Re: Cloudstack guest password

[Craig Dunn]

Hi all,

think i`m getting somewhere with it now.

currently when I try to change the password it says NetworkManager is not
running

Jul 28 11:51:17 centos8-base-m1Dd4 cloud-set-guest-password[886]: Error:
NetworkManager is not running.
Jul 28 11:51:17 centos8-base-m1Dd4 cloud[924]: Unable to determine the
password server, falling back to data-server
Jul 28 11:51:17 centos8-base-m1Dd4 cloud[925]: Sending request to password
server at data-server
Jul 28 11:51:17 centos8-base-m1Dd4 cloud[936]: Failed to send request to
password server at data-server
Jul 28 11:51:17 centos8-base-m1Dd4 cloud[939]: Did not need to change
password.

i`m guessing this means its not running on boot when the script is running
as once i'm logged if I check the service status it says it running,
however I have noticed that under CGroup the is a dhclient line in my
working Centos7 Template which doesnt appear in my new Centos8 Template:

CGroup: /system.slice/NetworkManager.service
   ├─774 /usr/sbin/NetworkManager --no-daemon
   └─903 /sbin/dhclient -d -q -sf /usr/libexec/nm-dhcp-helper -pf
/var/run/dhclient-eno16777984.pid -lf
/var/lib/NetworkManager/dhclient-dcf53092-0072-3182-bb91-c33e31d619e3-eno16777984.lease
-cf /var/lib/NetworkManager/dhc...

not sure if that is whats wrong at the minute more looking at differences
between working and not working.

On Fri, 24 Jul 2020 at 14:13, Craig Dunn  wrote:

> thanks Andrija,
>
> I`m having vpn trouble atm so cant get onto the VR to check but, if I run
> the wget command manually against the VR I get back the password which was
> shown in the UI.
>
> I`m guessing the VR has done its but its just not being applied to the VM
> itself for some reason.
>
>
>
> On Fri, 24 Jul 2020 at 12:20, Andrija Panic 
> wrote:
>
>> When you change the password for a VM (while VM is stopped) - ACS will
>> store the password (in readable form) inside the VR in
>> /var/cache/cloud/password- - so if you see the password here (as
>> given in the UI) - that means that nobody fetched the password from the VR
>> - but if it's missing (replaced with word "saved") that means that the
>> script/cloud-init already downloaded the password (but it's question if it
>> was applied or not on the OS/VM itself)
>>
>> Hope that helps
>>
>> On Fri, 24 Jul 2020 at 12:34, Craig Dunn > .invalid>
>> wrote:
>>
>> > Hey all,
>> >
>> > So, spent yesterday messing around with it, if I run wget -q -t 3 -T 20
>> -O
>> > - --header "DomU_Request: send_my_password" $192.168.81.1:8080. Replace
>> > $PASSWORD_SERVER manually it shows nothing.
>> >
>> > [image: image.png]
>> > but this behaviour seems normal as a working template does not respond
>> > either.
>> >
>> > I have also noticed that the leases file in /var/lib/dhclient doesnt
>> > generate till you run dhclient (this is fine I can figure that out
>> later)
>> > but running reset password from cloudstack doesn't actually change it.
>> Only
>> > only thing I have done is a symbolic link on the script from the init.d
>> > folder to rc0.d folder (I had to do this with ubuntu so assumed I would
>> > here too)
>> >
>> > any troubleshooting tips anyone can offer?
>> >
>> > Thanks
>> >
>> > On Thu, 23 Jul 2020 at 12:27, Andrija Panic 
>> > wrote:
>> >
>> >> cloud-init has the "plugin" for cloudstack, so it "behaves" well with
>> it.
>> >>
>> >> best,
>> >>
>> >> On Thu, 23 Jul 2020 at 12:48, Craig Dunn > >> .invalid>
>> >> wrote:
>> >>
>> >> > Thanks Andrija i`ll have a look into that, does the platform need to
>> >> > support it, or does cloudstack support it by default?
>> >> >
>> >> > @Vivek strangely its generated a lease file in /var/lib/dhclient (not
>> >> sure
>> >> > if I did anything to force it) I have tried resetting both manually
>> and
>> >> via
>> >> > the UI with no changes
>> >> >
>> >> > On Thu, 23 Jul 2020 at 11:22, Andrija Panic > >
>> >> > wrote:
>> >> >
>> >> > > that script used to work only with initd and not systemd, so better
>> >> > invest
>> >> > > some time in cloud-init, and achieve the same thing (and more if
>> >> needed).
>> >> > > You can i.e. download the
>> http://dl.openvm.eu/cloudstack/macchinina/
>> >> > > template
>> >> > > and see how the cloud-init is configured there (afaik, it uses
>> >> > cloud-init)
>> >> > >
>> >> > > Best,
>> >> > >
>> >> > > On Thu, 23 Jul 2020 at 12:17, Vivek Kumar > >> > > .invalid>
>> >> > > wrote:
>> >> > >
>> >> > > > That won’t help because it fetch the password from router . Can
>> you
>> >> > just
>> >> > > > run “dhclient” and check the lease folder wether you are getting
>> >> lease
>> >> > > file
>> >> > > > generated or not.
>> >> > > >
>> >> > > > Vivek Kumar
>> >> > > > Manager - Cloud & DevOps
>> >> > > > IndiQus Technologies
>> >> > > > 24*7  O +91 11 4055 1411  |   M +91 7503460090
>> >> > > > www.indiqus.com 
>> >> > > >
>> >> > > > This message is intended only for the use of the individual or
>> >> entity
>> >> > to
>> >> > > > which it is addressed and may c

