Mailinglist Archive: opensuse (3901 mails)

[Örn Einar Hansen <orn.hansen@xxxxxxxxxx>]

Þann Mánudagur 03 janúar 2005 22:38 skrifaði David Krider:
> Want to know the single best way to hang a Unix box so badly as to need
> a reboot?
  That's an oldie, and not really valid ... but I do recall there were several 
buggy NFS implementation that caused the hang to remain, even after the 
server came up.

> And sometimes it's pretty dang inconvenient.
  The idea *is* that the server you use to "export" the shares is the robust 
machine on the network.  Not some lil'crapper, that nobody cares about and 
nobody watches after and one that sooner or later is gonna run out of 
diskspace, or get full with logs.

  This was a problem, when the clients had even their root file system 
exported from the server.  Or at least, most of the major system files ...  
where the clients were so called "terminals" ... they do freeze when the 
server is down but only because they're waiting for the server to reappear.

  I do recall, as previously stated, that some linux boxes remained frozen, 
even after the server came up ... don't think it does anymore, but then I 
haven't tested it.

  However, if you are trying to "imply" that CIFS are better ... than I 
suggest you start reading more carefully ... like "oplocks".  It's not a 
question on "faultlessness", but rather one of "convenience".   Of course 
you'll say hang is not "convenient" ... but it is, when you really need the 
file system to be there it's better to "hang around" until it is ... rather 
than to have the data vaporize into the ozone layer as it does in cifs.

My 2¢ worth,
  Örn