Colin Charles <[EMAIL PROTECTED]> wrote: >> Ever thought about runaway processes, hitting swap, on a slow disk (like those on laptops), that the OOM doesn't get to? Yes, this happens <<
It doesn't happen nearly as often as a system becoming overcommitted due to an unanticipated load - whether it's more clients hitting a daemon than expected, or the user simply trying to run every application on the system at once. >> If however there's no swap or a small swap, the OOM killer is given a chance to attack when it hits a runaway process << I don't recall seeing a "runaway process" of this type - presumably something that's leaking (allocating and not freeing) memory at an incredibly fast rate on any production system I've been involved with. I certainly would never advocate configuring a system to have no or very small swap on the basis that it makes it easier to deal with what is an exceptional occurrence. >> This was fixed in an update to OOo, so make sure this is via a stock FC-3 installation << And I certainly wouldn't advocate configuring a system with no swap just to mitigate what is obviously a bug; the correct approach is to get the bug fixed, as happened here. No swap at all, on office workstations, is asking for trouble. I'll be sticking with the old rule. Best, --- Les Bell, RHCE, CISSP [http://www.lesbell.com.au] _______________________________________________ lpi-examdev mailing list lpi-examdev@lpi.org http://list.lpi.org/mailman/listinfo/lpi-examdev
