Ovirt devel - Jan 2009 - ovirt-server patch for vm failover and actual data set to db

this patch restart vm from crashed host and set to db current data about vm.
please test or criticize it.
 fencing is ipmi , when domain['state'] set 'crashed' i will
reboot host


diff --git a/src/db-omatic/db_omatic.rb b/src/db-omatic/db_omatic.rb
index 4afffb1..bdab392 100755
--- a/src/db-omatic/db_omatic.rb
+++ b/src/db-omatic/db_omatic.rb
@@ -74,6 +76,65 @@ class DbOmatic < Qpid::Qmf::Console
         domain[:synced] = true
     end

+    #find hostname from values['node'] where
values['class_type'] == 'domain'
+    def get_host_id(abank,bbank) #?????????? ???????? ?? ?? ??? ????
- ???????? ????????????? ???? ??????????
+     begin
+	@cached_objects.keys.each do |objkey|
+  	  if @cached_objects[objkey][:agent_bank].to_s == abank and
@cached_objects[objkey][:broker_bank].to_s == bbank and
@cached_objects[objkey][:class_type].to_s == 'node'
+		return Host.find(:first, :conditions => ['hostname ?', at
cached_objects[objkey]["hostname"].to_s]).id
+		    break
+	  end
+	end
+      rescue => ex
+	log("error in get_host_id")
+	log(ex)
+     end
+    end
+
+    def set_host(values,digit)
+      begin
+	vm = Vm.find(:first, :conditions => ['description =
?',values["name"].to_s])
+	if vm and digit
+	vm.host_id = digit
+	vm.save!
+	else
+		log("this vm not exist #{values["name"]}")
+	end
+      rescue => ex
+	puts "error when set_host for #{values["name"]}"
+	puts ex
+      end
+    end
+
+    def start_crashed_vm(vm)
+	task = VmTask.new( :user => 'db-omatic', :task_target => vm,
:action
=> 'start_vm', :state => 'queued')
+	task.save!
+	log("set task for start crashed vm #{vm.id}")
+    end
+
+    def set_domain_stopped(domain)
+      begin
+        vm = Vm.find(:first, :conditions => ['uuid = ?',
domain['uuid']])
+        if vm != nil
+	  curstate = vm.state
+          vm.state = Vm::STATE_STOPPED
+          vm.host_id = nil
+          vm.save
+	  domain['state'] = 'crashed'  # and now i will use ipmi for
reboot
anavailable host - simple fencing
+	  if curstate == Vm::STATE_RUNNING
+            start_crashed_vm(vm)
+	  end
+        else
+          log('vm == nil ')
+        end
+        log("domain  #{domain['id']} already stopped")
+      rescue => ex
+        log("can\'t set domain #{domain['id']} stopped")
+	log(ex)
+      end
+    end
+
+
     def update_host_state(host_info, state)
         db_host = Host.find(:first, :conditions => [ "hostname =
?",
host_info['hostname'] ])
         if db_host

@@ -130,20 +190,36 @@ class DbOmatic < Qpid::Qmf::Console
             end

             domain_state_change = false
-
+	    change_node = false
             obj.properties.each do |key, newval|
                 if values[key.to_s] != newval
                     values[key.to_s] = newval
                     #log "new value for property #{key} : #{newval}"
                     if type == "domain" and key.to_s ==
"state"
-                        domain_state_change = true
+			domain_state_change = true
                     end
+		    if type == "domain" and key.to_s == "node"
+			change_node = true
+		    end
+
                 end
             end

             if domain_state_change
                 update_domain_state(values)
             end
+	if change_node		
+	values.each do |key,val|
+	  if key == 'state' and val == 'running'
+		abank = values['node'].to_s.split('-')[3]
+		bbank = values['node'].to_s.split('-')[4]
+		@@host_id = get_host_id(abank,bbank)
+		set_host(values,@@host_id)
+		log("update node data for #{values['name']}")
+		break
+	  end
+	end
+	end

             if new_object
                 if type == "node"
@@ -187,11 +263,6 @@ class DbOmatic < Qpid::Qmf::Console
         end
     end

