The agent has always logged a percentage as it fetches weights, with a comment saying a replica pulling a hundred gigabytes and one that is wedged look identical from the control plane. They still did: the log is on the node, and the node is not where anyone is looking when a replica has been starting for ten minutes. Progress now rides on the shard status and appears under the replica in `ctl replicas` — a line rather than a column, because most replicas are not downloading anything and a column would be empty for all of them. It is deliberately not a state transition. Treating each chunk as one would re-mark the shard as starting on every report and reset the timer that decides when a slow start has become a failed one, so an hour-long download would keep a doomed replica alive indefinitely. It arrives on its own method for that reason, and a real transition clears it — otherwise a finished download sits in the listing at ninety-nine percent under a shard that has been serving for an hour. Watchers are keyed by source rather than by shard, because a fetch is single-flight: eight shards starting together produce one download and all eight are waiting on it. No percentage is invented when the server sent no length. One that stops moving is worse than a byte count that keeps rising, so the field is absent and the listing says "so far". Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
This commit is contained in:
1 parent
100f10ac17
commit
a35e8796d2
11 files changed
+447
-60
No files matched your search
@@ -24,6 +24,10 @@ import (
|
||||
// agents report they are already running. The reconciler implements it.
|
||||
type StatusSink interface {
|
||||
OnShardStatus(replicaID, shardID string, state supervisor.State, errMsg string)
|
||||
// OnFetchProgress is separate because it is not a transition: the shard is
|
||||
// still starting, and treating each report as a state change would reset
|
||||
// the timers that decide when a slow start has become a failed one.
|
||||
OnFetchProgress(replicaID, shardID, source string, done, total int64)
|
||||
Adopt(nodeID string, shards []reconciler.AdoptedShard) (unclaimed []reconciler.AdoptedShard)
|
||||
}
|
||||
|
||||
@@ -233,6 +237,14 @@ func (s *Service) handle(nodeID string, msg *llmgwv1.AgentMessage) {
|
||||
s.applyHeartbeat(nodeID, m.Heartbeat)
|
||||
|
||||
case *llmgwv1.AgentMessage_Shard:
|
||||
// Progress rather than a transition. Falling through to OnShardStatus
|
||||
// would re-mark the shard as starting on every chunk, which is how a
|
||||
// download that takes an hour would keep a start timer at zero.
|
||||
if f := m.Shard.GetFetch(); f != nil {
|
||||
s.status.OnFetchProgress(m.Shard.GetReplicaId(), m.Shard.GetShardId(),
|
||||
f.GetSource(), f.GetBytesDone(), f.GetBytesTotal())
|
||||
return
|
||||
}
|
||||
st := StateFromWire(m.Shard.GetState())
|
||||
s.status.OnShardStatus(m.Shard.GetReplicaId(), m.Shard.GetShardId(), st, m.Shard.GetError())
|
||||
if m.Shard.GetState() == llmgwv1.ShardStatus_STATE_FAILED {
|
||||
|
||||
Reference in new issue
Block a user