You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@cloudstack.apache.org by Alex Ough <al...@sungard.com> on 2013/12/02 16:45:18 UTC
Re: A question on vm migrations when hosts are set into a maintenance mode.
All,
It's been a while since this review was requested, so can anyone review
this to move on?
Thanks in advance.
Alex Ough
On Fri, Nov 22, 2013 at 11:51 AM, Alex Ough <al...@sungard.com> wrote:
> Thanks to help from many developers,
> I sent my first review request in the cloudstack,
> https://reviews.apache.org/r/15763/,
> so please take a look at it and let me know if there is anything
> missing/incorrect.
>
> Thanks again.
> Alex Ough
>
>
> On Thu, Nov 21, 2013 at 10:09 AM, Alex Ough <al...@sungard.com> wrote:
>
>> Hi Wido,
>>
>> Related with the 'disk_offering.cache_mode',
>> I found that there is a sql file to add that column, which is 'cloudstack/
>> developer/target/db/db/schema-430to440.sql'.
>> But after running it manually, I have another error saying that the
>> 'cache_mode' cannot be null, so I just changed the column to allow null to
>> move on.
>>
>> So you may include the sql file and allow null in that column.
>>
>> Thanks
>> Alex Ough
>>
>>
>> On Wed, Nov 20, 2013 at 3:45 PM, Alex Huang <Al...@citrix.com>wrote:
>>
>>> Wido,
>>>
>>>
>>>
>>> Looks like you didn’t update your schema file or forgot to add a schema
>>> upgrade file in master? See the error described by Alex below.
>>>
>>>
>>>
>>> @Alex
>>>
>>> You can probably remove the cachemode field in the DiskOfferingVO for now. Or revert 1edaa36 on your local repo to keep moving forward.
>>>
>>>
>>>
>>> --Alex
>>>
>>>
>>>
>>> *From:* Alex Ough [mailto:alex.ough@sungard.com]
>>> *Sent:* Wednesday, November 20, 2013 9:56 AM
>>> *To:* Alex Huang
>>> *Cc:* dev@cloudstack.apache.org
>>>
>>> *Subject:* Re: A question on vm migrations when hosts are set into a
>>> maintenance mode.
>>>
>>>
>>>
>>> Hi Alex,
>>>
>>>
>>>
>>> It looks like you moved the 'ExecuteInSequence' to the vm level from the
>>> management server, which seems to be ok, but at this time I cannot test it
>>>
>>> because when I try to start up the management server, it fails because
>>> of database schema error, which is 'Unknown column
>>> 'disk_offering.cache_mode' in 'field list' even if I re-built the database.
>>>
>>>
>>>
>>> Is the schema change part of your changes or something other developer
>>> changed?
>>>
>>>
>>>
>>> If that is from any other developer, can you fix this?
>>>
>>>
>>>
>>> Thanks
>>>
>>> Alex Ough
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>> On Tue, Nov 19, 2013 at 7:25 PM, Alex Huang <Al...@citrix.com>
>>> wrote:
>>>
>>> Alex,
>>>
>>>
>>>
>>> Can you do a pull from master and see if my fix fits your needs.
>>> Unfortunately, I’m on the road and couldn’t do an actual test of it.
>>>
>>>
>>>
>>> --Alex
>>>
>>>
>>>
>>> *From:* Alex Huang
>>> *Sent:* Tuesday, November 19, 2013 4:45 AM
>>> *To:* 'Alex Ough'; dev@cloudstack.apache.org
>>> *Subject:* RE: A question on vm migrations when hosts are set into a
>>> maintenance mode.
>>>
>>>
>>>
>>> Alex,
>>>
>>>
>>>
>>> Sorry for the late reply. Been travelling the last couple of weeks.
>>> I’ll look into this today.
>>>
>>>
>>>
>>> --Alex
>>>
>>>
>>>
>>> *From:* Alex Ough [mailto:alex.ough@sungard.com <al...@sungard.com>]
>>>
>>> *Sent:* Monday, November 18, 2013 6:17 AM
>>> *To:* dev@cloudstack.apache.org
>>> *Cc:* Alex Huang
>>> *Subject:* Re: A question on vm migrations when hosts are set into a
>>> maintenance mode.
>>>
>>>
>>>
>>> Thank Parasanna & Sebastien,
>>>
>>> I also got his email and sent an email.
>>>
>>> Waiting for his reply...
>>>
>>>
>>>
>>> Thanks
>>>
>>> Alex Ough
>>>
>>>
>>>
>>> On Sat, Nov 16, 2013 at 3:05 PM, Sebastien Goasguen <ru...@gmail.com>
>>> wrote:
>>>
>>> cc Alex Huang to get his attention:
>>>
>>>
>>>
>>> On Nov 15, 2013, at 10:17 PM, Prasanna Santhanam <ts...@apache.org> wrote:
>>>
>>> > Alex, Could you just do a git blame on the file and copy the emails of
>>> > people who changed that bit of code? They may be able to help if Cc-ed
>>> > directly.
>>> >
>>> > Thanks,
>>> >
>>> > On Fri, Nov 15, 2013 at 01:49:07PM -0600, Alex Ough wrote:
>>> >> I hate to sending the same emails over and over again, but I really
>>> need to
>>> >> finalize this feature to be included in the next code freeze because
>>> this
>>> >> feature is very critical in our inside project.
