Building Debian style package fails on 2.0 branch and 1.4.0 tag about walinuxagent

[simpl1g]

#!/usr/bin/make -f

%:
    dh $@ --with python2
override_dh_auto_build override_dh_auto_install:
    @

Change debian/rules files to look like this code and it will build, however I can't launch instances on azure with this agent, provision fails. When there will be update for debian??

[szarkos]

Can you please try without using setup.py? We've only tested that install method with the RPM packaging, but the Debian packaging predates it and should not require it.

Ubuntu is also still on the 1.3.x branch as well (with a few patches backported), and it works just fine on Azure. We're working to have a 2.0.x agent packaged in Ubuntu sometime in the near future, although I do not have a timeframe for that. If there are any packaging changes required for 2.0.x we will be sure to get those patches pushed upstream here for Debian as well.

In the meantime, if you want to test out the newest agent you can also install the agent "manually" via "waagent -install". See the readme for more details. It's not ideal, but for testing it should work. We have Debian 7.x running in our test automation, and we use this install method to "slipstream" in the latest agent before deploying on Azure.

Thanks,
Steve

[TriJetScud]

As I am very unfamiliar with Debian style packaging, how does one remove the setup.py process in the debian/ directory for packaging purposes?

[szarkos]

setup.py is not being called by anything in the debian/ directory but by dpkg-buildpackage. You can probably just remove setup.py (or rename it) or add the parameters simpl1g was suggesting.

@simpl1g, I'm not sure why a 1.4.0 or 2.0.1 package would fail to provision. Can you try to mount the VHD of the failed VM (you can actually do this from another working VM in Azure) and try to grab the logs in /var/log? I'd love to take a look and see what's wrong there.

[simpl1g]

@szarkos This is what i have in waagent.log. Look like i have 2 agents running?

