Compare commits

...

7 Commits

Author SHA1 Message Date
hysyeah
738198350f fix: skip nginx reload if configuration has not changed (#2555) 2026-02-25 13:54:09 +08:00
eball
5de7ea01bc fix: skip nginx reload if configuration has not changed 2026-02-25 11:45:13 +08:00
hysyeah
ff7765bcb9 fix: skip invalid expose port (#2434) 2026-01-22 12:00:32 +08:00
wiy
c98c855099 feat(olares-app): update version to v1.8.2 (#2433)
* feat(olares-app): update version to v1.8.2

* feat(olares-app): update version to v1.8.2
2026-01-22 00:06:13 +08:00
hysyeah
4ae552f33f bfl: myapps api add rawAppName (#2432)
* fix: myapps api add rawAppName field

* update bfl api image tag to v0.4.39
2026-01-22 00:05:42 +08:00
hysyeah
f2aad6d9f6 cli: feat amdgpu install (#2430) 2026-01-22 00:05:01 +08:00
dkeven
02bc4fafd5 feat(cli): collect nginx logs stored temporarily in some containers (#2429) 2026-01-22 00:04:29 +08:00
19 changed files with 585 additions and 9 deletions

View File

@@ -317,7 +317,7 @@ spec:
chown -R 1000:1000 /uploadstemp && \
chown -R 1000:1000 /appdata
- name: olares-app-init
image: beclab/system-frontend:v1.7.7
image: beclab/system-frontend:v1.8.2
imagePullPolicy: IfNotPresent
command:
- /bin/sh

View File

@@ -0,0 +1,21 @@
package amdgpu
import (
"log"
"github.com/beclab/Olares/cli/pkg/pipelines"
"github.com/spf13/cobra"
)
func NewCmdAmdGpuInstall() *cobra.Command {
cmd := &cobra.Command{
Use: "install",
Short: "Install AMD ROCm stack via amdgpu-install",
Run: func(cmd *cobra.Command, args []string) {
if err := pipelines.AmdGpuInstall(); err != nil {
log.Fatalf("error: %v", err)
}
},
}
return cmd
}

View File

@@ -0,0 +1,16 @@
package amdgpu
import "github.com/spf13/cobra"
func NewCmdAmdGpu() *cobra.Command {
cmd := &cobra.Command{
Use: "amdgpu",
Short: "Manage AMD GPU ROCm stack",
}
cmd.AddCommand(NewCmdAmdGpuInstall())
cmd.AddCommand(NewCmdAmdGpuUninstall())
cmd.AddCommand(NewCmdAmdGpuStatus())
return cmd
}

View File

@@ -0,0 +1,21 @@
package amdgpu
import (
"log"
"github.com/beclab/Olares/cli/pkg/pipelines"
"github.com/spf13/cobra"
)
func NewCmdAmdGpuStatus() *cobra.Command {
cmd := &cobra.Command{
Use: "status",
Short: "Show AMD GPU driver and ROCm status",
Run: func(cmd *cobra.Command, args []string) {
if err := pipelines.AmdGpuStatus(); err != nil {
log.Fatalf("error: %v", err)
}
},
}
return cmd
}

View File

@@ -0,0 +1,21 @@
package amdgpu
import (
"log"
"github.com/beclab/Olares/cli/pkg/pipelines"
"github.com/spf13/cobra"
)
func NewCmdAmdGpuUninstall() *cobra.Command {
cmd := &cobra.Command{
Use: "uninstall",
Short: "Uninstall AMD ROCm stack via amdgpu-install",
Run: func(cmd *cobra.Command, args []string) {
if err := pipelines.AmdGpuUninstall(); err != nil {
log.Fatalf("error: %v", err)
}
},
}
return cmd
}