System vm template

[Ajit Singh]

Hi

Can anyone guide me on how to register or install system vm templates for
vsphere and is it mandatory to reboot acs management servers after system
vm templates installation or registration.

Any link would also help

Thanks
Ajit

Re: Cloudstack guest password

[Richard Lawley]

Do you have a redundant network (2 VRs)?  If so, you cannot
communicate with the password server on the shared .1 IP - you need to
use the VR's unique IP address (this is why the scripts are trying to
find the DHCP server address - this is the only way to find this IP).

Also note that there's a long-standing bug where ACS only stores the
password on the VR with the lowest ID, rather than the one which is
active, so if you've got v-5-VM and v-6-VM but 6 is master, your
password reset won't work.

On Tue, 28 Jul 2020 at 16:59, Craig Dunn
 wrote:
>
> Hi all,
>
> think i`m getting somewhere with it now.
>
> currently when I try to change the password it says NetworkManager is not
> running
>
> Jul 28 11:51:17 centos8-base-m1Dd4 cloud-set-guest-password[886]: Error:
> NetworkManager is not running.
> Jul 28 11:51:17 centos8-base-m1Dd4 cloud[924]: Unable to determine the
> password server, falling back to data-server
> Jul 28 11:51:17 centos8-base-m1Dd4 cloud[925]: Sending request to password
> server at data-server
> Jul 28 11:51:17 centos8-base-m1Dd4 cloud[936]: Failed to send request to
> password server at data-server
> Jul 28 11:51:17 centos8-base-m1Dd4 cloud[939]: Did not need to change
> password.
>
> i`m guessing this means its not running on boot when the script is running
> as once i'm logged if I check the service status it says it running,
> however I have noticed that under CGroup the is a dhclient line in my
> working Centos7 Template which doesnt appear in my new Centos8 Template:
>
> CGroup: /system.slice/NetworkManager.service
>├─774 /usr/sbin/NetworkManager --no-daemon
>└─903 /sbin/dhclient -d -q -sf /usr/libexec/nm-dhcp-helper -pf
> /var/run/dhclient-eno16777984.pid -lf
> /var/lib/NetworkManager/dhclient-dcf53092-0072-3182-bb91-c33e31d619e3-eno16777984.lease
> -cf /var/lib/NetworkManager/dhc...
>
> not sure if that is whats wrong at the minute more looking at differences
> between working and not working.
>
> On Fri, 24 Jul 2020 at 14:13, Craig Dunn  wrote:
>
> > thanks Andrija,
> >
> > I`m having vpn trouble atm so cant get onto the VR to check but, if I run
> > the wget command manually against the VR I get back the password which was
> > shown in the UI.
> >
> > I`m guessing the VR has done its but its just not being applied to the VM
> > itself for some reason.
> >
> >
> >
> > On Fri, 24 Jul 2020 at 12:20, Andrija Panic 
> > wrote:
> >
> >> When you change the password for a VM (while VM is stopped) - ACS will
> >> store the password (in readable form) inside the VR in
> >> /var/cache/cloud/password- - so if you see the password here (as
> >> given in the UI) - that means that nobody fetched the password from the VR
> >> - but if it's missing (replaced with word "saved") that means that the
