On Fri, 2003-09-19 at 22:45, Anthony Towns wrote: > On Fri, Sep 19, 2003 at 03:04:36PM -0500, Jeff Licquia wrote: > > - Interestingly, sarge's failures are nearly an order of magnitude > > higher than woody's, which was an unexpected result. I'm going to see > > what happened. > > 520|370 1 00032590 2 1|Effective UID returned (1001) not equal to VSX_UID1 > (1002 > > Those sort of bugs happen when you use "su" to become the vsx user instead > of logging in as it, IIRC. I found the most reliable way to run the tests > was logging in at the console as vsx0 (?) and running the test by hand.
Actually, these runs were under ssh, as in "ssh vsx0@<testbox>". What's really weird: my last sarge test run (on the 19th) doubled the number of failures from the previous run (1669 total). This was: - logged in to vc1 as vsx0 - running the kernel from kernel-image-2.4.21-5-586tsc The predominant error was the same as you identified: the effective UID and GID were wrong. > > - I'm using the newest test suite. > > You need to be running a more recent kernel than 2.4.18-bf2.4 to get useful > results. kernel-image-2.4.22-i386 should hit testing tomorrow. I'll get that, do a dist-upgrade, and run another test, after poking around for any obvious problems with the test suite that would cause the UID problem.
