Correction du lancement du service onenode

Le service onenode ne se lance pas a cause d'une
boucle dans les dépendances de services.

Pour régler le problème on le lance après la multi-user.target

De plus au moment ou le service se lance opennebula n'est pas
complètement lancé alors le script essaye d'ouvrir des connections
pendant 20 secondes avant de remonter un problème.

Enfin on garde une liste des machines qui sont "running" avant l'arrêt
du serveur pour pouvoir les relance proprement.

Contribution de Cadoles (htts://www.cadoles.com)

ref #20338 @6h

init/onenode.service

@@ -2,6 +2,7 @@
 Description=OpenNebula Node starter
 After=opennebula.service opennebula-sunstone.service libvirt-bin.service
 ConditionPathExists=/etc/eole/release
+After=multi-user.target
 
 [Service]
 Type=oneshot
@@ -10,8 +11,8 @@ Environment=ENDPOINT=http://127.0.0.1:2633/RPC2
 TimeoutSec=1min
 RemainAfterExit=yes
 Restart=no
-ExecStart=/usr/share/eole/sbin/onevm-all -w -c ${CREDS} -e ${ENDPOINT} -a "resume"
+ExecStart=/usr/share/eole/sbin/onevm-all -t 20 -w -c ${CREDS} -e ${ENDPOINT} -a "resume"
-ExecStop=/usr/share/eole/sbin/onevm-all -w -c ${CREDS} -e ${ENDPOINT} -a "suspend"
+ExecStop=/usr/share/eole/sbin/onevm-all -t 20 -w -c ${CREDS} -e ${ENDPOINT} -a "suspend"
 
 [Install]
 WantedBy=multi-user.target

scripts/onevm-all

@@ -5,6 +5,11 @@
 ##############################################################################
 ONE_LOCATION=ENV["ONE_LOCATION"]
 USER=ENV["user"]
+RUNVMFILE="/var/lib/one/running.bck"
+TIMEOUT=20
+
+# oneadmin user flag value
+USERFLAG=-2
 
 if !ONE_LOCATION
     RUBY_LIB_LOCATION="/usr/lib/one/ruby"
@@ -55,13 +60,16 @@ end
 # AIM: Suspend a virtual machine
 #
 def _do_suspend(vm, wait)
+    fd = File.open(RUNVMFILE,'a')
     if vm.status == "runn"
         puts("Suspending #{vm.name} ...")
+        fd.write("#{vm.id}\n")
         vm.suspend
         if wait
             _wait(vm, "susp")
         end
     end
+    fd.close
 end
 
 #
@@ -69,28 +77,34 @@ end
 # PARAM: OpenNebula::VirtualMachine object
 # AIM: Resum a suspended virtual machines
 #
-def _do_resume(vm, wait)
+def _do_resume(vm, wait, force=FALSE)
+    if force
+      vm.resume
+    else
       if vm.status == "susp"
         puts("Resume on #{vm.name}")
         vm.resume
-#    elsif vm.status == 'save'
+        #    elsif vm.status == 'save'
-#      puts("Recover on #{vm.name}")
+        #      puts("Recover on #{vm.name}")
-#      # Try to recover VM with retry action
+        #      # Try to recover VM with retry action
-#      vm.recover(2)
+        #      vm.recover(2)
-#      vm.resume
+        #      vm.resume
       elsif vm.status == 'unkn'
         puts("Resume on #{vm.name}")
         vm.resume
       else
         return -1
       end
+    end
+
     if wait
       _wait(vm, "runn")
     end
 end
 
 
-options = {:creds => nil, :action => nil, :endpoint => nil}
+options = {:creds => nil, :action => nil, :endpoint => nil,
+           :timeout => nil}
 
 parser = OptionParser.new do|opts|
   opts.banner = "Usage: #{File.basename(__FILE__)} [options]"
@@ -106,6 +120,10 @@ parser = OptionParser.new do|opts|
     options[:endpoint] = value;
   end
 
+  opts.on('-t', '--timeout timeout', 'Timeout for opennebula connection') do |value|
+      options[:timeout] = value.to_i;
+  end
+
   opts.on('-w', '--wait', 'Wait for action ends') do |w|
       options[:wait] = w
   end
@@ -114,6 +132,8 @@ parser = OptionParser.new do|opts|
     puts opts
     exit
   end
+
+
 end
 
 parser.parse!
@@ -133,6 +153,10 @@ if not options[:endpoint]
   options[:endpoint] = "http://#{ip}:2633/RPC2"
 end
 
+if not options[:timeout]
+  options[:timeout] = TIMEOUT
+end
+
 # Actions
 SUPPORTED = ['status', 'boot', 'resume', 'shutdown', 'suspend']
 
@@ -154,13 +178,25 @@ end
 begin
   client = Client.new(CREDENTIALS, options[:endpoint])
 
-  vm_pool = VirtualMachinePool.new(client, -1)
+  vm_pool = VirtualMachinePool.new(client, USERFLAG)
+
+  if File.exist?(RUNVMFILE)
+    running_vms = File.open(RUNVMFILE,'r')
+  else
+    running_vms = []
+  end
 
   rc = vm_pool.info
-  if OpenNebula.is_error?(rc)
+  cnt = 0
+  while OpenNebula.is_error?(rc)
+    if cnt == options[:timeout]
         puts rc.message
         exit(-1)
     end
+    rc = vm_pool.info
+    sleep(1)
+    cnt += 1
+  end
 
   vm_pool.each do |vm|
     case options[:action]
@@ -175,11 +211,18 @@ begin
     when "suspend"
         _do_suspend(vm, options[:wait])
     when "resume"
-        _do_resume(vm, options[:wait])
+      force = FALSE
+      if running_vms.include?('vm.id')
+        force = TRUE
+      end
+      _do_resume(vm, options[:wait], force)
     else
       puts("#{vm.name}\t#{vm.status}")
     end
   end
+  if options[:action] == "resume"
+    File.truncate(RUNVMFILE, 0)
+  end
 rescue Exception => e
   puts e.message
   exit(-1)