>>> >>
>>> >> Anyone who can help, please?
>>> >> Thanks
>>> >> Alex Ough
>>> >>
>>> >>
>>> >> On Thu, Nov 14, 2013 at 1:27 PM, Alex Ough <al...@sungard.com>
>>> wrote:
>>> >>
>>> >>> Not sure if Alex Huang checked this, but can anyone help to resolve
>>> this?
>>> >>>
>>> >>> Thanks
>>> >>> Alex Ough
>>> >>>
>>> >>>
>>> >>> On Wed, Nov 13, 2013 at 11:39 AM, Alex Ough <al...@sungard.com>
>>> wrote:
>>> >>>
>>> >>>> It sounds a little scary...
>>> >>>>
>>> >>>> I looked at the history and found these.
>>> >>>>
>>> >>>> 8/9/ : file moved to engine by Alex Huang
>>> >>>> 9/16 : '_mgmtServer.getExecuteInSequence()' changed to
>>> >>>> 'getExecuteInSequence()' by Alex Huang
>>> >>>>
>>> >>>>
>>> >>>> Hi Alex Huang,
>>> >>>> I'm not sure if you're aware of this, but can you check this for me?
>>> >>>>
>>> >>>> Thanks
>>> >>>> Alex Ough
>>> >>>>
>>> >>>>
>>> >>>>
>>> >>>> On Wed, Nov 13, 2013 at 11:18 AM, Marcus Sorensen <
>>> shadowsor@gmail.com>wrote:
>>> >>>>
>>> >>>>> I'm not sure. I know in the past when I've seen files change
>>> locations
>>> >>>>> it has also clobbered updates to that file. Someone branched, did
>>> the
>>> >>>>> reorganization work, and merged, while in-between the original file
>>> >>>>> changed.
>>> >>>>>
>>> >>>>> On Wed, Nov 13, 2013 at 9:21 AM, Alex Ough <al...@sungard.com>
>>> >>>>> wrote:
>>> >>>>>> All,
>>> >>>>>>
>>> >>>>>> While merging my changes to 4.3 branch, I found that the option,
>>> >>>>>> 'execute.in.sequence.hypervisor.commands' is NOT used in
>>> >>>>> Start/Stop/Copy
>>> >>>>>> commands in 'VirtualMachineManagerImpl.java' any more as below.
>>> >>>>>>
>>> >>>>>>
>>> >>>>>> *StopCommand stop = new StopCommand(vm, getExecuteInSequence());*
>>> >>>>>>
>>> >>>>>> *protected boolean getExecuteInSequence() {*
>>> >>>>>> * return false;*
>>> >>>>>> *}*
>>> >>>>>>
>>> >>>>>> As you see in the above, the function, 'getExecuteInSequence',
>>> just
>>> >>>>> returns
>>> >>>>>> false instead of getting the value from the global variable.
>>> >>>>>>
>>> >>>>>> And one more change is that the file has been moved to
>>> >>>>>> 'engine/orchestration/src/com/cloud/vm' from
>>> 'server/src/com/cloud/vm'.
>>> >>>>>>
>>> >>>>>> Am I missing something related with this or do we stop supporting
>>> this
>>> >>>>>> option in 4.3?
>>> >>>>>> I'm a little confused, so please help me resolve this.
>>> >>>>>>
>>> >>>>>> Thanks
>>> >>>>>> Alex Ough
>>> >>>>>>
>>> >>>>>>
>>> >>>>>>
>>> >>>>>>
>>> >>>>>> On Tue, Nov 12, 2013 at 4:20 PM, Alex Ough <alex.ough@sungard.com
>>> >
>>> >>>>> wrote:
>>> >>>>>>
>>> >>>>>>> Thanks a lot for your confirmation, Marcus.
>>> >>>>>>> I'll create a review request unless anyone has an objection.
>>> >>>>>>>
>>> >>>>>>> Thanks
>>> >>>>>>> Alex Ough
>>> >>>>>>>
>>> >>>>>>>
>>> >>>>>>> On Tue, Nov 12, 2013 at 3:37 PM, Marcus Sorensen <
>>> shadowsor@gmail.com
>>> >>>>>> wrote:
>>> >>>>>>>
>>> >>>>>>>> I have done parallel KVM migrations without issue, it's
>>> "supposed to
>>> >>>>>>>> work". Really I think it's in the same boat as parallel
>>> start/stop.
>>> >>>>> It
>>> >>>>>>>> should work, but the config option is there just in case. I
>>> think we
>>> >>>>>>>> should add it.
>>> >>>>>>>>
>>> >>>>>>>> On Thu, Oct 3, 2013 at 11:41 AM, Chip Childers
>>> >>>>>>>> <ch...@sungard.com> wrote:
>>> >>>>>>>>> On Thu, Oct 03, 2013 at 11:44:46AM -0500, Alex Ough wrote:
>>> >>>>>>>>>> I'm not sure what else commands 'MigrateCommand' actually
>>> execute
>>> >>>>> in
>>> >>>>>>>>>> addition to 'Start/Stop/CopyCommand', but can we include
>>> >>>>>>>> 'MigrateCommand'
>>> >>>>>>>>>> if it consists of only those 3 commands?
