Description of problem:
When starting several yum jobs in parallel, starvation can occur.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Run a "yum" job in a shell script loop.
2. Also run a yum job by hand in a separate window.
Observe that the shell loop will nearly always hold on to the yum lock, starving the separate one.
Yum locks should be more fair, so a yum job waiting for a long time should
get priority over one that is freshly started.
there's no internal communication between yum processes. They're just watching a lock file and checking back. The one that checks most recently after the last lock owner exits will get it.
I'm closing this cantfix but it really should be cantfix-without-a-herculean-effort.