Description of problem: In a very exciting moment, it has been theorized that sufficiently long hostnames, when combined with the length of the session URIs created/accessed by anaconda during a kickstart, can create a URL of greater length than anaconda can gracefully handle. Version-Release number of selected component (if applicable): 3.6 How reproducible: Always Steps to Reproduce: 1. Use a machine with a fairly long hostname, such as test01.x86_64.rhndev.redhat.com 2. Set up a few different kickstarts, including a rhel-3-as-i386-u3 and a rhel-3-desktop-x86_64-u3 3. The desktop kickstart should break in anaconda with a "failure to retrieve URL" message. 4. However, if you then create an alias of x.rhndev.redhat.com, thus shortening the hostname, the kickstart will work! See mspevack for more details. Actual results: Expected results: Additional info: This bug was discovered during the testing of bug #131094
bumping up priority
Ok - fix is in CVS. Test plan: Test both RHN-hosted kickstarts (on satellite), and externally hosted kickstarts (both on satellite and in qa). The RHN-hosted kickstarts should have the --url line changed transparently to be a 'tiny url'. Non-RHN hsoted kickstarts should not be affected.
QA push. {ON_DEV,QA_READY} --> ON_QA
DOC notes: We should document the fact that on satellite, we no longer use the long URL with the hash info (a42bc4398d...) - we use a much shorter URL with hostname/ty - this is because it was possible to hit the internal anaconda limit for the length of this URL with reasonable-but-long hostnames.
Functionality looks good on QA. Waiting to test docs.
Mass move from PROD_READY to CLOSED:CURRENTRELEASE