>>> >>>>>>>>>>
>>> >>>>>>>>>> Thanks
>>> >>>>>>>>>> Alex Ough
>>> >>>>>>>>>
>>> >>>>>>>>> In the case of VMware, the migrate command is executed via the
>>> >>>>>>>>> MigrateVMTask that's part of the VMware SDK (see
>>> >>>>>>>>>
>>> >>>>>
>>> vmware-base/src/com/cloud/hypervisor/vmware/mo/VirtualMachineMO.java).
>>> >>>>>>>>>
>>> >>>>>>>>> For VMware, I know that vCenter will queue and process
>>> concurrent
>>> >>>>>>>>> requests for migrations. Specifically, it will throttle the
>>> >>>>> migrations
>>> >>>>>>>>> happening, based on it's internal concurrency constraints, but
>>> the
>>> >>>>> task
>>> >>>>>>>>> queue will still accept more connections. Obviously the risk
>>> are
>>> >>>>> the
>>> >>>>>>>>> VMware layer tasks timing out if it takes too long for the task
>>> >>>>> queue to
>>> >>>>>>>>> complete.
>>> >>>>>>>>>
>>> >>>>>>>>> As for XenServer, it's happening in what appears to be a
>>> similar
>>> >>>>> way
>>> >>>>>>>>> (although the source host is the target for the migration API
>>> >>>>> call).
>>> >>>>>>>>>
>>> >>>>>>>>> Check
>>> >>>>>>>>>
>>> >>>>>>>>
>>> >>>>>
>>> plugins/hypervisors/xen/src/com/cloud/hypervisor/xen/resource/CitrixResourceBase.java.
>>> >>>>>>>>>
>>> >>>>>>>>> I'm not familiar enough with XenServer's concurrency model for
>>> >>>>>>>>> migrations. Any experts know the answer to if it can handle
>>> >>>>> concurrency
>>> >>>>>>>>> in a stable way?
>>> >>>>>>>>>
>>> >>>>>>>>> With KVM, it's obviously executing via the agent. Similarly to
>>> >>>>>>>>> XenServer, I'm not familiar enough to know about concurrent
>>> >>>>> operations.
>>> >>>>>>>>>
>>> >>>>>>>>> So do the HV experts on the list have any opinions about
>>> XenServer
>>> >>>>> and
>>> >>>>>>>>> KVM migration concurrency?
>>> >>>>>>>>>
>>> >>>>>>>>> -chip
>>> >>>>>>>>>
>>> >>>>>>>>>
>>> >>>>>>>>
>>> >>>>>>>>
>>> >>>>>>>
>>> >>>>>
>>> >>>>>
>>> >>>>
>>> >>>
>>> >
>>> > --
>>> > Prasanna.,
>>> >
>>> > ------------------------
>>> > Powered by BigRock.com
>>> >
>>>
>>>
>>>
>>>
>>>
>>
>>
>
Re: A question on vm migrations when hosts are set into a maintenance mode.
Posted by Alex Ough <al...@sungard.com>.
Thank Koushik, and I modified what you pointed out (for 'MigrateCommand'
text) and updated the diffs.
Let me know if there is anything missing/incorrect.
Thanks
Alex Ough
On Tue, Dec 3, 2013 at 2:46 AM, Koushik Das <ko...@citrix.com> wrote:
> I have posted my review comments. Except for a few minor comments changes
> look good to me.
>
> -Koushik
>
> On 02-Dec-2013, at 9:15 PM, Alex Ough <al...@sungard.com> wrote:
>
> > All,
> >
> > It's been a while since this review was requested, so can anyone review
> > this to move on?
> >
> > Thanks in advance.
> > Alex Ough
> >
> >
> > On Fri, Nov 22, 2013 at 11:51 AM, Alex Ough <al...@sungard.com>
> wrote:
> >
> >> Thanks to help from many developers,
> >> I sent my first review request in the cloudstack,
> >> https://reviews.apache.org/r/15763/,
> >> so please take a look at it and let me know if there is anything
> >> missing/incorrect.
> >>
> >> Thanks again.
> >> Alex Ough
> >>
> >>
> >> On Thu, Nov 21, 2013 at 10:09 AM, Alex Ough <al...@sungard.com>
> wrote:
> >>
> >>> Hi Wido,
> >>>
> >>> Related with the 'disk_offering.cache_mode',
> >>> I found that there is a sql file to add that column, which is
> 'cloudstack/
> >>> developer/target/db/db/schema-430to440.sql'.
> >>> But after running it manually, I have another error saying that the
> >>> 'cache_mode' cannot be null, so I just changed the column to allow
> null to
> >>> move on.
> >>>
> >>> So you may include the sql file and allow null in that column.
> >>>
> >>> Thanks
> >>> Alex Ough
> >>>
> >>>
> >>> On Wed, Nov 20, 2013 at 3:45 PM, Alex Huang <Alex.Huang@citrix.com
> >wrote:
> >>>
> >>>> Wido,
> >>>>
> >>>>
> >>>>
> >>>> Looks like you didn’t update your schema file or forgot to add a
> schema
> >>>> upgrade file in master? See the error described by Alex below.
> >>>>
> >>>>
> >>>>
> >>>> @Alex
> >>>>
> >>>> You can probably remove the cachemode field in the DiskOfferingVO for
> now. Or revert 1edaa36 on your local repo to keep moving forward.
