On Wed, 16 Feb 2000, Derek Martin wrote: >> Hmmmm, by filesystem corruption, do you mean actual damage to the >> filesystem structure itself, or simply lost user data? > > I can't answer that, but does it matter? Well, yes. :-) If I lose my network connection long enough that things start timing out, I expect uncommitted writes to be lost. If it is that vital, I'll take steps to make sure the network connection stays up. In other words, I expect a loss of connectivity to result in the loss of access to the resources provided by that connectivity. But I *don't* expect a loss of connectivity to result in the widespread destruction of the filesystem on the server! That's ridiculous! (However, as a long-time computer user, I have learned that "ridiculous" and "reality" are far from mutually exclusive...) > If the write was modifying the inode table (I have no idea how this is > handled on NFS...) then I imagine it's possible that you'll hose the FS. Ugh. I was hoping that NFS didn't go that low-level, but given the fact that NFS has been moved into the kernel with Linux 2.2 ... blech. > if you're paging in a swapped process on some systems, they will re-read > the text segment from the disk ... and if the system loses contact with > the server that the executable was shared from, it will cause > system-dependent problems ranging from swap errors to kernel panic. Double ugh. Nice error recovery there -- NOT! I've noticed that there seems to be an "error threshold" in Unix. Up to a certain point, everything works fine, but if things get hosed badly enough, the system just throws up its hands and says "I give up!". Linux doesn't seem to do that as often as the commercial Unixes; one of the (many) things I like about it. :-) Oh, well. Mucho thanks for the good info, Derek. I think I'll add that NFS book from O'Reilly to my To-Buy list.... :-) -- Ben Scott [EMAIL PROTECTED] ********************************************************** To unsubscribe from this list, send mail to [EMAIL PROTECTED] with the following text in the *body* (*not* the subject line) of the letter: unsubscribe gnhlug **********************************************************
