bulltrackers-module 1.0.303 → 1.0.304
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
|
@@ -20,9 +20,9 @@ const SECONDS_PER_CALC_MARGIN = 25; // 20s base + safety margin
|
|
|
20
20
|
async function getHighMemReroutes(db, date, pass, tasks) {
|
|
21
21
|
const reroutes = [];
|
|
22
22
|
for (const task of tasks) {
|
|
23
|
-
const name
|
|
23
|
+
const name = normalizeName(task.name);
|
|
24
24
|
const ledgerPath = `computation_audit_ledger/${date}/passes/${pass}/tasks/${name}`;
|
|
25
|
-
const doc
|
|
25
|
+
const doc = await db.doc(ledgerPath).get();
|
|
26
26
|
|
|
27
27
|
if (doc.exists) {
|
|
28
28
|
const data = doc.data();
|
|
@@ -41,15 +41,15 @@ async function getHighMemReroutes(db, date, pass, tasks) {
|
|
|
41
41
|
|
|
42
42
|
async function dispatchComputationPass(config, dependencies, computationManifest, reqBody = {}) {
|
|
43
43
|
const { logger, db } = dependencies;
|
|
44
|
-
const pubsubUtils
|
|
44
|
+
const pubsubUtils = new PubSubUtils(dependencies);
|
|
45
45
|
|
|
46
46
|
// Inputs from Workflow Cursor
|
|
47
|
-
const passToRun
|
|
47
|
+
const passToRun = String(reqBody.pass || config.COMPUTATION_PASS_TO_RUN);
|
|
48
48
|
const targetCursorN = parseInt(reqBody.cursorIndex || 1);
|
|
49
|
-
const dateLimitStr
|
|
49
|
+
const dateLimitStr = reqBody.date || config.date;
|
|
50
50
|
|
|
51
|
-
const manifestMap
|
|
52
|
-
const passes
|
|
51
|
+
const manifestMap = new Map(computationManifest.map(c => [normalizeName(c.name), c]));
|
|
52
|
+
const passes = groupByPass(computationManifest);
|
|
53
53
|
const calcsInThisPass = passes[passToRun] || [];
|
|
54
54
|
|
|
55
55
|
if (!calcsInThisPass.length) {
|
|
@@ -58,35 +58,35 @@ async function dispatchComputationPass(config, dependencies, computationManifest
|
|
|
58
58
|
|
|
59
59
|
// 1. Discover all "Dirty" Dates (Dates needing work)
|
|
60
60
|
const earliestDates = await getEarliestDataDates(config, dependencies);
|
|
61
|
-
const allDates
|
|
61
|
+
const allDates = getExpectedDateStrings(earliestDates.absoluteEarliest, new Date(dateLimitStr + 'T00:00:00Z'));
|
|
62
62
|
|
|
63
63
|
const dirtyDates = [];
|
|
64
64
|
for (const d of allDates) {
|
|
65
|
-
const dailyStatus
|
|
65
|
+
const dailyStatus = await fetchComputationStatus(d, config, dependencies);
|
|
66
66
|
const availability = await checkRootDataAvailability(d, config, dependencies, DEFINITIVE_EARLIEST_DATES);
|
|
67
67
|
|
|
68
68
|
const report = analyzeDateExecution(d, calcsInThisPass, availability.status, dailyStatus, manifestMap, null);
|
|
69
|
-
const tasks
|
|
69
|
+
const tasks = [...report.runnable, ...report.reRuns];
|
|
70
70
|
|
|
71
71
|
if (tasks.length > 0) {
|
|
72
72
|
dirtyDates.push({ date: d, tasks });
|
|
73
73
|
}
|
|
74
74
|
}
|
|
75
75
|
|
|
76
|
-
let selectedDate
|
|
76
|
+
let selectedDate = null;
|
|
77
77
|
let selectedTasks = [];
|
|
78
|
-
let isReroute
|
|
79
|
-
let isSweep
|
|
78
|
+
let isReroute = false;
|
|
79
|
+
let isSweep = false;
|
|
80
80
|
|
|
81
81
|
// 2. Logic: Prioritize Reroute for N-1
|
|
82
82
|
if (targetCursorN > 1 && (targetCursorN - 2) < dirtyDates.length) {
|
|
83
83
|
const prevEntry = dirtyDates[targetCursorN - 2];
|
|
84
|
-
const reroutes
|
|
84
|
+
const reroutes = await getHighMemReroutes(db, prevEntry.date, passToRun, prevEntry.tasks);
|
|
85
85
|
|
