You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Updated SNARK worker protocol. Now more parallelism can be utilized when constructing proofs for a zkApp command. Redeployment of SNARK worker and SNARK coordinator(i.e. any daemon that's attached to a SNARK worker) would be needed for this to take effect.
3
+
4
+
## Metrics
5
+
The metrics related to SNARK workers has been updated as well.
6
+
7
+
Following metrics are maintained by the worker:
8
+
-`snark_worker_sub_zkapp_command_segment_time`
9
+
-`snark_worker_sub_zkapp_command_merge_time`
10
+
-`snark_worker_merge_time`
11
+
-`snark_worker_nonzkapp_transition_time`
12
+
13
+
addtionally, the coordinator maintains all metrics above, plus an extra metric:
14
+
-`snark_worker_zkapp_transition_time`
15
+
16
+
While having same name, these shared metrics have different meanings across coordinator and worker: Any time tracked by a worker represents the time this specific worker spent on that genre, while any time tracked by a coordinator represents the sum of time any worker attached to that coordinator spent on that genre.
17
+
18
+
## Foot Note
19
+
This is part of the Slot Reduction MIP from Mesa Hardfork.
(* remove it from seen jobs after attempting to adding it to the pool to
926
+
avoid this work being reassigned.
902
927
* If the diff is accepted then remove it from the seen jobs.
903
-
* If not then the work should have already been in the pool with a lower fee or the statement isn't referenced anymore or any other error. In any case remove it from the seen jobs so that it can be picked up if needed *)
904
-
Work_selector.remove t.snark_job_state spec
928
+
* If not then the work should have already been in the pool with a
929
+
lower fee or the statement isn't referenced anymore or any other
930
+
error. In any case remove it from the seen jobs so that it can be
931
+
picked up if needed *)
932
+
Work_selector.remove t.work_selector stmts ;
933
+
Result.iter_error result ~f:(funerr ->
934
+
(* Possible reasons of failure: receiving pipe's capacity exceeded,
935
+
fee that isn't the lowest, failure in verification or application to the pool *)
936
+
[%log warn] "Failed to add completed work to the pool"
0 commit comments