2013/11/08 14:43:27 Configured SSH client probing to keep connections alive.
2013/11/08 15:22:13 Windows Azure Linux Agent Version: WALinuxAgent-2.0.1
2013/11/08 15:22:13 Linux Distribution Detected      : Ubuntu
2013/11/08 15:22:13 Windows Azure Linux Agent Version: WALinuxAgent-2.0.1
2013/11/08 15:22:13 Linux Distribution Detected      : Ubuntu
2013/11/08 15:22:13 uname -r
2013/11/08 15:22:13 uname -r
2013/11/08 15:22:13 lsmod | grep ata_piix
2013/11/08 15:22:13 lsmod | grep ata_piix
2013/11/08 15:22:13 Module /lib/modules/3.2.0-56-virtual/kernel/drivers/ata/ata_piix.ko driver for ATAPI CD-ROM does not exist.
2013/11/08 15:22:13 LC_ALL=C fdisk -l /dev/sr0 | grep Disk
2013/11/08 15:22:13 Module /lib/modules/3.2.0-56-virtual/kernel/drivers/ata/ata_piix.ko driver for ATAPI CD-ROM does not exist.
2013/11/08 15:22:13 LC_ALL=C fdisk -l /dev/sr0 | grep Disk
2013/11/08 15:22:13 mount -v /dev/sr0 /mnt/cdrom/secure
2013/11/08 15:22:13 mount -v /dev/sr0 /mnt/cdrom/secure
2013/11/08 15:22:13 mount: block device /dev/sr0 is write-protected, mounting read-only
2013/11/08 15:22:13 mount: you didn't specify a filesystem type for /dev/sr0
2013/11/08 15:22:13        I will try type udf
2013/11/08 15:22:13 mount: you didn't specify a filesystem type for /dev/sr0
2013/11/08 15:22:13        I will try type udf
2013/11/08 15:22:13 /dev/sr0 on /mnt/cdrom/secure type udf (ro)
2013/11/08 15:22:13 mount succeeded on attempt #1
2013/11/08 15:22:13 umount /dev/sr0
2013/11/08 15:22:13 ERROR:CalledProcessError.  Error Code is 32
2013/11/08 15:22:13 ERROR:CalledProcessError.  Command string was mount -v /dev/sr0 /mnt/cdrom/secure
2013/11/08 15:22:13 ERROR:CalledProcessError.  Command result was mount: block device /dev/sr0 is write-protected, mounting read-only
2013/11/08 15:22:13 ERROR:mount: /dev/sr0 already mounted or /mnt/cdrom/secure busy
2013/11/08 15:22:13 ERROR:mount: according to mtab, /dev/sr0 is already mounted on /mnt/cdrom/secure
2013/11/08 15:22:13 ERROR:mount: you didn't specify a filesystem type for /dev/sr0
2013/11/08 15:22:13 ERROR:       I will try type udf
2013/11/08 15:22:13 ERROR:mount: you didn't specify a filesystem type for /dev/sr0
2013/11/08 15:22:13 ERROR:       I will try type udf
2013/11/08 15:22:13 mount: block device /dev/sr0 is write-protected, mounting read-only
2013/11/08 15:22:13 mount: /dev/sr0 already mounted or /mnt/cdrom/secure busy
2013/11/08 15:22:13 mount: according to mtab, /dev/sr0 is already mounted on /mnt/cdrom/secure
2013/11/08 15:22:13 mount: you didn't specify a filesystem type for /dev/sr0
2013/11/08 15:22:13        I will try type udf