View File

@@ -3,6 +3,7 @@ package os
import (
"archive/tar"
"compress/gzip"
"context"
"fmt"
"io"
"log"
@@ -13,6 +14,9 @@ import (
"strings"
"time"
corev1 "k8s.io/api/core/v1"
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
"k8s.io/client-go/kubernetes"
"k8s.io/client-go/rest"
"k8s.io/client-go/tools/clientcmd"
ctrl "sigs.k8s.io/controller-runtime"
@@ -276,7 +280,7 @@ func collectSystemdLogs(tw *tar.Writer, options *LogCollectOptions) error {
}
func collectDmesgLogs(tw *tar.Writer, options *LogCollectOptions) error {
cmd := exec.Command("dmesg")
cmd := exec.Command("dmesg -T")
output, err := cmd.Output()
if err != nil {
return err
@@ -399,6 +403,126 @@ func collectKubernetesLogs(tw *tar.Writer, options *LogCollectOptions) error {
}
}
if err := collectNginxLogsFromLabeledPods(tw); err != nil {
if !options.IgnoreKubeErrors {
return fmt.Errorf("failed to collect nginx logs from labeled pods: %v", err)
}
}
return nil
}
func collectNginxLogsFromLabeledPods(tw *tar.Writer) error {
if _, err := util.GetCommand("kubectl"); err != nil {
fmt.Printf("warning: kubectl not found, skipping collecting nginx logs from labeled pods\n")
return nil
}
cfg, err := ctrl.GetConfig()
if err != nil {
return fmt.Errorf("failed to get kubeconfig: %v", err)
}
clientset, err := kubernetes.NewForConfig(cfg)
if err != nil {
return fmt.Errorf("failed to create kube client: %v", err)
}
type selectorSpec struct {
LabelSelector string
ContainerName string
}
selectors := []selectorSpec{
{LabelSelector: "app=l4-bfl-proxy", ContainerName: ""},
{LabelSelector: "tier=bfl", ContainerName: "ingress"},
}
type targetPod struct {
Namespace string
Name string
ContainerName string
}
var targets []targetPod
ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
defer cancel()
for _, sel := range selectors {
podList, err := clientset.CoreV1().Pods(corev1.NamespaceAll).List(ctx, metav1.ListOptions{LabelSelector: sel.LabelSelector})
if err != nil {
return fmt.Errorf("failed to list pods by label %q: %v", sel.LabelSelector, err)
}
for _, pod := range podList.Items {
targets = append(targets, targetPod{
Namespace: pod.Namespace,
Name: pod.Name,
ContainerName: sel.ContainerName,
})
}
}
if len(targets) == 0 {
return nil
}
// simplest approach: use kubectl cp (it already implements copy via tar over exec)
tempDir, err := os.MkdirTemp("", "olares-nginx-logs-*")
if err != nil {
return fmt.Errorf("failed to create temp directory for nginx logs: %v", err)
}
defer os.RemoveAll(tempDir)
files := []string{"/var/log/nginx/access.log", "/var/log/nginx/error.log"}
for _, target := range targets {
for _, remotePath := range files {
base := filepath.Base(remotePath)
archivePath := filepath.Join("nginx", target.Namespace, target.Name, base)
dest := filepath.Join(tempDir, fmt.Sprintf("%s__%s__%s", target.Namespace, target.Name, base))
err := kubectlCopyFile(target.Namespace, target.Name, target.ContainerName, remotePath, dest)
if err != nil {
return fmt.Errorf("failed to kubectl cp %s/%s:%s: %v", target.Namespace, target.Name, remotePath, err)
}
fi, err := os.Stat(dest)
if err != nil {
return fmt.Errorf("failed to stat copied nginx log %s: %v", dest, err)
}
f, err := os.Open(dest)
if err != nil {
return fmt.Errorf("failed to open copied nginx log %s: %v", dest, err)
}
defer f.Close()
header := &tar.Header{
Name: archivePath,
Mode: 0644,
Size: fi.Size(),
ModTime: time.Now(),
}
if err := tw.WriteHeader(header); err != nil {
return fmt.Errorf("failed to write header for %s: %v", archivePath, err)
}
if _, err := io.CopyN(tw, f, header.Size); err != nil {
return fmt.Errorf("failed to write data for %s: %v", archivePath, err)
}
}
}
return nil
}
func kubectlCopyFile(namespace, pod, container, remotePath, destPath string) error {
args := []string{"-n", namespace, "cp"}
if container != "" {
args = append(args, "-c", container)
}
args = append(args, fmt.Sprintf("%s:%s", pod, remotePath), destPath)
cmd := exec.Command("kubectl", args...)
if out, err := cmd.CombinedOutput(); err != nil {
return fmt.Errorf("kubectl %s failed: %v, output: %s", strings.Join(args, " "), err, strings.TrimSpace(string(out)))
}
return nil
}

