Hen WOW, thanks, I will have to read this several times to digest. A big answer for a quarter word. Thanks so much for your time and effort.
John R. On Oct 25, 2004, at 1:50 PM, Henri Yandell wrote: > > (I don't know specifics, but this should be close enough, based on > journaling on linux) > > Journaling is a file-system feature whereby every change to the file > system is kept in a journal. So say you create a new file named BOB, > there will be the file named BOB, the pointers that say where BOB is > etc, but there will also be a file named 'JOURNAL' which contains a > line: > > 2004-10-24 13:43:56.982 CREATE FILE 'BOB' <data-in-file> > 2004-10-24 13:45:59.102 MODIFY FILE 'BOB' <diff-between-old-and-new> > 2004-10-24 13:47:12.134 DELETE FILE 'BOB' > > The reason for this is because while you think there is a file named > BOB, that file probably only exists in memory to start with and hasn't > been flushed down to the disk. If the machine crashes, anything that > hadn't been flushed is lost and the disk is often left in a slightly > broken state (thus the tool fsck (file system check) on non-journaling > systems). With a journal, instead of running something like fsck and > fixing the problems by deleting anything not quite correct, the > operating system can replay the journal from the last known flush > point. > > It's a classic programming problem. Talking to resources is slow, so > you create a cache above the resource. However, now the resource and > your cache get out of sync with each other and you have a journal > which is in the fastest part of the resource (aka the disk) and is the > only part kept up to date all the time. Whenever you flush to disk, > you also reset the journal. > > I would usually not bother to turn the journaling on until the first > day I have a crash and feel I've lost the data. Bear in mind, that a > crash for an external drive can be as simple as the cable being pulled > out (or maybe the power cable). That said, the cost, (assuming Apple's > code is mature and non-buggy) is pretty low for journalling (from a > user perspective). It's better to be safe than sorry. > > You may find that it's less desirable for lots of small files, or lots > of large files; it depends on the file-system implementation in > question. > > That's the brain dump :) Probably lots of corrections needed as this > is on the edge of my expertise area. > > Hen > > On Mon, 25 Oct 2004, John Robinson wrote: > >> I have a question that the folks in the real know will have an answer >> for. When I use TechTools Pro to optimize the hard drive I first have >> to turn off "Journaling". Once done I then turn it back on. >> >> What is Journaling? >> >> Also, when I mount a firewire drive the Journaling is turned off by >> default, so, once it is optimized is it better to turn "on" the >> journaling for the firewire drive? >> >> Thanks, just trying to understand what I am doing at the instruction >> of the utility software >> >> John R. >> >> >> >> | The next meeting of the Louisville Computer Society will >> | be October 26. The LCS Web page is <http://www.kymac.org>. >> | List posting address: <mailto:macgroup at erdos.math.louisville.edu> >> | List Web page: <http://erdos.math.louisville.edu/macgroup> >> > > > | The next meeting of the Louisville Computer Society will > | be October 26. The LCS Web page is <http://www.kymac.org>. > | List posting address: <mailto:macgroup at erdos.math.louisville.edu> > | List Web page: <http://erdos.math.louisville.edu/macgroup> > | The next meeting of the Louisville Computer Society will | be October 26. The LCS Web page is <http://www.kymac.org>. | List posting address: <mailto:macgroup at erdos.math.louisville.edu> | List Web page: <http://erdos.math.louisville.edu/macgroup>