> >> script/cloud-init already downloaded the password (but it's question if it
> >> was applied or not on the OS/VM itself)
> >>
> >> Hope that helps
> >>
> >> On Fri, 24 Jul 2020 at 12:34, Craig Dunn  >> .invalid>
> >> wrote:
> >>
> >> > Hey all,
> >> >
> >> > So, spent yesterday messing around with it, if I run wget -q -t 3 -T 20
> >> -O
> >> > - --header "DomU_Request: send_my_password" $192.168.81.1:8080. Replace
> >> > $PASSWORD_SERVER manually it shows nothing.
> >> >
> >> > [image: image.png]
> >> > but this behaviour seems normal as a working template does not respond
> >> > either.
> >> >
> >> > I have also noticed that the leases file in /var/lib/dhclient doesnt
> >> > generate till you run dhclient (this is fine I can figure that out
> >> later)
> >> > but running reset password from cloudstack doesn't actually change it.
> >> Only
> >> > only thing I have done is a symbolic link on the script from the init.d
> >> > folder to rc0.d folder (I had to do this with ubuntu so assumed I would
> >> > here too)
> >> >
> >> > any troubleshooting tips anyone can offer?
> >> >
> >> > Thanks
> >> >
> >> > On Thu, 23 Jul 2020 at 12:27, Andrija Panic 
> >> > wrote:
> >> >
> >> >> cloud-init has the "plugin" for cloudstack, so it "behaves" well with
> >> it.
> >> >>
> >> >> best,
> >> >>
> >> >> On Thu, 23 Jul 2020 at 12:48, Craig Dunn  >> >> .invalid>
> >> >> wrote:
> >> >>
> >> >> > Thanks Andrija i`ll have a look into that, does the platform need to
> >> >> > support it, or does cloudstack support it by default?
> >> >> >
> >> >> > @Vivek strangely its generated a lease file in /var/lib/dhclient (not
> >> >> sure
> >> >> > if I did anything to force it) I have tried resetting both manually
> >> and
> >> >> via
> >> >> > the UI with no changes
> >> >> >
> >> >> > On Thu, 23 Jul 2020 at 11:22, Andrija Panic  >> >
> >> >> > wrote:
> >> >> >
> >> >> > > that script used to work only with initd and not systemd, so better
> >> >> > invest
> >> >> > > some time in cloud-init, and achieve the same thing (and more if
> >> >> needed).
> >> >> > > You can i.e. download the
> >> http://dl.openvm.eu/cloudstack/macchinina/
> >> >> > > template
> >> >> > > and see how the cloud-init is

RE: System vm template

[Corey, Mike]

Ajit,

Once you create your Zone with the VMware resources (advanced configuration) 
you have to enable the Zone.  If your storage, mounts, and networking are all 
correctly configured the two system VM instances will deploy into vCenter 
automatically by CloudStack.

Make sure that your systemVM Template (vSphere) has a status of "Download 
Complete" and Ready is "Yes" before you enable the zone though.

Mike




-Original Message-
From: Ajit Singh  
Sent: Tuesday, July 28, 2020 12:05 PM
To: users@cloudstack.apache.org
Subject: System vm template

Hi

Can anyone guide me on how to register or install system vm templates for
vsphere and is it mandatory to reboot acs management servers after system
vm templates installation or registration.

