yahoo-eng-team team mailing list archive

[Chad Smith <1998998@xxxxxxxxxxxxxxxxxx>]

Public bug reported:

cloud-init 22.4.2

On Azure, datasource details change across reboot cloud-init incorrectly
removes /run/cloud-init/cloud-id-azure symlink.


2022-12-07 04:14:17,920 - util.py[DEBUG]: Creating symbolic link from '/run/cloud-init/cloud-id' => '/run/cloud-init/cloud-id-azure'
2022-12-07 04:14:17,920 - util.py[DEBUG]: Attempting to remove /run/cloud-init/cloud-id-azure


# FRESH LAUNCH
ubuntu@test-focal:~$ ls -l /run/cloud-init/cloud-id
lrwxrwxrwx 1 root root 30 Dec  7 04:10 /run/cloud-init/cloud-id -> /run/cloud-init/cloud-id-azure
ubuntu@test-focal:~$ grep cloud-id /var/log/cloud-init.log 
2022-12-07 04:10:34,734 - util.py[DEBUG]: Writing to /run/cloud-init/cloud-id-azure - wb: [644] 6 bytes
2022-12-07 04:10:34,735 - util.py[DEBUG]: Creating symbolic link from '/run/cloud-init/cloud-id' => '/run/cloud-init/cloud-id-azure'

# ACROSS REBOOT


ubuntu@test-focal:~$ ls -l /run/cloud-init/cloud-id
lrwxrwxrwx 1 root root 30 Dec  7 04:14 /run/cloud-init/cloud-id -> /run/cloud-init/cloud-id-azure


2022-12-07 04:14:17,920 - util.py[DEBUG]: Writing to /run/cloud-init/cloud-id-azure - wb: [644] 6 bytes
2022-12-07 04:14:17,920 - util.py[DEBUG]: Creating symbolic link from '/run/cloud-init/cloud-id' => '/run/cloud-init/cloud-id-azure'
2022-12-07 04:14:17,920 - util.py[DEBUG]: Attempting to remove /run/cloud-init/cloud-id-azure


It looks like the symlink cleanup logic introduced in
https://github.com/canonical/cloud-
init/commit/217ef6ba6c52788f4363b998b6da08863fea5cd9  (which was
released in cloud-init v. 21.4) doesn't do a proper comparison of
previous-datasource to current datasource


Debug logs added to https://github.com/canonical/cloud-init/blob/main/cloudinit/sources/__init__.py#L443 shows that we are comparing the incorrect source file /root/cloud-init/cloud-id versus actually comparing the cloud-specific symlink name /run/cloud-init/cloud-id-<cloud>, resulting in a removal of the cloud-specific link name across reboot.

 
2022-12-07 04:25:58,162 - __init__.py[WARNING]: previous_cloud_id_file /run/cloud-init/cloud-id-azure != cloud_id_file /run/cloud-init/cloud-id


Likely all we need is a correction in logic like the following:
diff --git a/cloudinit/sources/__init__.py b/cloudinit/sources/__init__.py
index 85e094ac5..3fb08ead2 100644
--- a/cloudinit/sources/__init__.py
+++ b/cloudinit/sources/__init__.py
@@ -439,8 +439,9 @@ class DataSource(CloudInitPickleMixin, metaclass=abc.ABCMeta):
             prev_cloud_id_file = os.path.realpath(cloud_id_file)
         else:
             prev_cloud_id_file = cloud_id_file
-        util.sym_link(f"{cloud_id_file}-{cloud_id}", cloud_id_file, force=True)
-        if prev_cloud_id_file != cloud_id_file:
+        new_cloud_id_file = f"{cloud_id_file}-{cloud_id}"
+        util.sym_link(new_cloud_id_file, cloud_id_file, force=True)
+        if prev_cloud_id_file != new_cloud_id_file:
             util.del_file(prev_cloud_id_file)
         write_json(json_sensitive_file, processed_data, mode=0o600)
         json_file = self.paths.get_runpath("instance_data")

** Affects: cloud-init
     Importance: High
         Status: Triaged


** Tags: bitesize

** Changed in: cloud-init
       Status: New => Triaged

** Changed in: cloud-init
   Importance: Undecided => High

** Tags added: bitesize

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to cloud-init.
https://bugs.launchpad.net/bugs/1998998

Title:
  azure: /run/cloud-init/cloud-id broken symlink across reboot.