> >>>>
> >>>>
> >>>>
> >>>> --Alex
> >>>>
> >>>>
> >>>>
> >>>> *From:* Alex Ough [mailto:alex.ough@sungard.com]
> >>>> *Sent:* Wednesday, November 20, 2013 9:56 AM
> >>>> *To:* Alex Huang
> >>>> *Cc:* dev@cloudstack.apache.org
> >>>>
> >>>> *Subject:* Re: A question on vm migrations when hosts are set into a
> >>>> maintenance mode.
> >>>>
> >>>>
> >>>>
> >>>> Hi Alex,
> >>>>
> >>>>
> >>>>
> >>>> It looks like you moved the 'ExecuteInSequence' to the vm level from
> the
> >>>> management server, which seems to be ok, but at this time I cannot
> test it
> >>>>
> >>>> because when I try to start up the management server, it fails because
> >>>> of database schema error, which is 'Unknown column
> >>>> 'disk_offering.cache_mode' in 'field list' even if I re-built the
> database.
> >>>>
> >>>>
> >>>>
> >>>> Is the schema change part of your changes or something other developer
> >>>> changed?
> >>>>
> >>>>
> >>>>
> >>>> If that is from any other developer, can you fix this?
> >>>>
> >>>>
> >>>>
> >>>> Thanks
> >>>>
> >>>> Alex Ough
> >>>>
> >>>>
> >>>>
> >>>>
> >>>>
> >>>>
> >>>>
> >>>> On Tue, Nov 19, 2013 at 7:25 PM, Alex Huang <Al...@citrix.com>
> >>>> wrote:
> >>>>
> >>>> Alex,
> >>>>
> >>>>
> >>>>
> >>>> Can you do a pull from master and see if my fix fits your needs.
> >>>> Unfortunately, I’m on the road and couldn’t do an actual test of it.
> >>>>
> >>>>
> >>>>
> >>>> --Alex
> >>>>
> >>>>
> >>>>
> >>>> *From:* Alex Huang
> >>>> *Sent:* Tuesday, November 19, 2013 4:45 AM
> >>>> *To:* 'Alex Ough'; dev@cloudstack.apache.org
> >>>> *Subject:* RE: A question on vm migrations when hosts are set into a
> >>>> maintenance mode.
> >>>>
> >>>>
> >>>>
> >>>> Alex,
> >>>>
> >>>>
> >>>>
> >>>> Sorry for the late reply. Been travelling the last couple of weeks.
> >>>> I’ll look into this today.
> >>>>
> >>>>
> >>>>
> >>>> --Alex
> >>>>
> >>>>
> >>>>
> >>>> *From:* Alex Ough [mailto:alex.ough@sungard.com <
> alex.ough@sungard.com>]
> >>>>
> >>>> *Sent:* Monday, November 18, 2013 6:17 AM
> >>>> *To:* dev@cloudstack.apache.org
> >>>> *Cc:* Alex Huang
> >>>> *Subject:* Re: A question on vm migrations when hosts are set into a
> >>>> maintenance mode.
> >>>>
> >>>>
> >>>>
> >>>> Thank Parasanna & Sebastien,
> >>>>
> >>>> I also got his email and sent an email.
> >>>>
> >>>> Waiting for his reply...
> >>>>
> >>>>
> >>>>
> >>>> Thanks
> >>>>
> >>>> Alex Ough
> >>>>
> >>>>
> >>>>
> >>>> On Sat, Nov 16, 2013 at 3:05 PM, Sebastien Goasguen <runseb@gmail.com
> >
> >>>> wrote:
> >>>>
> >>>> cc Alex Huang to get his attention:
> >>>>
> >>>>
> >>>>
> >>>> On Nov 15, 2013, at 10:17 PM, Prasanna Santhanam <ts...@apache.org>
> wrote:
> >>>>
> >>>>> Alex, Could you just do a git blame on the file and copy the emails
> of
> >>>>> people who changed that bit of code? They may be able to help if
> Cc-ed
> >>>>> directly.
> >>>>>
> >>>>> Thanks,
> >>>>>
> >>>>> On Fri, Nov 15, 2013 at 01:49:07PM -0600, Alex Ough wrote:
> >>>>>> I hate to sending the same emails over and over again, but I really
> >>>> need to
> >>>>>> finalize this feature to be included in the next code freeze because
> >>>> this
> >>>>>> feature is very critical in our inside project.
> >>>>>>
> >>>>>> Anyone who can help, please?
> >>>>>> Thanks
> >>>>>> Alex Ough
> >>>>>>
> >>>>>>
> >>>>>> On Thu, Nov 14, 2013 at 1:27 PM, Alex Ough <al...@sungard.com>
> >>>> wrote:
> >>>>>>
> >>>>>>> Not sure if Alex Huang checked this, but can anyone help to resolve
> >>>> this?
> >>>>>>>
> >>>>>>> Thanks
> >>>>>>> Alex Ough
> >>>>>>>
> >>>>>>>
> >>>>>>> On Wed, Nov 13, 2013 at 11:39 AM, Alex Ough <alex.ough@sungard.com
> >
> >>>> wrote:
> >>>>>>>
> >>>>>>>> It sounds a little scary...
