X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/cb2d522176c17f2d388098b70fdbaa90fb30e682..3e0786c27f85e715ac3d624a397ad84ae938384d:/lib/dispatchcloud/node_size.go

diff --git a/lib/dispatchcloud/node_size.go b/lib/dispatchcloud/node_size.go
index 6831d8d8e9..0b394f4cfe 100644
--- a/lib/dispatchcloud/node_size.go
+++ b/lib/dispatchcloud/node_size.go
@@ -56,7 +56,7 @@ func estimateDockerImageSize(collectionPDH string) int64 {
 // EstimateScratchSpace estimates how much available disk space (in
 // bytes) is needed to run the container by summing the capacity
 // requested by 'tmp' mounts plus disk space required to load the
-// Docker image.
+// Docker image plus arv-mount block cache.
 func EstimateScratchSpace(ctr *arvados.Container) (needScratch int64) {
 	for _, m := range ctr.Mounts {
 		if m.Kind == "tmp" {
@@ -80,20 +80,23 @@ func EstimateScratchSpace(ctr *arvados.Container) (needScratch int64) {
 	// Now reserve space for the extracted image on disk.
 	needScratch += dockerImageSize
 
+	// Now reserve space the arv-mount disk cache
+	needScratch += ctr.RuntimeConstraints.KeepCacheDisk
+
 	return
 }
 
 // compareVersion returns true if vs1 < vs2, otherwise false
-func versionLess(vs1 string, vs2 string) bool {
+func versionLess(vs1 string, vs2 string) (bool, error) {
 	v1, err := strconv.ParseFloat(vs1, 64)
 	if err != nil {
-		return false
+		return false, err
 	}
 	v2, err := strconv.ParseFloat(vs2, 64)
 	if err != nil {
-		return false
+		return false, err
 	}
-	return v1 < v2
+	return v1 < v2, nil
 }
 
 // ChooseInstanceType returns the cheapest available
@@ -110,11 +113,19 @@ func ChooseInstanceType(cc *arvados.Cluster, ctr *arvados.Container) (best arvad
 
 	needRAM := ctr.RuntimeConstraints.RAM + ctr.RuntimeConstraints.KeepCacheRAM
 	needRAM += int64(cc.Containers.ReserveExtraRAM)
-	needRAM += int64(cc.Containers.LocalKeepBlobBuffersPerVCPU * needVCPUs * (1 << 26))
+	if cc.Containers.LocalKeepBlobBuffersPerVCPU > 0 {
+		// + 200 MiB for keepstore process + 10% for GOGC=10
+		needRAM += 220 << 20
+		// + 64 MiB for each blob buffer + 10% for GOGC=10
+		needRAM += int64(cc.Containers.LocalKeepBlobBuffersPerVCPU * needVCPUs * (1 << 26) * 11 / 10)
+	}
 	needRAM = (needRAM * 100) / int64(100-discountConfiguredRAMPercent)
 
 	ok := false
 	for _, it := range cc.InstanceTypes {
+		driverInsuff, driverErr := versionLess(it.CUDA.DriverVersion, ctr.RuntimeConstraints.CUDA.DriverVersion)
+		capabilityInsuff, capabilityErr := versionLess(it.CUDA.HardwareCapability, ctr.RuntimeConstraints.CUDA.HardwareCapability)
+
 		switch {
 		// reasons to reject a node
 		case ok && it.Price > best.Price: // already selected a node, and this one is more expensive
@@ -124,8 +135,8 @@ func ChooseInstanceType(cc *arvados.Cluster, ctr *arvados.Container) (best arvad
 		case it.Preemptible != ctr.SchedulingParameters.Preemptible: // wrong preemptable setting
 		case it.Price == best.Price && (it.RAM < best.RAM || it.VCPUs < best.VCPUs): // same price, worse specs
 		case it.CUDA.DeviceCount < ctr.RuntimeConstraints.CUDA.DeviceCount: // insufficient CUDA devices
-		case ctr.RuntimeConstraints.CUDA.DeviceCount > 0 && versionLess(it.CUDA.DriverVersion, ctr.RuntimeConstraints.CUDA.DriverVersion): // insufficient driver version
-		case ctr.RuntimeConstraints.CUDA.DeviceCount > 0 && versionLess(it.CUDA.HardwareCapability, ctr.RuntimeConstraints.CUDA.HardwareCapability): // insufficient hardware capability
+		case ctr.RuntimeConstraints.CUDA.DeviceCount > 0 && (driverInsuff || driverErr != nil): // insufficient driver version
+		case ctr.RuntimeConstraints.CUDA.DeviceCount > 0 && (capabilityInsuff || capabilityErr != nil): // insufficient hardware capability
 			// Don't select this node
 		default:
 			// Didn't reject the node, so select it