MySQL has problems creating threads under RH 6.0 due to
what looks like a glibc2 bug. This is actually a regression
to a bug in an earlier release of RH.
If you compile MySQL and then try to run safe_mysqld, it
looks like everything works fine until you take a look and
see that only one MySQL process/thread is running when in
fact three threads are supposed to be created upon MySQL
startup. As a result, the mysqld daemon is useless. It does
not create its unix socket nor does it listen on port 3306
as it's supposed to.
When I had this problem on RH 5.1, upgrading to the latest
glibc libraries fixed the problem. After upgrading to RH
6.0, the problem is back again.
Note that this is on a sun4c workstation (SPARCstation 2).
I don't know whether the regression occured at GNU or at
RedHat, so I'm submitting it here for you to figure out. :)
Is the problem still experienced with glibc-2.1.2-1 and later?
Please reopen if the problem still persists.