Closed robododge closed 3 years ago
@egor-bryzgalov I am running into this same issue.
to work around this, I had to upload the OpenJKS package needed to the root of our blob storage container, rather than into an OpenJDK folder inside the container.
It also feed like a waist and a danger, that the installer searches the entire drop folder looking for all zip files in the container, if it is going to fail to install it if the zip file it finds is not in the root of the blob container...
It would be grate to have a separate input parameter for the "source path" relative from the drop container, that would both limit the search of files, as well as fix this duplication of directory names in the path.
Me too - we came across this problem today as well, really could do with a fix
Hi, @ArthwitRail We fixed this problem. Updated task version 0.194.0 will be deployed in the next 5 weeks. Feel free to ping us if you will have any problem with it.
Hi everyone, fix for this issue has been already rolled out - closing this at the moment. Please let us know if you have any other questions.
Question, Bug, or Feature?
Type: Bug
Enter Task Name: JavaTooInstarllerV0
Environment
Server - Azure Pipelines or TFS on-premises? Pipelines
Agent - Hosted or Private: Private: Ubuntu 18, agent version 2.164.8
Issue Description
When using the azure storage account to load in jdk gzip file, if a nested path to the jdk file is supplied, file extraction fails because the azureCommonVirtualFile path does nor handle a prefixed extra path as described in tool tips. The resulting logs indicate an attempt to stat the tar.jz file from a with a double stacked path $(Pipeline.Workspace)/jdk15/jdk15/openjdk-15.0.1_linux-x64_bin.tar.gz which does not exist. Note, the doubling of the jdk15 intermediate path.
Troubleshooting
Tried moving the jdk artifact to the top-level directory in the storage account blob. That worked because it removed the need to use an extra path in the azureCommonVirtualFile property
Error logs