> >>>>>>>>
> >>>>>>>> I looked at the history and found these.
> >>>>>>>>
> >>>>>>>> 8/9/ : file moved to engine by Alex Huang
> >>>>>>>> 9/16 : '_mgmtServer.getExecuteInSequence()' changed to
> >>>>>>>> 'getExecuteInSequence()' by Alex Huang
> >>>>>>>>
> >>>>>>>>
> >>>>>>>> Hi Alex Huang,
> >>>>>>>> I'm not sure if you're aware of this, but can you check this for
> me?
> >>>>>>>>
> >>>>>>>> Thanks
> >>>>>>>> Alex Ough
> >>>>>>>>
> >>>>>>>>
> >>>>>>>>
> >>>>>>>> On Wed, Nov 13, 2013 at 11:18 AM, Marcus Sorensen <
> >>>> shadowsor@gmail.com>wrote:
> >>>>>>>>
> >>>>>>>>> I'm not sure. I know in the past when I've seen files change
> >>>> locations
> >>>>>>>>> it has also clobbered updates to that file. Someone branched, did
> >>>> the
> >>>>>>>>> reorganization work, and merged, while in-between the original
> file
> >>>>>>>>> changed.
> >>>>>>>>>
> >>>>>>>>> On Wed, Nov 13, 2013 at 9:21 AM, Alex Ough <
> alex.ough@sungard.com>
> >>>>>>>>> wrote:
> >>>>>>>>>> All,
> >>>>>>>>>>
> >>>>>>>>>> While merging my changes to 4.3 branch, I found that the option,
> >>>>>>>>>> 'execute.in.sequence.hypervisor.commands' is NOT used in
> >>>>>>>>> Start/Stop/Copy
> >>>>>>>>>> commands in 'VirtualMachineManagerImpl.java' any more as below.
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>> *StopCommand stop = new StopCommand(vm,
> getExecuteInSequence());*
> >>>>>>>>>>
> >>>>>>>>>> *protected boolean getExecuteInSequence() {*
> >>>>>>>>>> * return false;*
> >>>>>>>>>> *}*
> >>>>>>>>>>
> >>>>>>>>>> As you see in the above, the function, 'getExecuteInSequence',
> >>>> just
> >>>>>>>>> returns
> >>>>>>>>>> false instead of getting the value from the global variable.
> >>>>>>>>>>
> >>>>>>>>>> And one more change is that the file has been moved to
> >>>>>>>>>> 'engine/orchestration/src/com/cloud/vm' from
> >>>> 'server/src/com/cloud/vm'.
> >>>>>>>>>>
> >>>>>>>>>> Am I missing something related with this or do we stop
> supporting
> >>>> this
> >>>>>>>>>> option in 4.3?
> >>>>>>>>>> I'm a little confused, so please help me resolve this.
> >>>>>>>>>>
> >>>>>>>>>> Thanks
> >>>>>>>>>> Alex Ough
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>> On Tue, Nov 12, 2013 at 4:20 PM, Alex Ough <
> alex.ough@sungard.com
> >>>>>
> >>>>>>>>> wrote:
> >>>>>>>>>>
> >>>>>>>>>>> Thanks a lot for your confirmation, Marcus.
> >>>>>>>>>>> I'll create a review request unless anyone has an objection.
> >>>>>>>>>>>
> >>>>>>>>>>> Thanks
> >>>>>>>>>>> Alex Ough
> >>>>>>>>>>>
> >>>>>>>>>>>
> >>>>>>>>>>> On Tue, Nov 12, 2013 at 3:37 PM, Marcus Sorensen <
> >>>> shadowsor@gmail.com
> >>>>>>>>>> wrote:
> >>>>>>>>>>>
> >>>>>>>>>>>> I have done parallel KVM migrations without issue, it's
> >>>> "supposed to
> >>>>>>>>>>>> work". Really I think it's in the same boat as parallel
> >>>> start/stop.
> >>>>>>>>> It
> >>>>>>>>>>>> should work, but the config option is there just in case. I
> >>>> think we
> >>>>>>>>>>>> should add it.
> >>>>>>>>>>>>
> >>>>>>>>>>>> On Thu, Oct 3, 2013 at 11:41 AM, Chip Childers
> >>>>>>>>>>>> <ch...@sungard.com> wrote:
> >>>>>>>>>>>>> On Thu, Oct 03, 2013 at 11:44:46AM -0500, Alex Ough wrote:
> >>>>>>>>>>>>>> I'm not sure what else commands 'MigrateCommand' actually
> >>>> execute
> >>>>>>>>> in
> >>>>>>>>>>>>>> addition to 'Start/Stop/CopyCommand', but can we include
> >>>>>>>>>>>> 'MigrateCommand'
> >>>>>>>>>>>>>> if it consists of only those 3 commands?
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>> Thanks
> >>>>>>>>>>>>>> Alex Ough
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> In the case of VMware, the migrate command is executed via
> the
> >>>>>>>>>>>>> MigrateVMTask that's part of the VMware SDK (see
> >>>>>>>>>>>>>
> >>>>>>>>>
> >>>> vmware-base/src/com/cloud/hypervisor/vmware/mo/VirtualMachineMO.java).
