Rebuild the rollout cell around an explicit five-state machine - #2098
Open
fzyzcjy wants to merge 1 commit into
Open
Rebuild the rollout cell around an explicit five-state machine#2098fzyzcjy wants to merge 1 commit into
fzyzcjy wants to merge 1 commit into
Conversation
fzyzcjy
requested review from
Shi-Dong,
Zhichenzzz,
guapisolo,
jybsuper,
maocheng23,
yueming-yuan and
yushengsu-thu
as code owners
August 3, 2026 15:01
Contributor
|
Caution The consumer version of Gemini Code Assist on GitHub has been sunset. All code review activity has officially ceased. |
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-119
branch
from
August 3, 2026 23:43
b193c44 to
dde9aec
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-120
branch
2 times, most recently
from
August 4, 2026 00:38
12ac8dd to
d0a6fbb
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-119
branch
from
August 4, 2026 00:38
dde9aec to
a7b458d
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-120
branch
from
August 4, 2026 02:36
d0a6fbb to
655543b
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-119
branch
2 times, most recently
from
August 4, 2026 06:53
137f21d to
58a69cf
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-120
branch
from
August 4, 2026 06:53
655543b to
e2ba0f1
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-119
branch
from
August 4, 2026 07:24
58a69cf to
47c83b9
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-120
branch
from
August 4, 2026 07:24
e2ba0f1 to
1105a58
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-119
branch
from
August 4, 2026 07:31
47c83b9 to
9abc46a
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-120
branch
from
August 4, 2026 07:31
1105a58 to
5ad47c4
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-119
branch
from
August 6, 2026 03:46
9abc46a to
d0fbbff
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-120
branch
from
August 6, 2026 03:46
5ad47c4 to
e8d08e6
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-119
branch
from
August 6, 2026 07:04
d0fbbff to
d58254d
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-120
branch
from
August 6, 2026 07:04
e8d08e6 to
2b256fc
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-119
branch
from
August 9, 2026 23:31
d58254d to
a3ac4fb
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-120
branch
from
August 9, 2026 23:31
2b256fc to
151b9f0
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-119
branch
from
August 10, 2026 00:44
a3ac4fb to
d825a69
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-120
branch
from
August 10, 2026 00:44
151b9f0 to
976fea2
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-119
branch
from
August 10, 2026 16:06
29be034 to
042fa96
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-120
branch
from
August 10, 2026 16:33
132d90c to
f4509ea
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-119
branch
from
August 10, 2026 16:33
042fa96 to
ed0ca01
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-120
branch
from
August 12, 2026 02:48
f4509ea to
d663cd0
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-119
branch
from
August 12, 2026 09:32
88ea78b to
ca37002
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-120
branch
from
August 12, 2026 09:32
d663cd0 to
86c5325
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-119
branch
from
August 12, 2026 14:31
ca37002 to
6f8ab9a
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-120
branch
2 times, most recently
from
August 12, 2026 14:36
090417b to
3e27670
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-119
branch
from
August 12, 2026 14:36
6f8ab9a to
262ab1d
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-120
branch
from
August 12, 2026 23:56
3e27670 to
59be169
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-119
branch
from
August 13, 2026 01:13
45a326f to
e5c5f0c
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-120
branch
from
August 13, 2026 01:13
59be169 to
d5dbcca
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-119
branch
from
August 13, 2026 03:11
e5c5f0c to
3d17ff6
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-120
branch
from
August 13, 2026 03:11
d5dbcca to
23d2588
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-119
branch
from
August 13, 2026 03:25
3d17ff6 to
4cd949d
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-120
branch
from
August 13, 2026 03:25
23d2588 to
17282a3
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-119
branch
from
August 13, 2026 06:00
4cd949d to
8eee1c8
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-120
branch
2 times, most recently
from
August 13, 2026 06:39
6ffe4a1 to
9b7a55e
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-119
branch
2 times, most recently
from
August 14, 2026 04:45
72adb4a to
cdfe05f
Compare
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-120
branch
from
August 14, 2026 04:46
9b7a55e to
736001d
Compare
This was referenced Aug 27, 2026
guapisolo
reviewed
Aug 27, 2026
|
|
||
| if not self.meta.update_weights or self.args.debug_rollout_only: | ||
| await self.mark_weights_ready() | ||
| self._change_state("mark_pending_weights", StateInitializing, StatePendingWeights(addr_info=addr_info)) |
Collaborator
There was a problem hiding this comment.
[P2] Make initialization retries phase-aware
_tick_when_initializing() keeps the cell in StateInitializing across snapshot, release, and resume. If release succeeds server-side but its response is lost, resume fails, or the tick is cancelled after release, the next sweep starts over and snapshots weights that may already have been discarded. This can corrupt the weight-checker baseline or leave a colocated engine stuck after a transient HTTP failure. Please persist or reconcile the completed startup phase so retries resume from the first incomplete operation instead of replaying the whole sequence.
A cell now starts uninitialized, releases its engine's launch gate in init(), and is carried to pending-weights by the tick sweep once the engine answers, instead of blocking the caller inside add(). Readiness polling moves from a blocking wait to a single short-timeout probe per tick, and dispose() is accepted from every state.
fzyzcjy
force-pushed
the
tom/refactor-miles/op13-120
branch
from
August 29, 2026 12:38
736001d to
9a80842
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Part of #1837