platform is repaired, a…
- #298 17e5fea Failed work under Settings → System → Queue now says why each item failed: a platform fault (disk, model, encoder, timeout) that a Retry fixes after the platform is repaired, a source problem (a dead or private link, no audio, no usable clips) that fails the same way again, or a plan limit. Admins can dismiss failures nobody will retry, one at a time or every match of a filter, and bring them back from Show dismissed. Dismissing only clears the admin list: the owner's project keeps its error and its own Retry. Retry matching now retries platform failures unless you pick another cause. Videos refused for a plan limit in the last 30 days are retried automatically when the owner subscribes, changes plan or has their limits raised. The same actions are available as the dismiss_failures MCP tool, a cause filter on list_failures and retry_failures, and the reelsaw dismiss-failures and reelsaw retry-failures --cause commands.
- #297 ddf5f67 Sorting the admin Users list by a column now sorts every matching account on the server. Before, a column header reordered only the rows already loaded, so "Stored, biggest first" showed the biggest of the first 50 accounts, not the biggest overall. Sorting works with search, filters and grouping, and pages with "Load more". The sort is in the page link (#sort=storedBytes&dir=desc). Agents and operators get the same order: list_users takes sort and dir, and reelsaw users takes --sort and --dir.