This just bewildered me: in 3 years of running SharePoint 2007 I had never seen any major stability problems in any of the many different configurations we'd set up. Except now, when everything was brought down by one tiny little configuration change.
What confused the hell out of me was that we had installed SharePoint 2007 SP2 on this "1-Server Farm" a day before, so initially we thought it was an after-effect of the brand new service pack, but it wasn't.
What also was confusing was that even the non-SharePoint application pools (websites) on this machine were brought down. So it really looked much more like a general IIS problem, rather than a SharePoint issue. However, it turned out to be kick-started by a tiny little change in SharePoint.
As part of any install, we double-check the event logs and the main settings in the Central SharePoint Administration. As this Farm was not configured for Search, a warning was listed in the event log, and when checking the search settings in the SharePoint administration, the setting was then changed from "use all WFE servers for crawling" to the dedicated single server in this farm.
While we do run this SharePoint setup on an x64 version of Windows Server 2008 and also had IPv6 enabled on this machine, the entry that was created in the "hosts" file was an IPv4 address. Still, the problems were as described.
Deleting the entry from the hosts file and disabling IPv6 on this machine brought everything back to normal (app pools had to be restarted).
While I've seen, many times, that even the smallest change can have fatal consequences, this one got me really thinking: how can such a fatal flaw still not be fixed in SP2, practically almost a year after a Microsoft support staff had already isolated it??