|
86
86
|
if (reroutes.length > 0) {
|
|
87
|
-
selectedDate
|
|
87
|
+
selectedDate = prevEntry.date;
|
|
88
88
|
selectedTasks = reroutes;
|
|
89
|
-
isReroute
|
|
89
|
+
isReroute = true;
|
|
90
90
|
logger.log('INFO', `[Dispatcher] Reroute detected for ${selectedDate}. Pausing N increment.`);
|
|
91
91
|
}
|
|
92
92
|
}
|
|
@@ -94,14 +94,14 @@ async function dispatchComputationPass(config, dependencies, computationManifest
|
|
|
94
94
|
// 3. Logic: N-th Dirty Date or Final Sweep
|
|
95
95
|
if (!selectedDate) {
|
|
96
96
|
if (targetCursorN <= dirtyDates.length) {
|
|
97
|
-
const entry
|
|
98
|
-
selectedDate
|
|
97
|
+
const entry = dirtyDates[targetCursorN - 1];
|
|
98
|
+
selectedDate = entry.date;
|
|
99
99
|
selectedTasks = entry.tasks;
|
|
100
100
|
} else {
|
|
101
101
|
// Final Satiation Sweep: Check if anything was missed (recovery)
|
|
102
102
|
if (dirtyDates.length > 0) {
|
|
103
|
-
isSweep
|
|
104
|
-
selectedDate
|
|
103
|
+
isSweep = true;
|
|
104
|
+
selectedDate = dirtyDates[0].date;
|
|
105
105
|
selectedTasks = dirtyDates[0].tasks;
|
|
106
106
|
}
|
|
107
107
|
}
|
|
@@ -114,21 +114,21 @@ async function dispatchComputationPass(config, dependencies, computationManifest
|
|
|
114
114
|
|
|
115
115
|
// 5. Dispatch to PubSub (Standard vs. High-Mem)
|
|
116
116
|
const standardTasks = selectedTasks.filter(t => t.resources !== 'high-mem').map(t => ({ ...t, date: selectedDate, pass: passToRun }));
|
|
117
|
-
const highMemTasks
|
|
117
|
+
const highMemTasks = selectedTasks.filter(t => t.resources === 'high-mem').map(t => ({ ...t, date: selectedDate, pass: passToRun }));
|
|
118
118
|
|
|
119
119
|
const pubPromises = [];
|
|
120
120
|
if (standardTasks.length > 0) {
|
|
121
121
|
pubPromises.push(pubsubUtils.batchPublishTasks(dependencies, {
|
|
122
122
|
topicName: config.computationTopicStandard || 'computation-tasks',
|
|
123
|
-
tasks: standardTasks,
|
|
124
|
-
taskType: `pass-${passToRun}-std`
|
|
123
|
+
tasks : standardTasks,
|
|
124
|
+
taskType : `pass-${passToRun}-std`
|
|
125
125
|
}));
|
|
126
126
|
}
|
|
127
127
|
if (highMemTasks.length > 0) {
|
|
128
128
|
pubPromises.push(pubsubUtils.batchPublishTasks(dependencies, {
|
|
129
129
|
topicName: config.computationTopicHighMem || 'computation-tasks-highmem',
|
|
130
|
-
tasks: highMemTasks,
|
|
131
|
-
taskType: `pass-${passToRun}-high`
|
|
130
|
+
tasks : highMemTasks,
|
|
131
|
+
taskType : `pass-${passToRun}-high`
|
|
132
132
|
}));
|
|
133
133
|
}
|
|
134
134
|
await Promise.all(pubPromises);
|
|
@@ -138,11 +138,11 @@ async function dispatchComputationPass(config, dependencies, computationManifest
|
|
|
138
138
|
logger.log('INFO', `[Dispatcher] ${isReroute ? 'Reroute' : (isSweep ? 'Sweep' : 'Standard')} Run: ${selectedDate}. Tasks: ${selectedTasks.length}. ETA: ${etaSeconds}s`);
|
|
139
139
|
|
|
140
140
|
return {
|
|
141
|
-
status: isSweep ? 'RECOVERY' : 'CONTINUE_PASS',
|
|
142
|
-
dateProcessed: selectedDate,
|
|
143
|
-
dispatched: selectedTasks.length,
|
|
141
|
+
status : isSweep ? 'RECOVERY' : 'CONTINUE_PASS',
|
|
142
|
+
dateProcessed : selectedDate,
|
|
143
|
+
dispatched : selectedTasks.length,
|
|
144
144
|
n_cursor_ignored: isReroute, // Tell workflow to stay on same N
|
|
145
|
-
etaSeconds: etaSeconds
|
|
145
|
+
etaSeconds : etaSeconds
|
|
146
146
|
};
