Open charris-msft opened 1 year ago
Hey @charris-msft , thanks for reporting it. Performance is always tricky, and I'm glad that you found an anomaly by comparing with the deployment center timings.
Would you mind rerunning azd up --template https://github.com/tonybaloney/django-on-azure
with debug
enabled (assuming it reproduces consistently), and sending the logs offline (debug logs might contain az
debug logs which itself can be sensitive)? I'm going to try and do the same on my end.
The following command should work:
azd up --template https://github.com/tonybaloney/django-on-azure --debug 1>azd.out 2>azd.err
.
Hi @weikanglim - I tried the command you suggested, but it just hung for about 20 minutes - I'm guessing the prompt for the environment name, region, etc was swallowed up by stdout.
I'll get the --debug, but without redirecting the output and share it in just a minute
Hey @charris-msft did you get the debug output?
@charris-msft Thanks for submitting the debug output.
Looking at the output, I see all the correct behavior on azd
. A deployment is created through az deployment sub create
, and every 30 seconds, az
was polling the ARM API to see if the deployment was completed. At 00:15:12 UTC, it completed and azd
correctly continued with application deploy. The ARM deploy started at 00:06:57 UTC, so this took about 9 minutes.
For the application deploy, that took 8 minutes to complete. These two operations took majority the execution time and are all dependent on Azure services on how long it might take.
Output from
azd version
azd version 0.2.0-beta.1 (commit 817a81ab44ac788ec2d3050e94ba0138c3a80ac3)
Output from
az version
Describe the bug With the previous version of azd I observed
azd up
templates for this repo taking 3 minutes.After upgrading to 0.2.0-beta.1, I've deployed twice.
I noticed when comparing my recording to the Deployment Center logs for one of the attemptes, there appears to be a 4 minute gap between when the Deployment Center indicated the deployment was complete and when the CLI updated to indicate the deployment was complete.
JonG thought there might be a delay in the api.
To Reproduce
azd up --tempate https://github.com/tonybaloney/django-on-azure
Expected behavior After talking with JonG, about improvements in the current release I expected the deployment to be at least as fast if not faster than the previous version.
Environment Information on your environment:
Additional context Add any other context about the problem here.