Current-Users archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: netbsd-6 instability - vmem

  I'm having new stability problems on two recent -rnetbsd-6
installations as well.  In one case, on an Intel quad core with
4GiB RAM (amd64) used as my workstation with X on console.  In the
other, on a Mac Mini with 512MiB (macppc) used as a server.  They
both lock-up non-repeatably--but I believe always when some aspect
of the system is heavily loaded: I haven't figured out whether it's
memory consumption, disk I/O or CPU utilization, or comb. thereof.
This happens somewhere between roughly a couple times a week and
several times a day, depending on what I'm doing.

  Both systems use cgd devices backed by raidframe for storage
(including for swap space), with WAPBL enabled on the filesystems;
and both have IPSEC enabled in the kernels.  A lot of unused en-
tries are commented out of both of their kernel configurations and
some entries added in (using GENERIC as a starting point).  I don't
think I've tested pure GENERIC for stability.  I may be able to if
that's highly advisable, but would prefer not to; e.g., no IPSEC,
and I forget what else I'd be missing, although IPSEC has been
inoperative for me since the upgrade anyway (a problem I may need
to ask about in a separate thread).

  When the lock-ups have occurred, I haven't been able to get into
ddb due to unresponsive consoles.  Usually, wdog timers haven't
managed to cause reboots, but they did in at least one instance.

  Enabling "option DEBUG" hasn't fixed stability.  I'll be trying
LOCKDEBUG and DIAGNOSTIC next.  I can report the result if that's
desirable.  Input on how else to debug/fix this issue is welcome.

-Dave B.

Boston, MA

Home | Main Index | Thread Index | Old Index