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> For VMware, I know that vCenter will queue and process
> >>>> concurrent
> >>>>>>>>>>>>> requests for migrations. Specifically, it will throttle the
> >>>>>>>>> migrations
> >>>>>>>>>>>>> happening, based on it's internal concurrency constraints,
> but
> >>>> the
> >>>>>>>>> task
> >>>>>>>>>>>>> queue will still accept more connections. Obviously the risk
> >>>> are
> >>>>>>>>> the
> >>>>>>>>>>>>> VMware layer tasks timing out if it takes too long for the
> task
> >>>>>>>>> queue to
> >>>>>>>>>>>>> complete.
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> As for XenServer, it's happening in what appears to be a
> >>>> similar
> >>>>>>>>> way
> >>>>>>>>>>>>> (although the source host is the target for the migration API
> >>>>>>>>> call).
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> Check
> >>>>>>>>>>>>>
> >>>>>>>>>>>>
> >>>>>>>>>
> >>>>
> plugins/hypervisors/xen/src/com/cloud/hypervisor/xen/resource/CitrixResourceBase.java.
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> I'm not familiar enough with XenServer's concurrency model
> for
> >>>>>>>>>>>>> migrations. Any experts know the answer to if it can handle
> >>>>>>>>> concurrency
> >>>>>>>>>>>>> in a stable way?
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> With KVM, it's obviously executing via the agent. Similarly
> to
> >>>>>>>>>>>>> XenServer, I'm not familiar enough to know about concurrent
> >>>>>>>>> operations.
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> So do the HV experts on the list have any opinions about
> >>>> XenServer
> >>>>>>>>> and
> >>>>>>>>>>>>> KVM migration concurrency?
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> -chip
> >>>>>>>>>>>>>
> >>>>>>>>>>>>>
> >>>>>>>>>>>>
> >>>>>>>>>>>>
> >>>>>>>>>>>
> >>>>>>>>>
> >>>>>>>>>
> >>>>>>>>
> >>>>>>>
> >>>>>
> >>>>> --
> >>>>> Prasanna.,
> >>>>>
> >>>>> ------------------------
> >>>>> Powered by BigRock.com
> >>>>>
> >>>>
> >>>>
> >>>>
> >>>>
> >>>>
> >>>
> >>>
> >>
>
>
>
Re: A question on vm migrations when hosts are set into a
maintenance mode.
Posted by Koushik Das <ko...@citrix.com>.
I have posted my review comments. Except for a few minor comments changes look good to me.
-Koushik
On 02-Dec-2013, at 9:15 PM, Alex Ough <al...@sungard.com> wrote:
> All,
>
> It's been a while since this review was requested, so can anyone review
> this to move on?
>
> Thanks in advance.
> Alex Ough
>
>
> On Fri, Nov 22, 2013 at 11:51 AM, Alex Ough <al...@sungard.com> wrote:
>
>> Thanks to help from many developers,
>> I sent my first review request in the cloudstack,
>> https://reviews.apache.org/r/15763/,
>> so please take a look at it and let me know if there is anything
>> missing/incorrect.
>>
>> Thanks again.
>> Alex Ough
>>
>>
>> On Thu, Nov 21, 2013 at 10:09 AM, Alex Ough <al...@sungard.com> wrote:
>>
>>> Hi Wido,
>>>
>>> Related with the 'disk_offering.cache_mode',
>>> I found that there is a sql file to add that column, which is 'cloudstack/
>>> developer/target/db/db/schema-430to440.sql'.
>>> But after running it manually, I have another error saying that the
>>> 'cache_mode' cannot be null, so I just changed the column to allow null to
>>> move on.
>>>
>>> So you may include the sql file and allow null in that column.
>>>
>>> Thanks
>>> Alex Ough
>>>
>>>
>>> On Wed, Nov 20, 2013 at 3:45 PM, Alex Huang <Al...@citrix.com>wrote:
>>>
>>>> Wido,
>>>>
>>>>
>>>>
>>>> Looks like you didn’t update your schema file or forgot to add a schema
>>>> upgrade file in master? See the error described by Alex below.
>>>>
>>>>
>>>>
>>>> @Alex
>>>>
>>>> You can probably remove the cachemode field in the DiskOfferingVO for now. Or revert 1edaa36 on your local repo to keep moving forward.
>>>>
>>>>
>>>>
>>>> --Alex
>>>>
>>>>
>>>>
>>>> *From:* Alex Ough [mailto:alex.ough@sungard.com]
>>>> *Sent:* Wednesday, November 20, 2013 9:56 AM
>>>> *To:* Alex Huang
>>>> *Cc:* dev@cloudstack.apache.org
>>>>
>>>> *Subject:* Re: A question on vm migrations when hosts are set into a
>>>> maintenance mode.
>>>>
>>>>
>>>>
>>>> Hi Alex,
>>>>
>>>>
>>>>
>>>> It looks like you moved the 'ExecuteInSequence' to the vm level from the
>>>> management server, which seems to be ok, but at this time I cannot test it
>>>>
>>>> because when I try to start up the management server, it fails because
>>>> of database schema error, which is 'Unknown column
>>>> 'disk_offering.cache_mode' in 'field list' even if I re-built the database.
>>>>
>>>>
>>>>
>>>> Is the schema change part of your changes or something other developer
>>>> changed?