View File

@@ -1,6 +1,7 @@
package ctl
import (
"github.com/beclab/Olares/cli/cmd/ctl/amdgpu"
"github.com/beclab/Olares/cli/cmd/ctl/disk"
"github.com/beclab/Olares/cli/cmd/ctl/gpu"
"github.com/beclab/Olares/cli/cmd/ctl/node"
@@ -33,6 +34,7 @@ func NewDefaultCommand() *cobra.Command {
cmds.AddCommand(os.NewOSCommands()...)
cmds.AddCommand(node.NewNodeCommand())
cmds.AddCommand(gpu.NewCmdGpu())
cmds.AddCommand(amdgpu.NewCmdAmdGpu())
cmds.AddCommand(user.NewUserCommand())
cmds.AddCommand(disk.NewDiskCommand())

133
cli/pkg/amdgpu/tasks.go Normal file
View File

@@ -0,0 +1,133 @@
package amdgpu
import (
"fmt"
"os/exec"
"path"
"path/filepath"
"github.com/beclab/Olares/cli/pkg/common"
cc "github.com/beclab/Olares/cli/pkg/core/common"
"github.com/beclab/Olares/cli/pkg/core/connector"
"github.com/beclab/Olares/cli/pkg/core/logger"
"github.com/beclab/Olares/cli/pkg/core/task"
"github.com/beclab/Olares/cli/pkg/utils"
"github.com/Masterminds/semver/v3"
"github.com/pkg/errors"
)
// InstallAmdRocmModule installs AMD ROCm stack on supported Ubuntu if AMD GPU is present.
type InstallAmdRocmModule struct {
common.KubeModule
}
func (m *InstallAmdRocmModule) Init() {
m.Name = "InstallAMDGPU"
installAmd := &task.RemoteTask{
Name: "InstallAmdRocm",
Hosts: m.Runtime.GetHostsByRole(common.Master),
Action: &InstallAmdRocm{
// no manifest needed
},
Parallel: false,
Retry: 1,
}
m.Tasks = []task.Interface{
installAmd,
}
}
// InstallAmdRocm installs ROCm using amdgpu-install on Ubuntu 22.04/24.04 for AMD GPUs.
type InstallAmdRocm struct {
common.KubeAction
}
func (t *InstallAmdRocm) Execute(runtime connector.Runtime) error {
si := runtime.GetSystemInfo()
if !si.IsLinux() || !si.IsUbuntu() || !(si.IsUbuntuVersionEqual(connector.Ubuntu2204) || si.IsUbuntuVersionEqual(connector.Ubuntu2404)) {
return nil
}
amdGPUExists, err := utils.HasAmdIGPU(runtime)
if err != nil {
return err
}
// skip rocm install
if !amdGPUExists {
return nil
}
rocmV, _ := utils.RocmVersion()
min := semver.MustParse("7.1.1")
if rocmV != nil && rocmV.LessThan(min) {
return fmt.Errorf("detected ROCm version %s, which is lower than required %s; please uninstall existing ROCm/AMDGPU components before installation with command: olares-cli amdgpu uninstall", rocmV.Original(), min.Original())
}
if rocmV != nil && rocmV.GreaterThan(min) {
logger.Warnf("Warning: detected ROCm version %s great than maximum tested version %s")
return nil
}
if rocmV != nil && rocmV.Equal(min) {
logger.Infof("detected ROCm version %s, skip rocm install...", min.Original())
return nil
}
// ensure python3-setuptools and python3-wheel
_, _ = runtime.GetRunner().SudoCmd("apt-get update", false, true)