Any link would also help

Thanks
Ajit

Re: System vm template

[Ajit Singh]

Thanks mike

I did checked systemvm template (vsphere)
It shows status as download complete and ready but it’s not showing the
size as systemvm template (Xenserver) shows

Regards
Ajit

On Tue, 28 Jul 2020 at 10:36 PM, Corey, Mike  wrote:

> Ajit,
>
> Once you create your Zone with the VMware resources (advanced
> configuration) you have to enable the Zone.  If your storage, mounts, and
> networking are all correctly configured the two system VM instances will
> deploy into vCenter automatically by CloudStack.
>
> Make sure that your systemVM Template (vSphere) has a status of "Download
> Complete" and Ready is "Yes" before you enable the zone though.
>
> Mike
>
>
>
>
> -Original Message-
> From: Ajit Singh 
> Sent: Tuesday, July 28, 2020 12:05 PM
> To: users@cloudstack.apache.org
> Subject: System vm template
>
> Hi
>
> Can anyone guide me on how to register or install system vm templates for
> vsphere and is it mandatory to reboot acs management servers after system
> vm templates installation or registration.
>
> Any link would also help
>
> Thanks
> Ajit
>

RE: Windows Template & Multiple SCSI Controllers

[Corey, Mike]

I could use some advice on this.  I completely recreated my Window10 
VM/Template and imported it into ACS.  Deploying an instance runs; however a 
couple things are out of the normal and I could use some guidance on 
troubleshooting.

1 - The ASC template that ASC deploys does not include the network adapter that 
my imported OVA has.  Is this expected behavior of the template?

2 - Booting the deployed Instance VM to UEFI can't see the system volume 
(device 0:0) and fails to boot up Windows.

3 - Booting the deployed Instance VM to BIOS gets a Windows unrecoverable error 
- fails to boot up Windows correctly.

Cloning a VM from the ASC Deployed Template VM (adding a vnic after) works and 
the VM loads as expected.

So my question to the greater forum is what is ASC doing under the covers to 
the VM hardware that could prevent the OS / System Volume / primary partition 
from loading in the ASC deployed VM Instance?  

Anyone with a VMware & CloudStack deployment running out here that can shed 
some light?

Thanks!

Mike



 

-Original Message-
From: Corey, Mike  
Sent: Monday, July 27, 2020 11:48 AM
To: users@cloudstack.apache.org
Subject: [CAUTION] RE: Windows Template & Multiple SCSI Controllers

Why would you hardcode the addition of unrequired/unnecessary "hardware" to the 
VM instance?  This wasn't the case for the CentOS deployment so why do it with 
Windows OS?

I can't say for certain, but the VM instance that ACS creates doesn't start the 
OS (blue screen recovery console).  However, cloning from the template ACS 
creates in vCenter through the tradition vCenter method, the VM loads as normal.

Can this be changed via a global setting or other config file edit?



-Original Message-
From: Andrija Panic  
Sent: Saturday, July 25, 2020 2:44 PM
To: users 
Subject: Re: Windows Template & Multiple SCSI Controllers

Hardcoded behavior of having 4 identical controllers, with all the volumes
attached to the first one.

Why is this a problem for you?

Best,

On Fri, 24 Jul 2020, 19:55 Corey, Mike,  wrote:

> Hi,
>
>
>
> As I progress with my ACS & VMware setup I seem to hit a bump at every
> turn.  I’ve gotten to the point where I’m now able to upload a Windows 10
> template that we use in production (VMware) into ACS.  However, when I
> create a new instance through the GUI it is deployed with a total of 4 SCSI
> controllers when it should only have ONE.
>
>
>
> What is strange is that the clone of the template that ACS copies into
> vCenter only has a single SCSI controller (as expected).  In fact, when I
> clone a VM (traditional vCenter method) from the ACS template that was
> create – that VM only has the single controller and boots to the OS fine.
>
>
>
> Any ideas are welcome as to why this behavior is occurring.
>
>
>
> Thanks!
>
> Mike
>
>
>
>
>
>
>
> *Mike Corey*
>
>
> Technology Senior Consultant, IT CS CTW Operation & Virtualization Service
> US
>
>
> *SAP AMERICA, INC.* 3999 West Chester Pike, Newtown Square, 19073 United
> States
>
>
> T +1 610 661 0905, M +1 484 274 2658, E mike.co...@sap.com
>
>
>
>
>
>
>

Re: Windows Template & Multiple SCSI Controllers

[Marc-Andre Jutras]

Hey Mike,

On 2020-07-28 3:37 PM, Corey, Mike wrote:

I could use some advice on this.  I completely recreated my Window10 
VM/Template and imported it into ACS.  Deploying an instance runs; however a 
couple things are out of the normal and I could use some guidance on 
troubleshooting.

