Closed GreatScott000 closed 3 years ago
My assumption is that the problem is in your network, I see it very very seldom in my pipelines. AzureEdge is a CDN, from which downloading is much faster than directly from the blob storage. I don't know why it sometimes fails, but I added the retry in order to make sure that the entire pipeline doesn't fail due to a CDN download fail - and when it retries, it retries from the blob storage. It would be fairly simple to add one retry from the CDN before reverting to blob storage - not sure whether that would take more time though. If you want to test whether a second download from CDN works, you can modify this line: https://github.com/microsoft/navcontainerhelper/blob/7cb1bf561c6dfe8f68a9430256f36cfdaf816748/Misc/Download-File.ps1#L66 to just use $sourceUrl instead of $newSourceUrl - then it would retry from CDN Let me know if that works?
I see failures when downloading from CDN once a week (after looking at logs) and from looking at other issues with people downloading from CDN, an immediate retry from CDN is not a solution.
I do realize that this problem might be more outspoken on some networks, but it isn't something I can fix, the problem is in your network or on your machine.'
If you want to avoid the retry and always just go for the blob download, you can set the fully qualified storage account when getting the artifact url:
Get-BCArtifactUrl -storageAccount 'bcartifacts.blob.core.windows.net' -country us
Then it will not retry, but it will download from blob storage directly, which is somewhat slower than CDN.
PLEASE DO NOT INCLUDE ANY PASSWORDS OR TOKENS IN YOUR ISSUE!!!
Describe the issue I often see the artifact download fail and fall back to another URL. This fail/failover process seems to take an inordinate amount of time (a number of minutes).
Why is this download failure happening? Is there a problem with the default artifact URL? Is there a throttling issue?
Can the "fail/failover" happen faster?
Scripts used to create container and cause the issue
I have include the full results below but it is not really really relevant. Here is the relevant section
When it doesn't have an issue downloading the artifact it takes less than 5 minutes to build the container
When it fails the first time it takes nearly 3 times as long (13+ minutes)
Scripts used to create container and cause the issue
Screenshots If applicable, add screenshots to help explain your problem.
Additional context
Full output of scripts
Screenshots If applicable, add screenshots to help explain your problem.
Additional context