-
-    def del_agent(agent)
-        agent_disconnected(agent)
-    end
-
     # This method marks objects associated with the given agent as
timed out/invalid.  Called either
     # when the agent heartbeats out, or we get a del_agent callback.
     def agent_disconnected(agent)
@@ -204,11 +275,13 @@ class DbOmatic < Qpid::Qmf::Console
                 if values[:timed_out] == false
                     if values[:class_type] == 'node'
                         update_host_state(values, Host::STATE_UNAVAILABLE)
+			values[:timed_out] = true
                     elsif values[:class_type] == 'domain'
-                        update_domain_state(values, Vm::STATE_UNREACHABLE)
-                    end
+			set_domain_stopped(values)
+			values[:timed_out] = true
+		@cached_objects.delete(objkey)
+		    end
                 end
-            values[:timed_out] = true
             end
         end
     end
@@ -248,6 +320,7 @@ class DbOmatic < Qpid::Qmf::Console
         db_vm = Vm.find(:all)
         db_vm.each do |vm|
             log "Marking vm #{vm.description} as stopped."
+	    vm.host_id = nil
             vm.state = Vm::STATE_STOPPED
             vm.save
         end




-- 
? ?????????, ???????.

Hello.
are this patch not needed ?
i can rewrite if bad code

-- 
? ?????????, ???????.

Sorry I took so long to get back to you on this one.  There are some
issues with formatting in this patch etc. but mostly I've been thinking
that we need to have this configurable in the WUI.

I'm curious what other people think about rebooting crashed hosts?  Is
this a desired feature?