2013/11/08 15:22:13 mount: you didn't specify a filesystem type for /dev/sr0
2013/11/08 15:22:13        I will try type udf
2013/11/08 15:22:13 Device /dev/sr0 is already mounted on /mnt/cdrom/secure.1
2013/11/08 15:22:13 umount /dev/sr0
2013/11/08 15:22:13 VMM Init script not found.  Provisioning for Azure
2013/11/08 15:22:13 IPv4 address: 0.0.0.0
2013/11/08 15:22:13 VMM Init script not found.  Provisioning for Azure
2013/11/08 15:22:13 IPv4 address: 0.0.0.0
2013/11/08 15:22:13 ERROR:Traceback (most recent call last):
2013/11/08 15:22:13 ERROR:  File "/usr/sbin/waagent", line 4147, in main
2013/11/08 15:22:13 ERROR:    WaAgent.Run()
2013/11/08 15:22:13 ERROR:  File "/usr/sbin/waagent", line 3700, in Run
2013/11/08 15:22:13 ERROR:    Log("MAC  address: " + ":".join(["%02X" % Ord(a) for a in MyDistro.GetMacAddress()]))
2013/11/08 15:22:13 ERROR:  File "/usr/sbin/waagent", line 407, in GetMacAddress
2013/11/08 15:22:13 ERROR:    return GetMacAddress()
2013/11/08 15:22:13 ERROR:  File "/usr/sbin/waagent", line 2045, in GetMacAddress
2013/11/08 15:22:13 ERROR:    a = Linux_ioctl_GetInterfaceMac(ifname)
2013/11/08 15:22:13 ERROR:  File "/usr/sbin/waagent", line 1996, in Linux_ioctl_GetInterfaceMac
2013/11/08 15:22:13 ERROR:    info = fcntl.ioctl(s.fileno(), 0x8927,  struct.pack('256s', (ifname[:15]+('\0'*241)).encode('latin-1')))
2013/11/08 15:22:13 ERROR:IOError: [Errno 19] No such device
2013/11/08 15:22:13 ERROR:
2013/11/08 15:22:13 ERROR:Exception: [Errno 19] No such device
2013/11/08 15:22:13 ERROR:Traceback (most recent call last):
2013/11/08 15:22:13 ERROR:  File "/usr/sbin/waagent", line 4147, in main
2013/11/08 15:22:13 ERROR:    WaAgent.Run()
2013/11/08 15:22:13 ERROR:  File "/usr/sbin/waagent", line 3700, in Run
2013/11/08 15:22:13 ERROR:    Log("MAC  address: " + ":".join(["%02X" % Ord(a) for a in MyDistro.GetMacAddress()]))
2013/11/08 15:22:13 ERROR:  File "/usr/sbin/waagent", line 407, in GetMacAddress
2013/11/08 15:22:13 ERROR:    return GetMacAddress()
2013/11/08 15:22:13 ERROR:  File "/usr/sbin/waagent", line 2045, in GetMacAddress
2013/11/08 15:22:13 ERROR:    a = Linux_ioctl_GetInterfaceMac(ifname)
2013/11/08 15:22:13 ERROR:  File "/usr/sbin/waagent", line 1996, in Linux_ioctl_GetInterfaceMac
2013/11/08 15:22:13 ERROR:    info = fcntl.ioctl(s.fileno(), 0x8927,  struct.pack('256s', (ifname[:15]+('\0'*241)).encode('latin-1')))
2013/11/08 15:22:13 ERROR:IOError: [Errno 19] No such device
2013/11/08 15:22:13 ERROR:
2013/11/08 15:22:13 ERROR:Exception: [Errno 19] No such device

[szarkos]

Yes, it looks like it's running twice for some reason. Can you check your init scripts and see if it's listed twice somehow?