Status in cloud-init:
  Triaged

Bug description:
  cloud-init 22.4.2

  On Azure, datasource details change across reboot cloud-init
  incorrectly removes /run/cloud-init/cloud-id-azure symlink.

  
  2022-12-07 04:14:17,920 - util.py[DEBUG]: Creating symbolic link from '/run/cloud-init/cloud-id' => '/run/cloud-init/cloud-id-azure'
  2022-12-07 04:14:17,920 - util.py[DEBUG]: Attempting to remove /run/cloud-init/cloud-id-azure

  
  # FRESH LAUNCH
  ubuntu@test-focal:~$ ls -l /run/cloud-init/cloud-id
  lrwxrwxrwx 1 root root 30 Dec  7 04:10 /run/cloud-init/cloud-id -> /run/cloud-init/cloud-id-azure
  ubuntu@test-focal:~$ grep cloud-id /var/log/cloud-init.log 
  2022-12-07 04:10:34,734 - util.py[DEBUG]: Writing to /run/cloud-init/cloud-id-azure - wb: [644] 6 bytes
  2022-12-07 04:10:34,735 - util.py[DEBUG]: Creating symbolic link from '/run/cloud-init/cloud-id' => '/run/cloud-init/cloud-id-azure'

  # ACROSS REBOOT

  
  ubuntu@test-focal:~$ ls -l /run/cloud-init/cloud-id
  lrwxrwxrwx 1 root root 30 Dec  7 04:14 /run/cloud-init/cloud-id -> /run/cloud-init/cloud-id-azure

  
  2022-12-07 04:14:17,920 - util.py[DEBUG]: Writing to /run/cloud-init/cloud-id-azure - wb: [644] 6 bytes
  2022-12-07 04:14:17,920 - util.py[DEBUG]: Creating symbolic link from '/run/cloud-init/cloud-id' => '/run/cloud-init/cloud-id-azure'
  2022-12-07 04:14:17,920 - util.py[DEBUG]: Attempting to remove /run/cloud-init/cloud-id-azure


  It looks like the symlink cleanup logic introduced in
  https://github.com/canonical/cloud-
  init/commit/217ef6ba6c52788f4363b998b6da08863fea5cd9  (which was
  released in cloud-init v. 21.4) doesn't do a proper comparison of
  previous-datasource to current datasource

  
  Debug logs added to https://github.com/canonical/cloud-init/blob/main/cloudinit/sources/__init__.py#L443 shows that we are comparing the incorrect source file /root/cloud-init/cloud-id versus actually comparing the cloud-specific symlink name /run/cloud-init/cloud-id-<cloud>, resulting in a removal of the cloud-specific link name across reboot.

   
  2022-12-07 04:25:58,162 - __init__.py[WARNING]: previous_cloud_id_file /run/cloud-init/cloud-id-azure != cloud_id_file /run/cloud-init/cloud-id

  
  Likely all we need is a correction in logic like the following:
  diff --git a/cloudinit/sources/__init__.py b/cloudinit/sources/__init__.py
  index 85e094ac5..3fb08ead2 100644
  --- a/cloudinit/sources/__init__.py
  +++ b/cloudinit/sources/__init__.py
  @@ -439,8 +439,9 @@ class DataSource(CloudInitPickleMixin, metaclass=abc.ABCMeta):
               prev_cloud_id_file = os.path.realpath(cloud_id_file)
           else:
               prev_cloud_id_file = cloud_id_file
  -        util.sym_link(f"{cloud_id_file}-{cloud_id}", cloud_id_file, force=True)
  -        if prev_cloud_id_file != cloud_id_file:
  +        new_cloud_id_file = f"{cloud_id_file}-{cloud_id}"
  +        util.sym_link(new_cloud_id_file, cloud_id_file, force=True)
  +        if prev_cloud_id_file != new_cloud_id_file:
               util.del_file(prev_cloud_id_file)
           write_json(json_sensitive_file, processed_data, mode=0o600)
           json_file = self.paths.get_runpath("instance_data")

To manage notifications about this bug go to:
https://bugs.launchpad.net/cloud-init/+bug/1998998/+subscriptions