Panic in proportion plugin

[svileex]

What happened:
The Proportion plugin panicked when I had queues with guaranteed resources and queues without guaranteed resources. The problem lies in resource division; we must subtract guaranteed resources before performing the remaining division among all queues.

I have already added a test for this behavior and fixed this bug.

~/go/volcano/pkg/scheduler/plugins (proporion-guarantee-fix*) » go test ./...                                                                                                                     svilex@svilexLin
?       volcano.sh/volcano/pkg/scheduler/plugins        [no test files]
ok      volcano.sh/volcano/pkg/scheduler/plugins/binpack        (cached)
ok      volcano.sh/volcano/pkg/scheduler/plugins/cdp    (cached)
?       volcano.sh/volcano/pkg/scheduler/plugins/conformance    [no test files]
ok      volcano.sh/volcano/pkg/scheduler/plugins/drf    (cached)
?       volcano.sh/volcano/pkg/scheduler/plugins/extender       [no test files]
?       volcano.sh/volcano/pkg/scheduler/plugins/gang   [no test files]
?       volcano.sh/volcano/pkg/scheduler/plugins/nodeorder      [no test files]
?       volcano.sh/volcano/pkg/scheduler/plugins/numaaware      [no test files]
ok      volcano.sh/volcano/pkg/scheduler/plugins/numaaware/policy       (cached)
ok      volcano.sh/volcano/pkg/scheduler/plugins/numaaware/provider/cpumanager  (cached)
?       volcano.sh/volcano/pkg/scheduler/plugins/overcommit     [no test files]
ok      volcano.sh/volcano/pkg/scheduler/plugins/predicates     (cached)
?       volcano.sh/volcano/pkg/scheduler/plugins/priority       [no test files]
E1012 21:47:39.004831  214480 utils.go:43] init kubeclient in 4pdvgpu failed: invalid configuration: no configuration has been provided, try setting KUBERNETES_MASTER environment variable
--- FAIL: TestGuarantee (0.00s)
panic: resource is not sufficient to do operation: <cpu 8000.00, memory 8000.00, nvidia.com/gpu 0.00> sub <cpu 10000.00, memory 10000.00> [recovered]
        panic: resource is not sufficient to do operation: <cpu 8000.00, memory 8000.00, nvidia.com/gpu 0.00> sub <cpu 10000.00, memory 10000.00>

goroutine 54 [running]:
testing.tRunner.func1.2({0x278d4a0, 0xc000132b10})
        /usr/local/go/src/testing/testing.go:1396 +0x24e
testing.tRunner.func1()
        /usr/local/go/src/testing/testing.go:1399 +0x39f
panic({0x278d4a0, 0xc000132b10})
        /usr/local/go/src/runtime/panic.go:884 +0x212
volcano.sh/volcano/pkg/scheduler/util/assert.Assert(0x2f?, {0xc000426000?, 0xc000655230?})
        /home/svilex/go/volcano/pkg/scheduler/util/assert/assert.go:33 +0x174
volcano.sh/volcano/pkg/scheduler/util/assert.Assertf(0x0, {0x2bdf02f?, 0xf7941d?}, {0xc000655230?, 0x2a41620?, 0x0?})
        /home/svilex/go/volcano/pkg/scheduler/util/assert/assert.go:43 +0x56
volcano.sh/volcano/pkg/scheduler/api.(*Resource).Sub(0xc0000a67c0, 0xc0000a6800)
        /home/svilex/go/volcano/pkg/scheduler/api/resource_info.go:246 +0x9c
volcano.sh/volcano/pkg/scheduler/plugins/proportion.(*proportionPlugin).OnSessionOpen(0xc0000a6360, 0xc0005143c0)
        /home/svilex/go/volcano/pkg/scheduler/plugins/proportion/proportion.go:241 +0x15c5
volcano.sh/volcano/pkg/scheduler/framework.OpenSession({0x2e7dcf8?, 0xc0001ff900?}, {0xc000012348, 0x1, 0x1}, {0x0, 0x0, 0x0})
        /home/svilex/go/volcano/pkg/scheduler/framework/framework.go:45 +0x327
volcano.sh/volcano/pkg/scheduler/plugins/proportion.TestGuarantee(0xc000602b60)
        /home/svilex/go/volcano/pkg/scheduler/plugins/proportion/proportion_test.go:417 +0xfb8
testing.tRunner(0xc000602b60, 0x2c84658)
        /usr/local/go/src/testing/testing.go:1446 +0x10b
created by testing.(*T).Run
        /usr/local/go/src/testing/testing.go:1493 +0x35f
FAIL    volcano.sh/volcano/pkg/scheduler/plugins/proportion     9.018s
?       volcano.sh/volcano/pkg/scheduler/plugins/rescheduling   [no test files]
?       volcano.sh/volcano/pkg/scheduler/plugins/resourcequota  [no test files]
?       volcano.sh/volcano/pkg/scheduler/plugins/sla    [no test files]
ok      volcano.sh/volcano/pkg/scheduler/plugins/task-topology  (cached)
ok      volcano.sh/volcano/pkg/scheduler/plugins/tdm    (cached)

?       volcano.sh/volcano/pkg/scheduler/plugins/usage  [no test files]
?       volcano.sh/volcano/pkg/scheduler/plugins/util   [no test files]
?       volcano.sh/volcano/pkg/scheduler/plugins/util/k8s       [no test files]
?       volcano.sh/volcano/pkg/scheduler/plugins/util/nodelock  [no test files]
FAIL

What you expected to happen:
I don't expect panic
How to reproduce it (as minimally and precisely as possible):
Create three queues, one with guarantee resources > allResourcesInCluster/3, and two other queues.
Then run job in each queue which takes all resources in this queue.

Anything else we need to know?:
I have already fixed this bug, and now I'm going to create a MR. You can then check how to reproduce this bug in the test from my MR

Environment:

• Volcano Version:
• Kubernetes version (use kubectl version):
• Cloud provider or hardware configuration:
• OS (e.g. from /etc/os-release):
• Kernel (e.g. uname -a):
• Install tools:
• Others:

[svileex]

Mr: #3156

[lowang-bh]

Is it same with #3127?

[svileex]

No, in my case, the guaranteed resource is less than allocatable resources in the cluster

[svileex]

Hey, there hasn't been any news since October. Did I forget to do something?