Skip to content

Move per-cell recovery into ServerCell.recover - #1881

Open
fzyzcjy wants to merge 1 commit into
tom/refactor-miles/op5-23from
tom/refactor-miles/op5-17
Open

Move per-cell recovery into ServerCell.recover#1881
fzyzcjy wants to merge 1 commit into
tom/refactor-miles/op5-23from
tom/refactor-miles/op5-17

Conversation

@fzyzcjy

@fzyzcjy fzyzcjy commented Jul 28, 2026

Copy link
Copy Markdown
Collaborator

Part of #1837

@gemini-code-assist

Copy link
Copy Markdown
Contributor

Caution

The consumer version of Gemini Code Assist on GitHub has been sunset. All code review activity has officially ceased.

@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op5-17 branch from 7493517 to cfe8621 Compare July 28, 2026 13:31
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op5-23 branch 2 times, most recently from 92a733e to a9eb34e Compare August 3, 2026 23:45
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op5-17 branch 2 times, most recently from 1c910b1 to 0e9f0d5 Compare August 4, 2026 00:39
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op5-23 branch from a9eb34e to 04e65d7 Compare August 4, 2026 00:39
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op5-17 branch from 0e9f0d5 to 5510389 Compare August 4, 2026 06:57
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op5-23 branch from 04e65d7 to 975ce31 Compare August 4, 2026 06:57
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op5-17 branch 2 times, most recently from c70d67b to 00bb032 Compare August 10, 2026 06:07
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op5-23 branch from 81ebf4e to c326b1e Compare August 10, 2026 06:07
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op5-17 branch from 00bb032 to a3a3f81 Compare August 10, 2026 16:07
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op5-23 branch from 39948d0 to 9ed4407 Compare August 10, 2026 16:34
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op5-17 branch 2 times, most recently from 16f1d68 to 786ee9a Compare August 12, 2026 02:48
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op5-23 branch from 081b08f to ac37584 Compare August 12, 2026 14:31
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op5-17 branch 2 times, most recently from 167d712 to ff2dd3a Compare August 12, 2026 23:56
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op5-23 branch from 045a39b to ff36578 Compare August 13, 2026 01:13
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op5-17 branch from ff2dd3a to 68a93b4 Compare August 13, 2026 01:13
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op5-17 branch 2 times, most recently from 81fdc60 to d0044dd Compare August 13, 2026 06:01
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op5-23 branch from 2fb2661 to ccfb2cc Compare August 13, 2026 06:01
Recovering a cell is start + offload/resume + mark alive + publish to the
router, all of which act on that one cell, so the whole sequence moves
down and the offload policy fields (needs_offload, model_path,
update_weights) follow it into ServerCell. ServerGroup.recover becomes a
thin parallel fan-out that only decides whether a router is there to
register with, and offload/onload gate per cell instead of per group.
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op5-23 branch from ccfb2cc to d6616bb Compare August 14, 2026 04:45
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op5-17 branch from d0044dd to 33ff0ae Compare August 14, 2026 04:46

@guapisolo guapisolo left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Approving in stack context: #2117 completes the provider-reconcile replacement of the aggregate-recovery path, and #2094 fixes missed engine reconnects; both land before safe merge checkpoint #2176 and remain present through the checked tip.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants