Support version in platform image

- Override UbuntuServer 16.04-LTS latest to prior version due to
  linux-azure kernel issues

Author: alfpark

CHANGELOG.md

@@ -2,9 +2,16 @@
 
 ## [Unreleased]
 
+### Added
+- Optional `version` support for `platform_image`. This property can be
+used to set a host OS version to prevent possible issues that occur with
+`latest` image versions.
+
 ### Changed
 - Prevent invalid configuration of HPC offers with non-RDMA VM sizes
 - Expanded network tuning exemptions for new Dv3 and Ev3 sizes
+- Temporarily override Canonical UbuntuServer 16.04-LTS latest version to
+a prior version due to recent linux-azure kernel issues.
 
 ### Fixed
 - NV driver updates

config_templates/pool.json

@@ -5,7 +5,8 @@
             "platform_image": {
                 "publisher": "Canonical",
                 "offer": "UbuntuServer",
-                "sku": "16.04-LTS"
+                "sku": "16.04-LTS",
+                "version": "latest"
             },
             "custom_image": {
                 "image_uris": [

convoy/fleet.py

@@ -934,6 +934,19 @@ def _construct_pool_object(
                      pool_settings.vm_configuration.offer,
                      pool_settings.vm_configuration.publisher,
                      pool_settings.vm_configuration.sku))
+        # set image version to use
+        image_ref_to_use.version = pool_settings.vm_configuration.version
+        # TODO temporarily pin Ubuntu to version 16.04.201708151 due to
+        # Docker <-> kernel issues in the linux-azure kernel
+        if (image_ref_to_use.publisher.lower() == 'canonical' and
+                image_ref_to_use.offer.lower() == 'ubuntuserver' and
+                image_ref_to_use.sku.lower() == '16.04-lts' and
+                image_ref_to_use.version.lower() == 'latest'):
+            image_ref_to_use.version = '16.04.201708151'
+            logger.warning(
+                ('overriding Canonical UbuntuServer 16.04-LTS latest version '
+                 'to {}').format(image_ref_to_use.version))
+        logger.info('deploying vm config: {}'.format(image_ref_to_use))
         vmconfig = batchmodels.VirtualMachineConfiguration(
             image_reference=image_ref_to_use,
             node_agent_sku_id=sku_to_use.id,

convoy/settings.py

@@ -88,6 +88,7 @@
         'publisher',
         'offer',
         'sku',
+        'version',
     ]
 )
 PoolVmCustomImageSettings = collections.namedtuple(
@@ -616,6 +617,7 @@ def _populate_pool_vm_configuration(config):
             publisher=conf['publisher'],
             offer=conf['offer'],
             sku=conf['sku'],
+            version=_kv_read_checked(conf, 'version', default='latest'),
         )
     else:
         conf = pool_vm_configuration(config, 'custom_image')

docs/13-batch-shipyard-configuration-pool.md

@@ -13,7 +13,8 @@ The pool schema is as follows:
             "platform_image": {
                 "publisher": "Canonical",
                 "offer": "UbuntuServer",
-                "sku": "16.04-LTS"
+                "sku": "16.04-LTS",
+                "version": "latest"
             },
             "custom_image": {
                 "image_uris": [
@@ -123,6 +124,8 @@ specify both. If using a custom image, please see the
       Marketplace VM image.
     * (required for platform image) `sku` is the sku name of the Marketplace
       VM image.
+    * (optional) `version` is the image version to use. The default is
+      `latest`.
   * (required for custom image) `custom_image` defines the custom image to
     use:
     * (required for custom image) `image_uris` defines a list of page blob

scripts/shipyard_nodeprep.sh

@@ -369,9 +369,9 @@ docker_pull_image() {
         # non-zero exit code: check if pull output has toomanyrequests or
         # connection resets
         if [ contains "$pull_out" "toomanyrequests" ] || [ contains "$pull_out" "connection reset by peer" ]; then
-            echo "WARNING: will retry:\n$pull_out"
+            echo "WARNING: will retry: $pull_out"
         else
-            echo "ERROR:\n$pull_out"
+            echo "ERROR: $pull_out"
             exit $rc
         fi
         let retries=retries-1