1 - The ASC template that ASC deploys does not include the network adapter that 
my imported OVA has.  Is this expected behavior of the template?



Yes, It's expected, vNIC will be added to your VM only when you will 
start your VM...





2 - Booting the deployed Instance VM to UEFI can't see the system volume 
(device 0:0) and fails to boot up Windows.

make sure you boot up your vm from cloudstack and not from vmware...


3 - Booting the deployed Instance VM to BIOS gets a Windows unrecoverable error 
- fails to boot up Windows correctly.
same... make sure you boot up your vm from cloudstack and not from 
vmware...


Cloning a VM from the ASC Deployed Template VM (adding a vnic after) works and 
the VM loads as expected.


whoa, party on dude ;)

to use vmxnet3 driver or any others specific settings in your template, 
on acs, select your template and under settings, add :


nicAdapter = Vmxnet3
and define any others settings there... ( keyboard, root scsi driver, etc... )



So my question to the greater forum is what is ASC doing under the covers to 
the VM hardware that could prevent the OS / System Volume / primary partition 
from loading in the ASC deployed VM Instance?
ACS drives vmware through API calls, ACS will push configuration to your 
vm only when needed : always manage your VM via Cloudstack and don't do 
anything directly on vmware ;)
  


Anyone with a VMware & CloudStack deployment running out here that can shed 
some light?
yup, me ;)  6 ACS regions with vmware 5.5, 6.0 and 6.7 ( upgrade to 6.7 
on all regions on hold for now // covid... )


Thanks!

Mike



  


-Original Message-
From: Corey, Mike 
Sent: Monday, July 27, 2020 11:48 AM
To: users@cloudstack.apache.org
Subject: [CAUTION] RE: Windows Template & Multiple SCSI Controllers

Why would you hardcode the addition of unrequired/unnecessary "hardware" to the 
VM instance?  This wasn't the case for the CentOS deployment so why do it with Windows OS?

I can't say for certain, but the VM instance that ACS creates doesn't start the 
OS (blue screen recovery console).  However, cloning from the template ACS 
creates in vCenter through the tradition vCenter method, the VM loads as normal.

Can this be changed via a global setting or other config file edit?



-Original Message-
From: Andrija Panic 
Sent: Saturday, July 25, 2020 2:44 PM
To: users 
Subject: Re: Windows Template & Multiple SCSI Controllers

Hardcoded behavior of having 4 identical controllers, with all the volumes
attached to the first one.

Why is this a problem for you?

Best,

On Fri, 24 Jul 2020, 19:55 Corey, Mike,  wrote:


Hi,



As I progress with my ACS & VMware setup I seem to hit a bump at every
turn.  I’ve gotten to the point where I’m now able to upload a Windows 10
template that we use in production (VMware) into ACS.  However, when I
create a new instance through the GUI it is deployed with a total of 4 SCSI
controllers when it should only have ONE.



What is strange is that the clone of the template that ACS copies into
vCenter only has a single SCSI controller (as expected).  In fact, when I
clone a VM (traditional vCenter method) from the ACS template that was
create – that VM only has the single controller and boots to the OS fine.



Any ideas are welcome as to why this behavior is occurring.



Thanks!

Mike







*Mike Corey*


Technology Senior Consultant, IT CS CTW Operation & Virtualization Service
US


*SAP AMERICA, INC.* 3999 West Chester Pike, Newtown Square, 19073 United
States


T +1 610 661 0905, M +1 484 274 2658, E mike.co...@sap.com

Re: Windows Template & Multiple SCSI Controllers

[Thomas Joseph]

To add few more troubleshooting tips:
1. In template/VM settings tab, add entries for  Vnic (vmxnet3) and root &
data disk controllers (osdefault) from the dropdown menu.
2. You would need to have vmtools in the template.

