-1

The resource group created with Azure python SDK is reported as created, but actually gets created only hours later.

I'm creating the resource group with the SDK using. 

client.resource_groups.create_or_update(resource_group, {"location": location}})

I wait for it to be created by checking the value of:

client.resource_groups.get(resource_group).properties.provisioning_state

Which eventually has the value of "Succeeded.

Several seconds after that, my script attempts to create an Azure Network within that resource group, again, using the same python azure sdk. This fails with "resource groupe could not be found".

client.virtual_networks.create_or_update(resource_group,
                                         name, {
                                               "location": location,
                                               "address_space": {
                                                    "address_prefixes": [netmask]
                                                }
                                                }).result()

Note: up until recently this worked perfectly fine.

Only late last week things started failing (approx 40-60% of the time).

And indeed, if shortly after such a failure I list the azure resource groups in my subscription using the "az" cli, the resource group indeed is not in the list.

However, in some cases (not always), the resource group does eventually (e.g. after 30-120 minutes) show up in the resource group listing.

I'm perplexed, to say the least.

Packages I'm using:

python-azure-sdk-2.0.0
python-msrestazure-0.4.7

I know that AWS has a concept of eventual consistency. I.e. some resources, after being created, might take some time before they are available in the entire region. But in this case we're talking about hours. And in some case the resource group does not eventually show up at all.

Again, I must stress this very same code with the very same SDK packages used to work just fine until late last week. I'm eager to conclude something changed on the Azure's end.

I should add that I run several instances of this script in parallel (several different resource groups and networks created at the same time). This is when I often see this problem. So far I did not see it if I run only a single instance of this script.

I'm seeing this issue for several different regions.

Md Farid Uddin Kiron
  • 16,817
  • 3
  • 17
  • 43
pomagranad
  • 1

1 Answers1

0

it looks like azure is having some issues in the last couple of days, just give it some time, it will settle just fine. I observe random failures all the time last couple of days. It is starting to recover today.

4c74356b41
  • 69,186
  • 6
  • 100
  • 141
  • Indeed, this might be something with Azure. We're still seeing the same issues. though in the past 5 or so days there was 1 single day during which everything worked fine. But the day after that the issues resumed. – pomagranad Feb 04 '19 at 07:03
  • i dont see those issues anymore – 4c74356b41 Feb 04 '19 at 07:05
  • We seem to be talking about the same issues, indeed, the last we saw the issues during the night from 3rd to 4th March. But we didn't see them during the last night, from 4th to 5th. that said, last weekend there was also one night during which the issues were gone, but they returned the next day. – pomagranad Feb 05 '19 at 11:01
  • transient failures cloud is famous for. My infrastructure tests are failing all the time with random errors, so sometimes things just dont work in the cloud. – 4c74356b41 Feb 05 '19 at 11:03