	Ian

On Wed, 21 Jan 2009 18:51:44 +0300
??????? ??????? <pronix.service at gmail.com> wrote:
> this patch restart vm from crashed host and set to db current data about
vm.
> please test or criticize it.
>  fencing is ipmi , when domain['state'] set 'crashed' i
will reboot host
> 
> 
> diff --git a/src/db-omatic/db_omatic.rb b/src/db-omatic/db_omatic.rb
> index 4afffb1..bdab392 100755
> --- a/src/db-omatic/db_omatic.rb
> +++ b/src/db-omatic/db_omatic.rb
> @@ -74,6 +76,65 @@ class DbOmatic < Qpid::Qmf::Console
>          domain[:synced] = true
>      end
> 
> +    #find hostname from values['node'] where
values['class_type'] == 'domain'
> +    def get_host_id(abank,bbank) #?????????? ???????? ?? ?? ??? ????
> - ???????? ????????????? ???? ??????????
> +     begin
> +	@cached_objects.keys.each do |objkey|
> +  	  if @cached_objects[objkey][:agent_bank].to_s == abank and
> @cached_objects[objkey][:broker_bank].to_s == bbank and
> @cached_objects[objkey][:class_type].to_s == 'node'
> +		return Host.find(:first, :conditions => ['hostname > ?',
at cached_objects[objkey]["hostname"].to_s]).id
> +		    break
> +	  end
> +	end
> +      rescue => ex
> +	log("error in get_host_id")
> +	log(ex)
> +     end
> +    end
> +
> +    def set_host(values,digit)
> +      begin
> +	vm = Vm.find(:first, :conditions => ['description =
?',values["name"].to_s])
> +	if vm and digit
> +	vm.host_id = digit
> +	vm.save!
> +	else
> +		log("this vm not exist #{values["name"]}")
> +	end
> +      rescue => ex
> +	puts "error when set_host for #{values["name"]}"
> +	puts ex
> +      end
> +    end
> +
> +    def start_crashed_vm(vm)
> +	task = VmTask.new( :user => 'db-omatic', :task_target =>
vm, :action
> => 'start_vm', :state => 'queued')
> +	task.save!
> +	log("set task for start crashed vm #{vm.id}")
> +    end
> +
> +    def set_domain_stopped(domain)
> +      begin
> +        vm = Vm.find(:first, :conditions => ['uuid = ?',
domain['uuid']])
> +        if vm != nil
> +	  curstate = vm.state
> +          vm.state = Vm::STATE_STOPPED
> +          vm.host_id = nil
> +          vm.save
> +	  domain['state'] = 'crashed'  # and now i will use ipmi
for reboot
> anavailable host - simple fencing
> +	  if curstate == Vm::STATE_RUNNING
> +            start_crashed_vm(vm)
> +	  end
> +        else
> +          log('vm == nil ')
> +        end
> +        log("domain  #{domain['id']} already stopped")
> +      rescue => ex
> +        log("can\'t set domain #{domain['id']}
stopped")
> +	log(ex)
> +      end
> +    end
> +
> +
>      def update_host_state(host_info, state)
>          db_host = Host.find(:first, :conditions => [ "hostname =
?",
> host_info['hostname'] ])
>          if db_host
> 
> @@ -130,20 +190,36 @@ class DbOmatic < Qpid::Qmf::Console
>              end
> 
>              domain_state_change = false
> -
> +	    change_node = false
>              obj.properties.each do |key, newval|
>                  if values[key.to_s] != newval
>                      values[key.to_s] = newval
>                      #log "new value for property #{key} :
#{newval}"
>                      if type == "domain" and key.to_s ==
"state"
> -                        domain_state_change = true
> +			domain_state_change = true
>                      end
> +		    if type == "domain" and key.to_s == "node"
> +			change_node = true
> +		    end
> +
>                  end
>              end
> 
>              if domain_state_change
>                  update_domain_state(values)
>              end
> +	if change_node		
> +	values.each do |key,val|
> +	  if key == 'state' and val == 'running'
> +		abank = values['node'].to_s.split('-')[3]
> +		bbank = values['node'].to_s.split('-')[4]
> +		@@host_id = get_host_id(abank,bbank)
> +		set_host(values,@@host_id)
> +		log("update node data for #{values['name']}")
> +		break
> +	  end
> +	end
> +	end
> 
>              if new_object
>                  if type == "node"
> @@ -187,11 +263,6 @@ class DbOmatic < Qpid::Qmf::Console
>          end
>      end
> 
> -
> -    def del_agent(agent)
> -        agent_disconnected(agent)
> -    end
> -
>      # This method marks objects associated with the given agent as
> timed out/invalid.  Called either
>      # when the agent heartbeats out, or we get a del_agent callback.
>      def agent_disconnected(agent)
> @@ -204,11 +275,13 @@ class DbOmatic < Qpid::Qmf::Console
>                  if values[:timed_out] == false
>                      if values[:class_type] == 'node'
>                          update_host_state(values, Host::STATE_UNAVAILABLE)
> +			values[:timed_out] = true
>                      elsif values[:class_type] == 'domain'
> -                        update_domain_state(values, Vm::STATE_UNREACHABLE)
> -                    end
> +			set_domain_stopped(values)
> +			values[:timed_out] = true
> +		@cached_objects.delete(objkey)
> +		    end
>                  end
> -            values[:timed_out] = true
>              end
>          end
>      end
> @@ -248,6 +320,7 @@ class DbOmatic < Qpid::Qmf::Console
>          db_vm = Vm.find(:all)
>          db_vm.each do |vm|
>              log "Marking vm #{vm.description} as stopped."
> +	    vm.host_id = nil
>              vm.state = Vm::STATE_STOPPED
>              vm.save
>          end
> 
> 
> 
> 
> -- 
> ? ?????????, ???????.
> 
> _______________________________________________
> Ovirt-devel mailing list
> Ovirt-devel at redhat.com
> https://www.redhat.com/mailman/listinfo/ovirt-devel

i split patch and reformat it
first part is  for set actual data to db about vm
second add wui button for enable/disable restart vm if crash host and code
for do it (without fencing mechanism)

if source corrupt by mail agent i attach both files

*PATCH.1*

diff --git a/src/db-omatic/db_omatic.rb b/src/db-omatic/db_omatic.rb
index 4afffb1..c499610 100755
--- a/src/db-omatic/db_omatic.rb
+++ b/src/db-omatic/db_omatic.rb
@@ -74,6 +74,64 @@ class DbOmatic < Qpid::Qmf::Console
         domain[:synced] = true
     end