|
|
147
147
|
}
|
|
148
148
|
|
|
@@ -1,12 +1,14 @@
|
|
|
1
1
|
# Cloud Workflows: Precision Cursor-Based Orchestrator
|
|
2
|
+
# PURPOSE: Orchestrates 5 passes with deterministic pacing and syntax fixes.
|
|
3
|
+
|
|
2
4
|
main:
|
|
3
5
|
params: [input]
|
|
4
6
|
steps:
|
|
5
7
|
- init:
|
|
6
8
|
assign:
|
|
7
|
-
- project: ${sys.get_env("GOOGLE_CLOUD_PROJECT_ID")}
|
|
9
|
+
- project: '${sys.get_env("GOOGLE_CLOUD_PROJECT_ID")}'
|
|
8
10
|
- passes: ["1", "2", "3", "4", "5"]
|
|
9
|
-
- date_to_run: ${default(map.get(input, "date"), "2025-01-01")}
|
|
11
|
+
- date_to_run: '${default(map.get(input, "date"), "2025-01-01")}'
|
|
10
12
|
|
|
11
13
|
- run_sequential_passes:
|
|
12
14
|
for:
|
|
@@ -25,29 +27,38 @@ main:
|
|
|
25
27
|
- call_dispatcher:
|
|
26
28
|
call: http.post
|
|
27
29
|
args:
|
|
28
|
-
url: ${"https://europe-west1-" + project + ".cloudfunctions.net/
|
|
30
|
+
url: '${"https://europe-west1-" + project + ".cloudfunctions.net/computation-pass-" + pass_id}'
|
|
29
31
|
body:
|
|
30
|
-
pass: ${pass_id}
|
|
31
|
-
cursorIndex: ${n_cursor}
|
|
32
|
-
date: ${date_to_run}
|
|
32
|
+
pass: '${pass_id}'
|
|
33
|
+
cursorIndex: '${n_cursor}'
|
|
34
|
+
date: '${date_to_run}'
|
|
33
35
|
auth: { type: OIDC }
|
|
34
36
|
result: dispatch_res
|
|
35
37
|
|
|
36
38
|
- evaluate_dispatch:
|
|
37
39
|
switch:
|
|
38
|
-
# State 1:
|
|
39
|
-
- condition: ${dispatch_res.body.status == "MOVE_TO_NEXT_PASS"}
|
|
40
|
-
assign:
|
|
40
|
+
# State 1: Pass exhausted and Satiation Sweep complete
|
|
41
|
+
- condition: '${dispatch_res.body.status == "MOVE_TO_NEXT_PASS"}'
|
|
42
|
+
assign:
|
|
43
|
+
- pass_complete: true
|
|
41
44
|
|
|
42
|
-
# State 2: Tasks dispatched (Standard, Reroute, or Recovery)
|
|
43
|
-
- condition: ${dispatch_res.body.dispatched > 0}
|
|
45
|
+
# State 2: Tasks dispatched (Standard, Reroute, or Recovery Sweep)
|
|
46
|
+
- condition: '${dispatch_res.body.dispatched > 0}'
|
|
44
47
|
steps:
|
|
45
|
-
-
|
|
48
|
+
- log_dispatch:
|
|
49
|
+
call: sys.log
|
|
50
|
+
args:
|
|
51
|
+
text: '${"Pass " + pass_id + " - Dispatching " + string(dispatch_res.body.dispatched) + " tasks for " + dispatch_res.body.dateProcessed + ". ETA: " + string(dispatch_res.body.etaSeconds) + "s."}'
|
|
52
|
+
- wait_for_completion:
|
|
46
53
|
call: sys.sleep
|
|
47
|
-
args:
|
|
54
|
+
args:
|
|
55
|
+
seconds: '${int(dispatch_res.body.etaSeconds)}'
|
|
48
56
|
- update_cursor:
|
|
49
57
|
assign:
|
|
50
|
-
# If
|
|
51
|
-
- n_cursor: ${if(dispatch_res.body.n_cursor_ignored, n_cursor, n_cursor + 1)}
|
|
52
|
-
-
|
|
53
|
-
next: sequential_date_loop
|
|
58
|
+
# If n_cursor_ignored is true (Reroute or Sweep Recovery), we stay on same N
|
|
59
|
+
- n_cursor: '${if(dispatch_res.body.n_cursor_ignored, n_cursor, n_cursor + 1)}'
|
|
60
|
+
- next_loop:
|
|
61
|
+
next: sequential_date_loop
|
|
62
|
+
|
|
63
|
+
- finish:
|
|
64
|
+
return: "Pipeline Execution Satiated and Complete"
|