Coincidentally, I see there is a new pull request that's supposed to fix up the package building for Debian: #19

Any chance you could pull that branch and see if it works?

Steve

[simpl1g]

I tried to run this update in ubuntu

2013/11/11 13:03:21 Windows Azure Linux Agent Version: WALinuxAgent-2.0.2-PRE
2013/11/11 13:03:21 Linux Distribution Detected      : Ubuntu
2013/11/11 13:03:21 Module /lib/modules/3.2.0-56-virtual/kernel/drivers/ata/ata_piix.ko driver for ATAPI CD-ROM does not exist.
2013/11/11 13:03:21 mount: block device /dev/sr0 is write-protected, mounting read-only
2013/11/11 13:03:21 mount: you didn't specify a filesystem type for /dev/sr0
2013/11/11 13:03:21        I will try type udf
2013/11/11 13:03:21 mount: you didn't specify a filesystem type for /dev/sr0
2013/11/11 13:03:21        I will try type udf
2013/11/11 13:03:21 /dev/sr0 on /mnt/cdrom/secure type udf (ro)
2013/11/11 13:03:21 mount succeeded on attempt #1
2013/11/11 13:03:21 VMM Init script not found.  Provisioning for Azure
2013/11/11 13:03:21 IPv4 address: 0.0.0.0
2013/11/11 13:03:22 ERROR:Traceback (most recent call last):
2013/11/11 13:03:22 ERROR:  File "/usr/sbin/waagent", line 4147, in main
2013/11/11 13:03:22 ERROR:    WaAgent.Run()
2013/11/11 13:03:22 ERROR:  File "/usr/sbin/waagent", line 3700, in Run
2013/11/11 13:03:22 ERROR:    Log("MAC  address: " + ":".join(["%02X" % Ord(a) for a in MyDistro.GetMacAddress()]))
2013/11/11 13:03:22 ERROR:  File "/usr/sbin/waagent", line 407, in GetMacAddress
2013/11/11 13:03:22 ERROR:    return GetMacAddress()
2013/11/11 13:03:22 ERROR:  File "/usr/sbin/waagent", line 2045, in GetMacAddress
2013/11/11 13:03:22 ERROR:    a = Linux_ioctl_GetInterfaceMac(ifname)
2013/11/11 13:03:22 ERROR:  File "/usr/sbin/waagent", line 1996, in Linux_ioctl_GetInterfaceMac
2013/11/11 13:03:22 ERROR:    info = fcntl.ioctl(s.fileno(), 0x8927,  struct.pack('256s', (ifname[:15]+('\0'*241)).encode('latin-1')))
2013/11/11 13:03:22 ERROR:IOError: [Errno 19] No such device
2013/11/11 13:03:22 ERROR:
2013/11/11 13:03:22 ERROR:Exception: [Errno 19] No such device
2013/11/11 13:03:25 Windows Azure Linux Agent Version: WALinuxAgent-2.0.2-PRE
2013/11/11 13:03:25 Linux Distribution Detected      : Ubuntu
2013/11/11 13:03:25 Module /lib/modules/3.2.0-56-virtual/kernel/drivers/ata/ata_piix.ko driver for ATAPI CD-ROM does not exist.
2013/11/11 13:03:25 mount: block device /dev/sr0 is write-protected, mounting read-only
2013/11/11 13:03:25 mount: you didn't specify a filesystem type for /dev/sr0
2013/11/11 13:03:25        I will try type udf
2013/11/11 13:03:25 mount: you didn't specify a filesystem type for /dev/sr0
2013/11/11 13:03:25        I will try type udf
2013/11/11 13:03:25 /dev/sr0 on /mnt/cdrom/secure type udf (ro)
2013/11/11 13:03:25 mount succeeded on attempt #1
2013/11/11 13:03:25 VMM Init script not found.  Provisioning for Azure
2013/11/11 13:03:25 IPv4 address: 100.86.88.122
2013/11/11 13:03:25 MAC  address: 00:15:5D:52:8F:30
2013/11/11 13:03:25 Probing for Windows Azure environment.
2013/11/11 13:03:25 DoDhcpWork: Setting socket.timeout=10, entering recv
2013/11/11 13:03:25 Discovered Windows Azure endpoint: 100.86.88.70
2013/11/11 13:03:25 Fabric preferred wire protocol version: 2012-11-30
2013/11/11 13:03:25 Negotiated wire protocol version: 2012-11-30
2013/11/11 13:03:25 EnvMonitor: Moved /lib/udev/rules.d/75-persistent-net-generator.rules -> /var/lib/waagent
2013/11/11 13:03:26 Retrieved GoalState from Windows Azure Fabric.
2013/11/11 13:03:26 ExpectedState: Started
2013/11/11 13:03:26 ContainerId: 667ad872-31f2-4efb-99f4-9072375eb35d
2013/11/11 13:03:26 RoleInstanceId: 65973f3422ce432b93a5767825d93727.test
2013/11/11 13:03:26 Public cert with thumbprint: 979AC96854E389270D3C1C896F990F3B3EB49FCD was retrieved.
2013/11/11 13:03:26 Provisioning image started.
2013/11/11 13:03:27 mount: block device /dev/sr0 is write-protected, mounting read-only
2013/11/11 13:03:27 mount succeeded on attempt #1
2013/11/11 13:03:27 Provisioning image using OVF settings in the DVD.
2013/11/11 13:03:27 Disabled SSH password-based authentication methods.
2013/11/11 13:03:28 Resource disk (/dev/sdb1) is mounted at /mnt/resource with fstype ext4
2013/11/11 13:03:28 Created user account: azureuser
2013/11/11 13:03:30 EnvMonitor: Detected host name change: ubuntu -> test
2013/11/11 13:03:31 Posted Role Properties. CertificateThumbprint=e9c7d03a7206ac813e8c74a2d565d662
2013/11/11 13:03:31 Provisioning image completed.

