From 0135f114d34bc3b3604db7a5c5cbc1c5a0164c8d Mon Sep 17 00:00:00 2001
From: nexxo <nexxo@clupilot.local>
Date: Sat, 25 Jul 2026 23:40:33 +0200
Subject: [PATCH] fix(traffic): keep retrying the throttle release after a
 Proxmox blip

Tying the release to the creation of the new month's row meant a single failed
call was never retried: the row exists from then on, and the current period
looks unthrottled, so nothing could detect the stale NIC limit. A customer would
have stayed slow for a whole month they had paid for. Checked on every sample
now, with a test that makes the first release fail.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
---
 .../Jobs/CollectInstanceTraffic.php           | 12 +++---
 tests/Feature/TrafficTest.php                 | 39 +++++++++++++++++++
 2 files changed, 45 insertions(+), 6 deletions(-)

diff --git a/app/Provisioning/Jobs/CollectInstanceTraffic.php b/app/Provisioning/Jobs/CollectInstanceTraffic.php
index 59de491..7839bb6 100644
--- a/app/Provisioning/Jobs/CollectInstanceTraffic.php
+++ b/app/Provisioning/Jobs/CollectInstanceTraffic.php
@@ -74,12 +74,12 @@ class CollectInstanceTraffic implements ShouldBeUnique, ShouldQueue
             ['last_netin' => $netin, 'last_netout' => $netout, 'sampled_at' => now()],
         );
 
-        // The month rolled over: the new row starts unthrottled, but the NIC
-        // still carries last month's limit and nothing below would ever take it
-        // off — a customer would stay slow into a month they have paid for.
-        if ($row->wasRecentlyCreated) {
-            $this->releasePreviousPeriodThrottle($client, $instance);
-        }
+        // Checked on every sample, not only when the row was just created: if
+        // the release failed once — a Proxmox blip during the rollover run —
+        // tying it to row creation would never try again, and the customer would
+        // stay slow for a month they have paid for. The lookup exits
+        // immediately when there is nothing to release.
+        $this->releasePreviousPeriodThrottle($client, $instance);
 
         // A fresh row starts from this sample: whatever the VM transferred
         // before we started counting this month is not ours to bill.
diff --git a/tests/Feature/TrafficTest.php b/tests/Feature/TrafficTest.php
index 24ea9f9..e9de2da 100644
--- a/tests/Feature/TrafficTest.php
+++ b/tests/Feature/TrafficTest.php
@@ -179,3 +179,42 @@ it('lets a throttled instance run again when the new month starts', function ()
     expect($pve->networkRates[1042])->toBeNull()
         ->and(InstanceTraffic::query()->where('throttled', true)->count())->toBe(0);
 });
+
+it('keeps trying to release a throttle when Proxmox refuses once', function () {
+    [$pve, , $instance] = trafficSetup();
+
+    InstanceTraffic::create([
+        'instance_id' => $instance->id,
+        'period' => now()->subMonth()->format('Y-m'),
+        'tx_bytes' => 9_999_999_999_999,
+        'throttled' => true,
+    ]);
+
+    // A hub that fails the first release and works afterwards — the transient
+    // error that would otherwise leave the customer throttled for good.
+    $flaky = new class extends FakeProxmoxClient
+    {
+        public int $attempts = 0;
+
+        public function setNetworkRate(string $node, int $vmid, ?float $mbytesPerSecond): void
+        {
+            $this->attempts++;
+            if ($mbytesPerSecond === null && $this->attempts === 1) {
+                throw new RuntimeException('proxmox unavailable');
+            }
+            parent::setNetworkRate($node, $vmid, $mbytesPerSecond);
+        }
+    };
+    $flaky->counters[1042] = ['netin' => 0, 'netout' => 0];
+    app()->instance(ProxmoxClient::class, $flaky);
+
+    (new CollectInstanceTraffic)->handle($flaky);
+    expect(InstanceTraffic::query()->where('throttled', true)->count())->toBe(1);
+
+    // Next run: the current row already exists, so a release tied to row
+    // creation would never fire again.
+    (new CollectInstanceTraffic)->handle($flaky);
+
+    expect(InstanceTraffic::query()->where('throttled', true)->count())->toBe(0)
+        ->and($flaky->networkRates[1042])->toBeNull();
+});