Skip to content

Drive the cell state machine through whole-cell transitions - #1885

Open
fzyzcjy wants to merge 1 commit into
tom/refactor-miles/op6-1from
tom/refactor-miles/op6-2
Open

Drive the cell state machine through whole-cell transitions#1885
fzyzcjy wants to merge 1 commit into
tom/refactor-miles/op6-1from
tom/refactor-miles/op6-2

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/op6-2 branch from c1cfca9 to 5f0435c Compare July 28, 2026 13:31
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op6-1 branch 2 times, most recently from 3fa06ea to 36a0815 Compare August 3, 2026 23:45
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op6-2 branch from 5f0435c to 202bd42 Compare August 3, 2026 23:45
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op6-1 branch from 36a0815 to 23d3301 Compare August 4, 2026 00:39
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op6-2 branch from 202bd42 to 2560c03 Compare August 4, 2026 00:39
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op6-1 branch from 23d3301 to cb9631e Compare August 4, 2026 06:58
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op6-2 branch 2 times, most recently from fdec901 to 5c242d1 Compare August 6, 2026 07:03
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op6-1 branch 2 times, most recently from 4c94b79 to 5d6d557 Compare August 10, 2026 06:08
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op6-2 branch 2 times, most recently from a1724e8 to 9e05465 Compare August 10, 2026 16:07
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op6-1 branch 2 times, most recently from 673fee2 to 5bafef7 Compare August 10, 2026 16:34
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op6-2 branch from 9e05465 to 7ad1221 Compare August 10, 2026 16:34
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op6-1 branch from 5bafef7 to 44ade85 Compare August 12, 2026 02:48
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op6-2 branch from 7ad1221 to 3ae7202 Compare August 12, 2026 02:48
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op6-1 branch from 44ade85 to ce85aa0 Compare August 12, 2026 14:31
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op6-2 branch from 3ae7202 to 9625673 Compare August 12, 2026 14:31
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op6-2 branch 3 times, most recently from 60b8e9f to 556d8ec Compare August 13, 2026 03:25
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op6-1 branch from 3e2b85d to 5541f37 Compare August 13, 2026 03:25
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op6-2 branch from 556d8ec to e7203e8 Compare August 13, 2026 06:01
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op6-1 branch from 5541f37 to a03824b Compare August 13, 2026 06:01
A cell's engines are the node-ranks of one distributed SGLang engine, so every
state change now goes through a cell-level _mark_* helper that touches all of
them. This is an equivalent refactor: it is a prerequisite for lifting the
state machine onto ServerCell, and it makes the "all engines of a cell change
state together" invariant explicit at each call site rather than implied by the
loops.

Squashed commits:
- Drive the cell state machine through whole-cell transitions
- Apply black auto-fix
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op6-1 branch from a03824b to 4744679 Compare August 14, 2026 04:46
@fzyzcjy
fzyzcjy force-pushed the tom/refactor-miles/op6-2 branch from e7203e8 to 2ad891c 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.

Approved in stack context: #1885 can leak staged actors on launch failure and its restart test is invalid, both fixed by #2112 with rollback, retry, and whole-cell restart coverage; #2115/#2175 further harden and cover the failure path. These fixes land before safe merge ckpt 1 (#2176) and remain through the checked chain tip #2649, so the findings no longer block this PR within the safe-merge range.

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