checkPkgs := "dpkg -s python3-setuptools python3-wheel >/dev/null 2>&1 || DEBIAN_FRONTEND=noninteractive apt-get install -y python3-setuptools python3-wheel"
if _, err := runtime.GetRunner().SudoCmd(checkPkgs, false, true); err != nil {
return errors.Wrap(errors.WithStack(err), "failed to install python3-setuptools and python3-wheel")
}
// ensure amdgpu-install exists
if _, err := exec.LookPath("amdgpu-install"); err != nil {
var debURL string
if si.IsUbuntuVersionEqual(connector.Ubuntu2404) {
debURL = "https://repo.radeon.com/amdgpu-install/7.1.1/ubuntu/noble/amdgpu-install_7.1.1.70101-1_all.deb"
} else {
debURL = "https://repo.radeon.com/amdgpu-install/7.1.1/ubuntu/jammy/amdgpu-install_7.1.1.70101-1_all.deb"
}
tmpDeb := path.Join(runtime.GetBaseDir(), cc.PackageCacheDir, "gpu", "amdgpu-install_7.1.1.70101-1_all.deb")
if _, err := runtime.GetRunner().SudoCmd(fmt.Sprintf("install -d -m 0755 %s", filepath.Dir(tmpDeb)), false, true); err != nil {
return err
}
cmd := fmt.Sprintf("sh -c 'wget -O %s %s'", tmpDeb, debURL)
if _, err := runtime.GetRunner().SudoCmd(cmd, false, true); err != nil {
return errors.Wrap(errors.WithStack(err), "failed to download amdgpu-install deb")
}
if _, err := runtime.GetRunner().SudoCmd(fmt.Sprintf("DEBIAN_FRONTEND=noninteractive apt-get install -y %s", tmpDeb), false, true); err != nil {
return errors.Wrap(errors.WithStack(err), "failed to install amdgpu-install deb")
}
}
// run installer for rocm usecase
if _, err := runtime.GetRunner().SudoCmd("amdgpu-install -y --usecase=rocm", false, true); err != nil {
return errors.Wrap(errors.WithStack(err), "failed to install AMD ROCm via amdgpu-install")
}
fmt.Println()
logger.Warn("Warning: To enable ROCm, please reboot your machine after installation.")
return nil
}
type AmdgpuInstallAction struct {
common.KubeAction
}
func (t *AmdgpuInstallAction) Execute(runtime connector.Runtime) error {
if _, err := runtime.GetRunner().SudoCmd("amdgpu-install -y --usecase=rocm", false, true); err != nil {
return errors.Wrap(errors.WithStack(err), "failed to install AMD ROCm via amdgpu-install")
}
return nil
}
type AmdgpuUninstallAction struct {
common.KubeAction
}
func (t *AmdgpuUninstallAction) Execute(runtime connector.Runtime) error {
if _, err := runtime.GetRunner().SudoCmd("amdgpu-install --uninstall -y", false, true); err != nil {
return errors.Wrap(errors.WithStack(err), "failed to uninstall AMD ROCm via amdgpu-install")
}
fmt.Println()
logger.Warn("Warning: Please reboot your machine after uninstall to fully remove ROCm components.")
return nil
}

View File

@@ -81,6 +81,7 @@ func (m *RunPrechecksModule) Init() {
new(NvidiaCardArchChecker),
new(NouveauChecker),
new(CudaChecker),
new(RocmChecker),
}
runPreChecks := &task.LocalTask{
Name: "RunPrechecks",