Looks like it was able to provision.

ps ax | grep waagent
962 ?        Sl     0:00 python /usr/sbin/waagent -daemon

waagent is running, but service not

service walinuxagent status
walinuxagent stop/waiting

sudo service walinuxagent start
Trying to start it brings

2013/11/11 13:41:13 Windows Azure Linux Agent Version: WALinuxAgent-2.0.2-PRE
2013/11/11 13:41:13 Linux Distribution Detected      : Ubuntu
2013/11/11 13:41:13 Module /lib/modules/3.2.0-56-virtual/kernel/drivers/ata/ata_piix.ko driver for ATAPI CD-ROM does not exist.
2013/11/11 13:41:13 VMM Init script not found.  Provisioning for Azure
2013/11/11 13:41:13 IPv4 address: 100.86.88.122
2013/11/11 13:41:13 MAC  address: 00:15:5D:52:8F:30
2013/11/11 13:41:13 Probing for Windows Azure environment.
2013/11/11 13:41:13 DoDhcpWork: Setting socket.timeout=10, entering recv
2013/11/11 13:41:13 Discovered Windows Azure endpoint: 100.86.88.70
2013/11/11 13:41:13 Fabric preferred wire protocol version: 2012-11-30
2013/11/11 13:41:13 Negotiated wire protocol version: 2012-11-30
2013/11/11 13:41:14 Retrieved GoalState from Windows Azure Fabric.
2013/11/11 13:41:14 ExpectedState: Started
2013/11/11 13:41:14 ContainerId: 667ad872-31f2-4efb-99f4-9072375eb35d
2013/11/11 13:41:14 RoleInstanceId: 65973f3422ce432b93a5767825d93727.test
2013/11/11 13:41:14 Public cert with thumbprint: 979AC96854E389270D3C1C896F990F3B3EB49FCD was retrieved.
2013/11/11 13:41:14 ActivateResourceDisk: /dev/sdb1 is already mounted.
2013/11/11 13:41:14 ERROR:Traceback (most recent call last):
2013/11/11 13:41:14 ERROR:  File "/usr/sbin/waagent", line 4147, in main
2013/11/11 13:41:14 ERROR:    WaAgent.Run()
2013/11/11 13:41:14 ERROR:  File "/usr/sbin/waagent", line 3774, in Run
2013/11/11 13:41:14 ERROR:    self.LoadBalancerProbeServer = LoadBalancerProbeServer(currentPort)
2013/11/11 13:41:14 ERROR:  File "/usr/sbin/waagent", line 2218, in __init__
2013/11/11 13:41:14 ERROR:    self.server = SocketServer.TCPServer((self.get_ip(), port), TCPHandler)
2013/11/11 13:41:14 ERROR:  File "/usr/lib/python2.7/SocketServer.py", line 408, in __init__
2013/11/11 13:41:14 ERROR:    self.server_bind()
2013/11/11 13:41:14 ERROR:  File "/usr/lib/python2.7/SocketServer.py", line 419, in server_bind
2013/11/11 13:41:14 ERROR:    self.socket.bind(self.server_address)
2013/11/11 13:41:14 ERROR:  File "/usr/lib/python2.7/socket.py", line 224, in meth
2013/11/11 13:41:14 ERROR:    return getattr(self._sock,name)(*args)
2013/11/11 13:41:14 ERROR:error: [Errno 98] Address already in use
2013/11/11 13:41:14 ERROR:
2013/11/11 13:41:14 ERROR:Exception: [Errno 98] Address already in use

What is the difference between walinuxagent and waagent, in standard azure ubuntu image they are both running

[szarkos]

You need to remove the walinuxagent package before installing a new agent manually or via a package. You may also want to disable the cloud-init service unless you are installing via Canonical's official package.

When we install the agent manually, the service is called "waagent," but when installed via Ubuntu's packaging the service is walinuxagent. This was done on purpose so that the agent can determine if it was packaged or installed manually. Having both running may work usually (although you'll see all that ugliness in the logs), but it's not guaranteed to provisioning properly all the time.

Another consideration is that Azure's Ubuntu cloud images for 12.04, 13.04 and 13.10 (but not 12.10) all have cloud-init enabled by default and use it to do much of the provisioning work. They also carry some patches upstream for compatibility with cloud-init: http://bazaar.launchpad.net/~ubuntu-branches/ubuntu/saucy/walinuxagent/saucy/view/head:/debian/patches/config_for_cloud-init.patch

So, on Ubuntu, if installing manually via "waagent -install" or via your own 1.4 package you will likely want to also disable the cloud-init service. We test this scenario all the time and it works fine. However, if you want to use cloud-init you'll have to sync with Canonical's packaging upstream patches, which may be a bit of work. When we get the 2.0 version packaged for Ubuntu it should be a lot easier for them to upstream those patches for cloud-init as well.

Thanks,
Steve

[simpl1g]

I'm installing it on self-made image of ubuntu 12.04, so there is no walinuxagent installed before.
I'm building package from #19 and installing it via update-rc.d waagent defaults. And as i can see cloud-init service disabled

[g-k-r]

@simpl1g ubuntu is using upstart, so you better not add the systemV init as well. For backward compatibility upstart also starts systemV init files if available so this would start the waagent twice. use update-rc.d waagent remove to clear sysV init. to investigate the upstart service use initctl status walinuxagent and as usual check the /var/log/syslog

the upstart file is installed from source debian/upstart to /etc/init/walinuxagent.conf. I do not know the upstart system sufficiently to verify that is enough to wait for the mount event of the root file system. maybe it would be better to also wait for the network? since you had to removed the network-manager, how does your /etc/network/interfaces look like ?

@szarkos it looks like GetFirstActiveNetworkInterfaceNonLoopback returns 0.0.0.0 in GetIpv4Address for a not existing interface (i.e interfaces not yet known to the kernel) instead of a None which would make the agent wait. is that possible ?