>>>>
>>>>
>>>>
>>>> If that is from any other developer, can you fix this?
>>>>
>>>>
>>>>
>>>> Thanks
>>>>
>>>> Alex Ough
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> On Tue, Nov 19, 2013 at 7:25 PM, Alex Huang <Al...@citrix.com>
>>>> wrote:
>>>>
>>>> Alex,
>>>>
>>>>
>>>>
>>>> Can you do a pull from master and see if my fix fits your needs.
>>>> Unfortunately, I’m on the road and couldn’t do an actual test of it.
>>>>
>>>>
>>>>
>>>> --Alex
>>>>
>>>>
>>>>
>>>> *From:* Alex Huang
>>>> *Sent:* Tuesday, November 19, 2013 4:45 AM
>>>> *To:* 'Alex Ough'; dev@cloudstack.apache.org
>>>> *Subject:* RE: A question on vm migrations when hosts are set into a
>>>> maintenance mode.
>>>>
>>>>
>>>>
>>>> Alex,
>>>>
>>>>
>>>>
>>>> Sorry for the late reply. Been travelling the last couple of weeks.
>>>> I’ll look into this today.
>>>>
>>>>
>>>>
>>>> --Alex
>>>>
>>>>
>>>>
>>>> *From:* Alex Ough [mailto:alex.ough@sungard.com <al...@sungard.com>]
>>>>
>>>> *Sent:* Monday, November 18, 2013 6:17 AM
>>>> *To:* dev@cloudstack.apache.org
>>>> *Cc:* Alex Huang
>>>> *Subject:* Re: A question on vm migrations when hosts are set into a
>>>> maintenance mode.
>>>>
>>>>
>>>>
>>>> Thank Parasanna & Sebastien,
>>>>
>>>> I also got his email and sent an email.
>>>>
>>>> Waiting for his reply...
>>>>
>>>>
>>>>
>>>> Thanks
>>>>
>>>> Alex Ough
>>>>
>>>>
>>>>
>>>> On Sat, Nov 16, 2013 at 3:05 PM, Sebastien Goasguen <ru...@gmail.com>
>>>> wrote:
>>>>
>>>> cc Alex Huang to get his attention:
>>>>
>>>>
>>>>
>>>> On Nov 15, 2013, at 10:17 PM, Prasanna Santhanam <ts...@apache.org> wrote:
>>>>
>>>>> Alex, Could you just do a git blame on the file and copy the emails of
>>>>> people who changed that bit of code? They may be able to help if Cc-ed
>>>>> directly.
>>>>>
>>>>> Thanks,
>>>>>
>>>>> On Fri, Nov 15, 2013 at 01:49:07PM -0600, Alex Ough wrote:
>>>>>> I hate to sending the same emails over and over again, but I really
>>>> need to
>>>>>> finalize this feature to be included in the next code freeze because
>>>> this
>>>>>> feature is very critical in our inside project.
>>>>>>
>>>>>> Anyone who can help, please?
>>>>>> Thanks
>>>>>> Alex Ough
>>>>>>
>>>>>>
>>>>>> On Thu, Nov 14, 2013 at 1:27 PM, Alex Ough <al...@sungard.com>
>>>> wrote:
>>>>>>
>>>>>>> Not sure if Alex Huang checked this, but can anyone help to resolve
>>>> this?
>>>>>>>
>>>>>>> Thanks
>>>>>>> Alex Ough
>>>>>>>
>>>>>>>
>>>>>>> On Wed, Nov 13, 2013 at 11:39 AM, Alex Ough <al...@sungard.com>
>>>> wrote:
>>>>>>>
>>>>>>>> It sounds a little scary...
>>>>>>>>
>>>>>>>> I looked at the history and found these.
>>>>>>>>
>>>>>>>> 8/9/ : file moved to engine by Alex Huang
>>>>>>>> 9/16 : '_mgmtServer.getExecuteInSequence()' changed to
>>>>>>>> 'getExecuteInSequence()' by Alex Huang
>>>>>>>>
>>>>>>>>
>>>>>>>> Hi Alex Huang,
>>>>>>>> I'm not sure if you're aware of this, but can you check this for me?
>>>>>>>>
>>>>>>>> Thanks
>>>>>>>> Alex Ough
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> On Wed, Nov 13, 2013 at 11:18 AM, Marcus Sorensen <
>>>> shadowsor@gmail.com>wrote:
>>>>>>>>
>>>>>>>>> I'm not sure. I know in the past when I've seen files change
>>>> locations
>>>>>>>>> it has also clobbered updates to that file. Someone branched, did
>>>> the
>>>>>>>>> reorganization work, and merged, while in-between the original file
>>>>>>>>> changed.
>>>>>>>>>
>>>>>>>>> On Wed, Nov 13, 2013 at 9:21 AM, Alex Ough <al...@sungard.com>
>>>>>>>>> wrote:
>>>>>>>>>> All,
>>>>>>>>>>
>>>>>>>>>> While merging my changes to 4.3 branch, I found that the option,
>>>>>>>>>> 'execute.in.sequence.hypervisor.commands' is NOT used in
>>>>>>>>> Start/Stop/Copy
>>>>>>>>>> commands in 'VirtualMachineManagerImpl.java' any more as below.