Regards
Thomas

On Wed, 29 Jul 2020, 6:50 am Marc-Andre Jutras, 
wrote:

> Hey Mike,
>
> On 2020-07-28 3:37 PM, Corey, Mike wrote:
> > I could use some advice on this.  I completely recreated my Window10
> VM/Template and imported it into ACS.  Deploying an instance runs; however
> a couple things are out of the normal and I could use some guidance on
> troubleshooting.
> >
> > 1 - The ASC template that ASC deploys does not include the network
> adapter that my imported OVA has.  Is this expected behavior of the
> template?
>
>
> Yes, It's expected, vNIC will be added to your VM only when you will
> start your VM...
>
>
> >
> > 2 - Booting the deployed Instance VM to UEFI can't see the system volume
> (device 0:0) and fails to boot up Windows.
> make sure you boot up your vm from cloudstack and not from vmware...
> >
> > 3 - Booting the deployed Instance VM to BIOS gets a Windows
> unrecoverable error - fails to boot up Windows correctly.
> same... make sure you boot up your vm from cloudstack and not from
> vmware...
> >
> > Cloning a VM from the ASC Deployed Template VM (adding a vnic after)
> works and the VM loads as expected.
>
> whoa, party on dude ;)
>
> to use vmxnet3 driver or any others specific settings in your template,
> on acs, select your template and under settings, add :
>
> nicAdapter = Vmxnet3
> and define any others settings there... ( keyboard, root scsi driver,
> etc... )
>
> >
> > So my question to the greater forum is what is ASC doing under the
> covers to the VM hardware that could prevent the OS / System Volume /
> primary partition from loading in the ASC deployed VM Instance?
> ACS drives vmware through API calls, ACS will push configuration to your
> vm only when needed : always manage your VM via Cloudstack and don't do
> anything directly on vmware ;)
> >
> >
> > Anyone with a VMware & CloudStack deployment running out here that can
> shed some light?
> yup, me ;)  6 ACS regions with vmware 5.5, 6.0 and 6.7 ( upgrade to 6.7
> on all regions on hold for now // covid... )
> >
> > Thanks!
> >
> > Mike
> >
> >
> >
> >
> >
> > -Original Message-
> > From: Corey, Mike 
> > Sent: Monday, July 27, 2020 11:48 AM
> > To: users@cloudstack.apache.org
> > Subject: [CAUTION] RE: Windows Template & Multiple SCSI Controllers
> >
> > Why would you hardcode the addition of unrequired/unnecessary "hardware"
> to the VM instance?  This wasn't the case for the CentOS deployment so why
> do it with Windows OS?
> >
> > I can't say for certain, but the VM instance that ACS creates doesn't
> start the OS (blue screen recovery console).  However, cloning from the
> template ACS creates in vCenter through the tradition vCenter method, the
> VM loads as normal.
> >
> > Can this be changed via a global setting or other config file edit?
> >
> >
> >
> > -Original Message-
> > From: Andrija Panic 
> > Sent: Saturday, July 25, 2020 2:44 PM
> > To: users 
> > Subject: Re: Windows Template & Multiple SCSI Controllers
> >
> > Hardcoded behavior of having 4 identical controllers, with all the
> volumes
> > attached to the first one.
> >
> > Why is this a problem for you?
> >
> > Best,
> >
> > On Fri, 24 Jul 2020, 19:55 Corey, Mike,  wrote:
> >
> >> Hi,
> >>
> >>
> >>
> >> As I progress with my ACS & VMware setup I seem to hit a bump at every
> >> turn.  I’ve gotten to the point where I’m now able to upload a Windows
> 10
> >> template that we use in production (VMware) into ACS.  However, when I
> >> create a new instance through the GUI it is deployed with a total of 4
> SCSI
> >> controllers when it should only have ONE.
> >>
> >>
> >>
> >> What is strange is that the clone of the template that ACS copies into
> >> vCenter only has a single SCSI controller (as expected).  In fact, when
> I
> >> clone a VM (traditional vCenter method) from the ACS template that was
> >> create – that VM only has the single controller and boots to the OS
> fine.
> >>
> >>
> >>
> >> Any ideas are welcome as to why this behavior is occurring.
> >>
> >>
> >>
> >> Thanks!
> >>
> >> Mike
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >> *Mike Corey*
> >>
> >>
> >> Technology Senior Consultant, IT CS CTW Operation & Virtualization
> Service
> >> US
> >>
> >>
> >> *SAP AMERICA, INC.* 3999 West Chester Pike, Newtown Square, 19073 United
> >> States
> >>
> >>
> >> T +1 610 661 0905, M +1 484 274 2658, E mike.co...@sap.com
> >>
> >>
> >>
> >>
> >>
> >>
> >>
>