I spent a while yesterday trying to figure out why reading the runs metadata array of a just-created large workflow was very slow. If the workflow is reloaded (via Workflow.reload()) the issue disappeared. It seems to be due to Windows defender scanning the chunks on read, in some specific scenarios, since switching Defender off fixed the issue.
The workaround is to add a custom directory-exclusion to Defender so it ignores activity within that directory.
I spent a while yesterday trying to figure out why reading the runs metadata array of a just-created large workflow was very slow. If the workflow is reloaded (via
Workflow.reload()) the issue disappeared. It seems to be due to Windows defender scanning the chunks on read, in some specific scenarios, since switching Defender off fixed the issue.The workaround is to add a custom directory-exclusion to Defender so it ignores activity within that directory.