View File

@@ -372,6 +372,48 @@ func (c *CudaChecker) Check(runtime connector.Runtime) error {
return nil
}
// RocmChecker checks AMD ROCm version for AMD GPU on Ubuntu 22.04/24.04 only.
type RocmChecker struct{}
func (r *RocmChecker) Name() string {
return "ROCm"
}
func (r *RocmChecker) Check(runtime connector.Runtime) error {
if !runtime.GetSystemInfo().IsLinux() {
return nil
}
si := runtime.GetSystemInfo()
if !si.IsUbuntu() || !(si.IsUbuntuVersionEqual(connector.Ubuntu2204) || si.IsUbuntuVersionEqual(connector.Ubuntu2404)) {
return nil
}
// detect AMD GPU presence
amdGPUExists, err := utils.HasAmdIGPU(runtime)
if err != nil {
return err
}
// no AMD GPU found, no need to check rocm
if !amdGPUExists {
return nil
}
curV, err := utils.RocmVersion()
if err != nil && !os.IsNotExist(err) {
return err
}
if os.IsNotExist(err) {
return nil
}
min := semver.MustParse("7.1.1")
if curV.LessThan(min) {
return fmt.Errorf("detected ROCm version %s, which is lower than required %s; please uninstall existing ROCm/AMDGPU components before installation with command: olares-cli amdgpu uninstall", curV.Original(), min.Original())
}
return nil
}
//////////////////////////////////////////////
// precheck - task

View File

@@ -51,10 +51,12 @@ func (d DebianVersion) String() string {
}
const (
Ubuntu20 UbuntuVersion = "20."
Ubuntu22 UbuntuVersion = "22."
Ubuntu24 UbuntuVersion = "24."
Ubuntu25 UbuntuVersion = "25."
Ubuntu20 UbuntuVersion = "20."
Ubuntu22 UbuntuVersion = "22."
Ubuntu24 UbuntuVersion = "24."
Ubuntu25 UbuntuVersion = "25."
Ubuntu2204 UbuntuVersion = "22.04"
Ubuntu2404 UbuntuVersion = "24.04"
Debian9 DebianVersion = "9"
Debian10 DebianVersion = "10"

View File

@@ -3,8 +3,7 @@ package system
import (
"strings"
"github.com/beclab/Olares/cli/pkg/gpu"
"github.com/beclab/Olares/cli/pkg/amdgpu"
"github.com/beclab/Olares/cli/pkg/bootstrap/os"
"github.com/beclab/Olares/cli/pkg/bootstrap/patch"
"github.com/beclab/Olares/cli/pkg/bootstrap/precheck"
@@ -12,6 +11,7 @@ import (
"github.com/beclab/Olares/cli/pkg/container"
"github.com/beclab/Olares/cli/pkg/core/module"
"github.com/beclab/Olares/cli/pkg/daemon"
"github.com/beclab/Olares/cli/pkg/gpu"
"github.com/beclab/Olares/cli/pkg/images"
"github.com/beclab/Olares/cli/pkg/k3s"
"github.com/beclab/Olares/cli/pkg/manifest"
@@ -82,6 +82,7 @@ func (l *linuxPhaseBuilder) build() []module.Module {
addModule(&terminus.WriteReleaseFileModule{}).
addModule(gpuModuleBuilder(func() []module.Module {
return []module.Module{
&amdgpu.InstallAmdRocmModule{},
&gpu.InstallDriversModule{
ManifestModule: manifest.ManifestModule{
Manifest: l.manifestMap,

