On Monday 23 Jan 2006 10:37, Gilles wrote: > The idea is that if one disk fails, and the swap on that one is corrupted, > the system will crash...
And that is *good* -- a crashed system has to be dealt with there and then, and is not going to keep limping along, possibly making matters slowly worse and worse; as I have seen happen in practice with old, legacy systems. After all, if your swap partition is corrupt, how do you know it's the only thing? While I think it's generally good to keep a server up as long as possible, I also believe there are times when it needs to come down for damage limitation. > > With the kind of hardware RAID that appears as a single > > SCSI drive, you obviously can't avoid RAIDing swap; but at least it isn't > > such a performance hit. > Some people have pointed out that linux's software RAID is probably more > efficient then the common cheap (fake) hardware RAID. Indeed; but I was talking about expensive, true hardware RAID. If your "RAID controller" is not really a RAID controller, then by all means use md. > On my system (using "mdadm"): > > # ls /etc/raidtab > ls: /etc/raidtab: No such file or directory The configuration file might well have moved since I last set up an md RAID system ..... that's the thing to beware of with any reliable system; the construction often outlasts the design. As always, check your own setup and make appropriate substitutions. -- AJS delta echo bravo six four at earthshod dot co dot uk -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]