Description of problem: If you install a patch cluster by scheduling the install in the WebUI and then running rhn_check on the client, the /tmp directory isn't cleaned up afterwards. If the patch cluster is huge, it could cause subsequent installs to fail due to lack of space. How reproducible: Always Steps to Reproduce: 1. Schedule a patch cluster install. 2. run rhn_check -vvv on the client Actual results: The patch cluster installs, but the files placed in /tmp are not removed afterwards. Expected results: The files placed in /tmp are removed. Additional info:
Fixed in SVN. Turns out that there was an if-statement that always evaluated to false, which was keeping stuff from being cleaned out of /tmp. Test Plan 1. Schedule a patch cluster install for a solaris client. 2. Run rhn_check -vvv on the solaris client. 3. After the patch cluster install, check the /tmp/ directory to make sure that the files are gone. The directory created in /tmp by the patch cluster install will still exist, and the directory under that named after the patch cluster will still exist, but there should be any files under that. For example, an ls would look like this: bash-3.00# ls -l /tmp/tmpf0NGmp/J2SE_Solaris_9_Recommended/ total 0
Correction: There should NOT be any files in the directory named after the patch cluster.
That's correct. PROD_READY root@rhnsun3 /tmp/tmpMVQXtL/8_Recommended % ls -la total 32 drwxr-xr-x 2 root other 69 Aug 10 16:33 ./ drwxr-xr-x 3 root other 115 Aug 10 15:44 ../