101
cli/pkg/pipelines/amdgpu.go Normal file
View File

@@ -0,0 +1,101 @@
package pipelines
import (
"strings"
"github.com/beclab/Olares/cli/pkg/amdgpu"
"github.com/beclab/Olares/cli/pkg/common"
"github.com/beclab/Olares/cli/pkg/core/action"
"github.com/beclab/Olares/cli/pkg/core/connector"
"github.com/beclab/Olares/cli/pkg/core/logger"
"github.com/beclab/Olares/cli/pkg/core/module"
"github.com/beclab/Olares/cli/pkg/core/pipeline"
"github.com/beclab/Olares/cli/pkg/core/task"
)
type singleTaskModule struct {
common.KubeModule
name string
act action.Action
}
func (m *singleTaskModule) Init() {
m.Name = m.name
m.Tasks = []task.Interface{
&task.LocalTask{
Name: m.name,
Action: m.act,
},
}
}
func AmdGpuInstall() error {
arg := common.NewArgument()
arg.SetConsoleLog("amdgpuinstall.log", true)
runtime, err := common.NewKubeRuntime(common.AllInOne, *arg)
if err != nil {
return err
}
p := &pipeline.Pipeline{
Name: "InstallAMDGPUDrivers",
Runtime: runtime,
Modules: []module.Module{
&amdgpu.InstallAmdRocmModule{},
},
}
return p.Start()
}
func AmdGpuUninstall() error {
arg := common.NewArgument()
arg.SetConsoleLog("amdgpuuninstall.log", true)
runtime, err := common.NewKubeRuntime(common.AllInOne, *arg)
if err != nil {
return err
}
p := &pipeline.Pipeline{
Name: "UninstallAMDGPUDrivers",
Runtime: runtime,
Modules: []module.Module{
&singleTaskModule{name: "AmdgpuUninstall", act: new(amdgpu.AmdgpuUninstallAction)},
},
}
return p.Start()
}
func AmdGpuStatus() error {
arg := common.NewArgument()
runtime, err := common.NewKubeRuntime(common.AllInOne, *arg)
if err != nil {
return err
}
runtime.SetRunner(
&connector.Runner{
Host: &connector.BaseHost{
Name: common.LocalHost,
Arch: runtime.GetSystemInfo().GetOsArch(),
Os: runtime.GetSystemInfo().GetOsType(),
},
},
)
amdModel, _ := runtime.GetRunner().SudoCmd("lspci | grep -iE 'VGA|3D|Display' | grep -iE 'AMD|ATI' | head -1 || true", false, false)
drvVer, _ := runtime.GetRunner().SudoCmd("modinfo amdgpu 2>/dev/null | awk -F': ' '/^version:/{print $2}' || true", false, false)
rocmVer, _ := runtime.GetRunner().SudoCmd("cat /opt/rocm/.info/version 2>/dev/null || true", false, false)
if strings.TrimSpace(amdModel) != "" {
logger.Infof("AMD GPU: %s", strings.TrimSpace(amdModel))
} else {
logger.Info("AMD GPU: not detected")
}
if strings.TrimSpace(drvVer) != "" {
logger.Infof("AMDGPU driver %s", strings.TrimSpace(drvVer))
} else {
logger.Info("AMDGPU driver version: unknown")
}
if strings.TrimSpace(rocmVer) != "" {
logger.Infof("ROCm version: %s", strings.TrimSpace(rocmVer))
} else {
logger.Info("ROCm version: not installed")
}
return nil
}

View File

@@ -9,6 +9,7 @@ import (
"github.com/beclab/Olares/cli/pkg/core/connector"
"github.com/beclab/Olares/cli/pkg/core/logger"
"github.com/beclab/Olares/cli/pkg/core/task"
"github.com/beclab/Olares/cli/pkg/utils"
)
type WelcomeMessage struct {
@@ -68,6 +69,15 @@ func (t *WelcomeMessage) Execute(runtime connector.Runtime) error {
logger.Infof("Username: %s", t.KubeConf.Arg.User.UserName)
logger.Infof("Password: %s", t.KubeConf.Arg.User.Password)
fmt.Printf("\n------------------------------------------------\n\n\n\n\n")
fmt.Println()
// If AMD GPU on Ubuntu 22.04/24.04, print warning about reboot for ROCm
if si := runtime.GetSystemInfo(); si.IsUbuntu() && (si.IsUbuntuVersionEqual(connector.Ubuntu2204) || si.IsUbuntuVersionEqual(connector.Ubuntu2404)) {
if hasAmd, _ := utils.HasAmdIGPU(runtime); hasAmd {
logger.Warnf("\x1b[31mWarning: To enable ROCm, please reboot your machine after activation.\x1b[0m")
fmt.Println()
}
}
return nil
}