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> *StopCommand stop = new StopCommand(vm, getExecuteInSequence());*
>>>>>>>>>>
>>>>>>>>>> *protected boolean getExecuteInSequence() {*
>>>>>>>>>> * return false;*
>>>>>>>>>> *}*
>>>>>>>>>>
>>>>>>>>>> As you see in the above, the function, 'getExecuteInSequence',
>>>> just
>>>>>>>>> returns
>>>>>>>>>> false instead of getting the value from the global variable.
>>>>>>>>>>
>>>>>>>>>> And one more change is that the file has been moved to
>>>>>>>>>> 'engine/orchestration/src/com/cloud/vm' from
>>>> 'server/src/com/cloud/vm'.
>>>>>>>>>>
>>>>>>>>>> Am I missing something related with this or do we stop supporting
>>>> this
>>>>>>>>>> option in 4.3?
>>>>>>>>>> I'm a little confused, so please help me resolve this.
>>>>>>>>>>
>>>>>>>>>> Thanks
>>>>>>>>>> Alex Ough
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> On Tue, Nov 12, 2013 at 4:20 PM, Alex Ough <alex.ough@sungard.com
>>>>>
>>>>>>>>> wrote:
>>>>>>>>>>
>>>>>>>>>>> Thanks a lot for your confirmation, Marcus.
>>>>>>>>>>> I'll create a review request unless anyone has an objection.
>>>>>>>>>>>
>>>>>>>>>>> Thanks
>>>>>>>>>>> Alex Ough
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> On Tue, Nov 12, 2013 at 3:37 PM, Marcus Sorensen <
>>>> shadowsor@gmail.com
>>>>>>>>>> wrote:
>>>>>>>>>>>
>>>>>>>>>>>> I have done parallel KVM migrations without issue, it's
>>>> "supposed to
>>>>>>>>>>>> work". Really I think it's in the same boat as parallel
>>>> start/stop.
>>>>>>>>> It
>>>>>>>>>>>> should work, but the config option is there just in case. I
>>>> think we
>>>>>>>>>>>> should add it.
>>>>>>>>>>>>
>>>>>>>>>>>> On Thu, Oct 3, 2013 at 11:41 AM, Chip Childers
>>>>>>>>>>>> <ch...@sungard.com> wrote:
>>>>>>>>>>>>> On Thu, Oct 03, 2013 at 11:44:46AM -0500, Alex Ough wrote:
>>>>>>>>>>>>>> I'm not sure what else commands 'MigrateCommand' actually
>>>> execute
>>>>>>>>> in
>>>>>>>>>>>>>> addition to 'Start/Stop/CopyCommand', but can we include
>>>>>>>>>>>> 'MigrateCommand'
>>>>>>>>>>>>>> if it consists of only those 3 commands?
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> Thanks
>>>>>>>>>>>>>> Alex Ough
>>>>>>>>>>>>>
>>>>>>>>>>>>> In the case of VMware, the migrate command is executed via the
>>>>>>>>>>>>> MigrateVMTask that's part of the VMware SDK (see
>>>>>>>>>>>>>
>>>>>>>>>
>>>> vmware-base/src/com/cloud/hypervisor/vmware/mo/VirtualMachineMO.java).
>>>>>>>>>>>>>
>>>>>>>>>>>>> For VMware, I know that vCenter will queue and process
>>>> concurrent
>>>>>>>>>>>>> requests for migrations. Specifically, it will throttle the
>>>>>>>>> migrations
>>>>>>>>>>>>> happening, based on it's internal concurrency constraints, but
>>>> the
>>>>>>>>> task
>>>>>>>>>>>>> queue will still accept more connections. Obviously the risk
>>>> are
>>>>>>>>> the
>>>>>>>>>>>>> VMware layer tasks timing out if it takes too long for the task
>>>>>>>>> queue to
>>>>>>>>>>>>> complete.
>>>>>>>>>>>>>
>>>>>>>>>>>>> As for XenServer, it's happening in what appears to be a
>>>> similar
>>>>>>>>> way
>>>>>>>>>>>>> (although the source host is the target for the migration API
>>>>>>>>> call).
>>>>>>>>>>>>>
>>>>>>>>>>>>> Check
>>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>
>>>> plugins/hypervisors/xen/src/com/cloud/hypervisor/xen/resource/CitrixResourceBase.java.
>>>>>>>>>>>>>
>>>>>>>>>>>>> I'm not familiar enough with XenServer's concurrency model for
>>>>>>>>>>>>> migrations. Any experts know the answer to if it can handle
>>>>>>>>> concurrency
>>>>>>>>>>>>> in a stable way?
>>>>>>>>>>>>>
>>>>>>>>>>>>> With KVM, it's obviously executing via the agent. Similarly to
>>>>>>>>>>>>> XenServer, I'm not familiar enough to know about concurrent
>>>>>>>>> operations.
>>>>>>>>>>>>>
>>>>>>>>>>>>> So do the HV experts on the list have any opinions about
>>>> XenServer
>>>>>>>>> and
>>>>>>>>>>>>> KVM migration concurrency?
>>>>>>>>>>>>>
>>>>>>>>>>>>> -chip
>>>>>>>>>>>>>
>>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>
>>>>> --
>>>>> Prasanna.,
>>>>>
>>>>> ------------------------
>>>>> Powered by BigRock.com
>>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>