[simpl1g]

/etc/network/interfaces looks like this one, i'm generating it manually

auto lo
iface lo inet loopback
auto eth0
iface eth0 inet dhcp

start on filesystem and net-device-up IFACE=eth0 in upstart fixed problem with 0.0.0.0 in GetIpv4Address

Now I have the following log. Despite the errors in log everything seems to be working. As I noticed, it's better to set ClientAliveInterval to 180 in sshd config to prevent freezing terminal, because this manual installation step is no longer running

2013/11/13 10:16:38 Windows Azure Linux Agent Version: WALinuxAgent-2.0.1
2013/11/13 10:16:38 Linux Distribution Detected      : Ubuntu
2013/11/13 10:16:38 Module /lib/modules/3.2.0-56-virtual/kernel/drivers/ata/ata_piix.ko driver for ATAPI CD-ROM does not exist.
2013/11/13 10:16:38 mount: block device /dev/sr0 is write-protected, mounting read-only
2013/11/13 10:16:38 mount: you didn't specify a filesystem type for /dev/sr0
2013/11/13 10:16:38        I will try type udf
2013/11/13 10:16:38 mount: you didn't specify a filesystem type for /dev/sr0
2013/11/13 10:16:38        I will try type udf
2013/11/13 10:16:38 /dev/sr0 on /mnt/cdrom/secure type udf (ro)
2013/11/13 10:16:38 mount succeeded on attempt #1
2013/11/13 10:16:38 VMM Init script not found.  Provisioning for Azure
2013/11/13 10:16:38 IPv4 address: 10.0.0.6
2013/11/13 10:16:38 MAC  address: 00:15:5D:67:39:47
2013/11/13 10:16:38 Probing for Windows Azure environment.
2013/11/13 10:16:38 DoDhcpWork: Setting socket.timeout=10, entering recv
2013/11/13 10:16:38 Discovered Windows Azure endpoint: 168.63.129.16
2013/11/13 10:16:38 Fabric preferred wire protocol version: 2012-11-30
2013/11/13 10:16:38 Negotiated wire protocol version: 2012-11-30
2013/11/13 10:16:38 EnvMonitor: Moved /lib/udev/rules.d/75-persistent-net-generator.rules -> /var/lib/waagent
2013/11/13 10:16:39 Retrieved GoalState from Windows Azure Fabric.
2013/11/13 10:16:39 ExpectedState: Started
2013/11/13 10:16:39 ContainerId: e9f8542c-9427-4a1e-9d3e-4bc4d3ceec84
2013/11/13 10:16:39 RoleInstanceId: 659959e840ec4cdca4497671bc1bd1b0.test
2013/11/13 10:16:39 Public cert with thumbprint: 979AC96854E389270D3C1C896F990F3B3EB49FCD was retrieved.
2013/11/13 10:16:39 Provisioning image started.
2013/11/13 10:16:40 ERROR:CalledProcessError.  Error Code is 1
2013/11/13 10:16:40 ERROR:CalledProcessError.  Command string was /usr/sbin/service ssh status | grep running
2013/11/13 10:16:40 ERROR:CalledProcessError.  Command result was 
2013/11/13 10:16:40 Resource disk (/dev/sdb1) is mounted at /mnt/resource with fstype ext4
2013/11/13 10:16:41 mount: block device /dev/sr0 is write-protected, mounting read-only
2013/11/13 10:16:41 mount succeeded on attempt #1
2013/11/13 10:16:41 Provisioning image using OVF settings in the DVD.
2013/11/13 10:16:41 Disabled SSH password-based authentication methods.
2013/11/13 10:16:41 CreateAccount: azureuser already exists. Will update password.
2013/11/13 10:16:41 Created user account: azureuser
2013/11/13 10:16:41 ERROR:CalledProcessError.  Error Code is 1
2013/11/13 10:16:41 ERROR:CalledProcessError.  Command string was /usr/sbin/service ssh status | grep running
2013/11/13 10:16:41 ERROR:CalledProcessError.  Command result was 
2013/11/13 10:16:41 ERROR:CalledProcessError.  Error Code is 1
2013/11/13 10:16:41 ERROR:CalledProcessError.  Command string was ssh-keygen -lf /etc/ssh/ssh_host_rsa_key.pub
2013/11/13 10:16:41 ERROR:CalledProcessError.  Command result was /etc/ssh/ssh_host_rsa_key.pub: No such file or directory
2013/11/13 10:16:41 Posted Role Properties. CertificateThumbprint=No
2013/11/13 10:16:41 Provisioning image completed.
2013/11/13 10:16:48 EnvMonitor: Detected dhcp client restart. Restoring routing table.