67
cli/pkg/utils/amdgpu.go Normal file
View File

@@ -0,0 +1,67 @@
package utils
import (
"fmt"
"os"
"strings"
"github.com/Masterminds/semver/v3"
"github.com/beclab/Olares/cli/pkg/core/connector"
)
func HasAmdIGPU(execRuntime connector.Runtime) (bool, error) {
// Detect by CPU model names that bundle AMD AI NPU/graphics
targets := []string{
"AMD Ryzen AI Max+ 395",
"AMD Ryzen AI Max 390",
"AMD Ryzen AI Max 385",
"AMD Ryzen AI 9 HX 375",
"AMD Ryzen AI 9 HX 370",
"AMD Ryzen AI 9 365",
}
// try lscpu first: extract 'Model name' field
out, err := execRuntime.GetRunner().SudoCmd("lscpu 2>/dev/null | awk -F': *' '/^Model name/{print $2; exit}' || true", false, false)
if err != nil {
return false, err
}
if out != "" {
lo := strings.ToLower(strings.TrimSpace(out))
for _, t := range targets {
if strings.Contains(lo, strings.ToLower(t)) {
return true, nil
}
}
}
// fallback to /proc/cpuinfo
out, err = execRuntime.GetRunner().SudoCmd("awk -F': *' '/^model name/{print $2; exit}' /proc/cpuinfo 2>/dev/null || true", false, false)
if err != nil {
return false, err
}
if out != "" {
lo := strings.ToLower(strings.TrimSpace(out))
for _, t := range targets {
if strings.Contains(lo, strings.ToLower(t)) {
return true, nil
}
}
}
return false, nil
}
func RocmVersion() (*semver.Version, error) {
const rocmVersionFile = "/opt/rocm/.info/version"
data, err := os.ReadFile(rocmVersionFile)
if err != nil {
// no ROCm installed, nothing to check
if os.IsNotExist(err) {
return nil, err
}
return nil, err
}
curStr := strings.TrimSpace(string(data))
cur, err := semver.NewVersion(curStr)
if err != nil {
return nil, fmt.Errorf("invalid rocm version: %s", curStr)
}
return cur, nil
}

View File

@@ -266,7 +266,7 @@ spec:
containers:
- name: api
image: beclab/bfl:v0.4.38
image: beclab/bfl:v0.4.39
imagePullPolicy: IfNotPresent
securityContext:
runAsUser: 1000

View File

@@ -240,6 +240,7 @@ func (c *Client) getAppListFromData(apps []map[string]interface{}) ([]*AppInfo,
res = append(res, &AppInfo{
ID: genAppID(appSpec),
Name: stringOrEmpty(appSpec["name"]),
RawAppName: stringOrEmpty(appSpec["rawAppName"]),
Namespace: stringOrEmpty(appSpec["namespace"]),
DeploymentName: stringOrEmpty(appSpec["deployment"]),
Owner: stringOrEmpty(appSpec["owner"]),

View File

@@ -11,6 +11,7 @@ import (
type AppInfo struct {
ID string `json:"id"`
Name string `json:"name"`
RawAppName string `json:"rawAppName"`
Namespace string `json:"namespace"`
DeploymentName string `json:"deployment"`
Owner string `json:"owner"`

View File

@@ -146,6 +146,8 @@ type Server struct {
ngxCmd *nginx.Command
ngxTmpl *template.Template
prevNgxConf []byte
}
func (s *Server) init() error {
@@ -906,6 +908,9 @@ func (s *Server) generateStreamServers() ([]StreamServer, error) {
if bflHost == "" {
return nil, fmt.Errorf("can not find bfl service for user=%s", app.Spec.Owner)
}
if p.ExposePort < 1 || p.ExposePort > 65535 {
continue
}
server := StreamServer{
Protocol: p.Protocol,
Port: p.ExposePort,
@@ -929,6 +934,13 @@ func (s *Server) renderAndReload() error {
return err
}
nginxConfig := buf.Bytes()
if bytes.Equal(nginxConfig, s.prevNgxConf) {
klog.Infof("nginx config not changed, skip reload")
return nil
}
s.prevNgxConf = nginxConfig
err = s.testTemplate(nginxConfig)
if err != nil {
return err