+    #find hostname from values['node'] where
values['class_type'] ='domain'
+    def get_host_id(abank,bbank)
+     begin
+       @cached_objects.keys.each do |objkey|
+         if @cached_objects[objkey][:agent_bank].to_s == abank and
@cached_objects[objkey][:broker_bank].to_s == bbank and
@cached_objects[objkey][:class_type].to_s == 'node'
+               return Host.find(:first, :conditions => ['hostname
?', at cached_objects[objkey]["hostname"].to_s]).id
+                   break
+         end
+       end
+      rescue => ex
+       log("error in get_host_id")
+       log(ex)
+     end
+    end
+
+    def set_host(values,digit)
+      begin
+       vm = Vm.find(:first, :conditions => ['description
?',values["name"].to_s])
+       if vm and digit
+       vm.host_id = digit
+       vm.save!
+       else
+               log("this vm not exist #{values["name"]}")
+       end
+      rescue => ex
+       puts "error when set_host for #{values["name"]}"
+       puts ex
+      end
+    end
+
+    def start_crashed_vm(vm)
+       task = VmTask.new( :user => 'db-omatic', :task_target =>
vm, :action
=> 'start_vm', :state => 'queued')
+       task.save!
+       log("set task for start crashed vm #{vm.id}")
+    end
+
+    def set_domain_stopped(domain)
+      begin
+        vm = Vm.find(:first, :conditions => ['uuid = ?',
domain['uuid']])
+        if vm != nil
+         curstate = vm.state
+          vm.state = Vm::STATE_STOPPED
+          vm.host_id = nil
+          vm.save
+         domain['state'] = 'crashed'  # and now i will use ipmi
for reboot
anavailable host - simple fencing
+         if curstate == Vm::STATE_RUNNING and vm.ha # vm.ha true or false
+            start_crashed_vm(vm)
+         end
+        else
+          log('vm == nil ')
+        end
+        log("domain  #{domain['id']} already stopped")
+      rescue => ex
+        log("can\'t set domain #{domain['id']} stopped")
+       log(ex)
+      end
+    end
+
     def update_host_state(host_info, state)
         db_host = Host.find(:first, :conditions => [ "hostname =
?",
host_info['hostname'] ])
         if db_host
@@ -131,6 +189,7 @@ class DbOmatic < Qpid::Qmf::Console

             domain_state_change = false

+        change_node = false
             obj.properties.each do |key, newval|
                 if values[key.to_s] != newval
                     values[key.to_s] = newval
@@ -138,12 +197,30 @@ class DbOmatic < Qpid::Qmf::Console
                     if type == "domain" and key.to_s ==
"state"
                         domain_state_change = true
                     end
+                   if type == "domain" and key.to_s ==
"node"
+                       change_node = true
+                   end
+
+
                 end
             end

             if domain_state_change
                 update_domain_state(values)
             end
+       if change_node
+       values.each do |key,val|
+         if key == 'state' and val == 'running'
+               abank = values['node'].to_s.split('-')[3]
+               bbank = values['node'].to_s.split('-')[4]
+               @@host_id = get_host_id(abank,bbank)
+               set_host(values,@@host_id)
+               log("update node data for #{values['name']}")
+               break
+         end
+       end
+       end
+

             if new_object
                 if type == "node"
@@ -187,11 +264,6 @@ class DbOmatic < Qpid::Qmf::Console
         end
     end

-
-    def del_agent(agent)
-        agent_disconnected(agent)
-    end
-
     # This method marks objects associated with the given agent as timed
out/invalid.  Called either
     # when the agent heartbeats out, or we get a del_agent callback.
     def agent_disconnected(agent)
@@ -205,8 +277,10 @@ class DbOmatic < Qpid::Qmf::Console
                     if values[:class_type] == 'node'
                         update_host_state(values, Host::STATE_UNAVAILABLE)
                     elsif values[:class_type] == 'domain'
-                        update_domain_state(values, Vm::STATE_UNREACHABLE)
-                    end
+                       set_domain_stopped(values)
+                       values[:timed_out] = true
+                   @cached_objects.delete(objkey)
+                   end
                 end
             values[:timed_out] = true
             end
@@ -248,6 +322,7 @@ class DbOmatic < Qpid::Qmf::Console
         db_vm = Vm.find(:all)
         db_vm.each do |vm|
             log "Marking vm #{vm.description} as stopped."
+        vm.host_id = nil
             vm.state = Vm::STATE_STOPPED
             vm.save
         end