[szarkos]

That is good news.

Regarding modifying the ClientAliveInterval, the Ubuntu community pushed back on the agent editing the sshd_config file because it is owned by another package, which is not allowed by some Debian packaging guidelines: https://bugs.launchpad.net/ubuntu/+source/walinuxagent/+bug/1188610

@g-k-r, I will have someone take a look at the GetFirstActiveNetworkInterfaceNonLoopback issue.

Thanks,
Steve

[g-k-r]

@simpl1g looks good, but which ssh-server are you using ?

[simpl1g]

@g-k-r, I don't know what to answer :) Just standart sshd openssh-server 1:5.9p1-5ubuntu1.1

[g-k-r]

ubuntu has 3 different possible ssh server the default is openssh-server
run dpkg-query -l openssh-server dropbear lsh-server and check the output

[simpl1g]

the answer is in previous post:)

[g-k-r]

:-) interesting. so it seems that the sshd is not yet running when the waagent is, and grep results in '1', which may also be the reason that the host key is not yet generated. try to depend the upstart job on sshd (if possible)

[simpl1g]

@g-k-r Depending upstart on sshd job didn't give results

This error occurs only if provisioning image with password, without ssh key

2013/11/13 10:16:41 ERROR:CalledProcessError.  Error Code is 1
2013/11/13 10:16:41 ERROR:CalledProcessError.  Command string was ssh-keygen -lf /etc/ssh/ssh_host_rsa_key.pub
2013/11/13 10:16:41 ERROR:CalledProcessError.  Command result was /etc/ssh/ssh_host_rsa_key.pub: No such file or directory

2013/11/13 10:16:41 ERROR:CalledProcessError.  Error Code is 1
2013/11/13 10:16:41 ERROR:CalledProcessError.  Command string was /usr/sbin/service ssh status | grep running
2013/11/13 10:16:41 ERROR:CalledProcessError.  Command result was

This error does not appears after manual copying of waagent script and running waagent --setup --force, but after provisioning Azure displays hostname as localhost, although vm shows correct hostname, it looks strange

And the final point, ssh connection freezes with package installation, I can't do anything with terminal for 5-10 minutes, even to disconnect, with manual installation - no problems, can it be connected with ClientAliveInterval?

[simpl1g]

@g-k-r, @szarkos, this is what I have to do to generate fully working package for ubuntu, without any errors
simpl1g@78a074f

[g-k-r]

ok so you disable the upstart job, but why do you also disable setup.py during package generation? the /etc/init.d/waagent file is not packaged !
you also disabled provisioning in the waagent.conf wouldn't that break as soon as you deprovision once ?

[henrich]

still opened?

FYI. I'll upload waagent 2.0.8 to debian unstable, this issue is dealt with specifying override_dh_auto_install in debian/rules to specify --init-system option.

[szarkos]

Hi Hideki,

Thanks for this, can you point me to the 2.0.8 package in Debian?

Also, would it be possible to submit a pull request for these package changes so we can better support Debian?

Thanks!
Steve

[henrich]

On Wed, 29 Oct 2014 13:00:07 -0700
"Stephen A. Zarkos" [email protected] wrote:

Thanks for this, can you point me to the 2.0.8 package in Debian?

Well, it's my local disk yet ;)
probably we should fix some issues with cloud-init that Ubuntu's
walinuxagent package deals with some patches.

Also, would it be possible to submit a pull request for these package changes so we can better support Debian?

Okay.

[henrich]

now put debian package to debian experimental repository
(not unstable since it needs to adjust with cloud-init to work with it...)

[szarkos]

The Ubuntu package currently pulls in cloud-init, but to be clear we don't necessarily need cloud-init for the agent to work well with Debian. Cloud-init is nice to have, but I'm not sure if it should be a blocker to keep the latest agent out of the unstable branch.

But if we want cloud-init, what more changes are needed? If you like feel free to email me at stephen.zarkos at microsoft.com and we can discuss in more detail.

Thanks!