*PATCH.2*

diff --git a/src/app/controllers/vm_controller.rb
b/src/app/controllers/vm_controller.rb
index 56501fd..0f43680 100644
--- a/src/app/controllers/vm_controller.rb
+++ b/src/app/controllers/vm_controller.rb
@@ -20,11 +20,22 @@ require 'socket'

 class VmController < ApplicationController
   # GETs should be safe (see
http://www.w3.org/2001/tag/doc/whenToUseGet.html)
-  verify :method => :post, :only => [ :destroy, :create, :update ],
+  verify :method => :post, :only => [ :destroy, :create, :update ,
:change_ha_vm],
          :redirect_to => { :controller => 'dashboard' }

   before_filter :pre_vm_action, :only => [:vm_action, :cancel_queued_tasks,
:console]

+   def change_ha_vm
+     vm = Vm.find_by_id(params[:id])
+     if vm.ha
+       then vm.ha = false
+       else vm.ha = true
+     end
+     vm.save!
+     alert = "vm was change ha successfully "
+     render :json => { :object => "vm", :success => true,
:alert => alert
}
+   end
+
   def index
       roles = "('" +
          
Permission::roles_for_privilege(Permission::PRIV_VIEW).join("',
'") +
diff --git a/src/app/views/vm/show.rhtml b/src/app/views/vm/show.rhtml
index f361131..4809bbe 100644
--- a/src/app/views/vm/show.rhtml
+++ b/src/app/views/vm/show.rhtml
@@ -29,6 +29,9 @@
         </a>
       <% end -%>
     <% end %>
+    <a href="#" onClick="change_ha_vm()"
rel="facebox[.bolder]">
+      <%= image_tag "icon_x.png" %> Change HA
+    </a>
     <a href="#confirm_cancel" 
rel="facebox[.bolder]">
       <%= image_tag "icon_x.png" %> Cancel queued tasks
     </a>
@@ -40,6 +43,19 @@
 <%= confirmation_dialog("confirm_cancel", "Are you
sure?",
"cancel_queued_tasks()") %>
 <%= confirmation_dialog("confirm_delete", "Are you
sure?", "delete_vm()")
%>
 <script type="text/javascript">
+   function change_ha_vm()
+   {
+     $(document).trigger('close.facebox');
+     $.post('<%= url_for :controller => "vm", :action =>
"change_ha_vm",
:id => @vm %>',
+            {x: 1},
+             function(data,status){
+               $("#vms_grid").flexReload();
+               if (data.alert) {
+                 $.jGrowl(data.alert);
+               }
+               empty_summary('vms_selection', 'Virtual
Machine');
+             }, 'json');
+   }
   function cancel_queued_tasks()
   {
     $(document).trigger('close.facebox');

diff --git a/src/db/migrate/006_create_vms.rb
b/src/db/migrate/006_create_vms.rb
index 74794b6..a5460f0 100644
--- a/src/db/migrate/006_create_vms.rb
+++ b/src/db/migrate/006_create_vms.rb
@@ -34,6 +34,7 @@ class CreateVms < ActiveRecord::Migration
       t.string  :boot_device,    :null => false
       t.integer :vnc_port
       t.integer :lock_version,   :default => 0
+      t.boolean :ha , :default => true
     end
     execute "alter table vms add constraint fk_vms_hosts
              foreign key (host_id) references hosts(id)"
-- 
? ?????????, ???????.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://listman.redhat.com/archives/ovirt-devel/attachments/20090130/6f6b42e9/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: part.1.patch
Type: text/x-diff
Size: 4870 bytes
Desc: not available
URL:
<http://listman.redhat.com/archives/ovirt-devel/attachments/20090130/6f6b42e9/attachment.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: part.2.patch
Type: text/x-diff
Size: 2858 bytes
Desc: not available
URL:
<http://listman.redhat.com/archives/ovirt-devel/attachments/20090130/6f6b42e9/attachment-0001.bin>