On 20/06/2013 22:14, Alessandro Selli wrote: > Alan McKinnon wrote: >> On 20/06/2013 12:32, Alessandro Selli wrote: >>> I'm just playing the devil's advocate here. >>> Most modern distributions use the ext4 driver to handle all ext* >>> filesystems. ext4 does support the most often used attributes, the LPIC >>> objective could only consider those. >>> I don't think we should worry about attributes' permanence between >>> filesystems or across archival tools, at least no more than we already >>> do about the vfat filesystem. Just assume the candidate is aware that >>> they could disappear under particular circumstances. >>> Worthiness of tools is a difficult matter to evaluate. Everything is >>> worthelss until you need it. Something is worthier than something else >>> even if you use it 1/100 the times. I do think the burden of >>> considering the append-only and immutable attributes to be pretty low, >>> and those attributes are supported by all the filesystems distributions >>> format installation storage by default. >>> >> So it's like lilo then (recently discussed here)? > > I don't think so. LiLO is no longer the default boot loader of most > (all but one?) major distros, and it's code is in maintenance only mode > AFAIK. Instead, every major distribution that I could check (Debian, > Fedora and Ubuntu) all install chattr and lsattr by default.
I forgot the tongue-in-cheek smiley I mentioned lilo only because it can be argued it's one of those things that could be dropped. The analogy doesn't stretch very far > > [...] > >> By all means mention in a >> class that they exist and which man pages to read when the student needs >> them, > > Looks like 90% the same as what I proposed. Yes > > [...] > >> But I don't know how to examine a knowledge area of "know that something >> exists" :-) > > Aren't there already several plain awareness areas? > One could be asked: > > *) What does the lsattr command accomplish? > > 1) It is the Linux System Accounting Tool for Terminal Resources > 2) It is a tool for listing extended filesystem attributes > 3) It is a popular alias for ls -atTr > 4) It is a system stress-testing tool > 5) It is an LPI candidate stress-testing tool That's a valid approach, thanks for thinking that up > >> I would *much* rather concentrate on things sysadmins *will* >> run into, such as why is it that to prevent a file being deleted the >> permission needed is applied to the containing directory and not the >> file itself? > > This is an different matter. Of course you know this does not > accomplish the same thing as the chattr command does: a directory > permissions affects *all* files inside the directory. And using chattr > you have two different ways to protect a file: you can set it immutable > or just undeletable. And you cannot implement an append-only or a > secure deletion mode on files acting on directory permissions. All > these four modes are supported by ext4. I never intended to imply they accomplish the same thing. In examination you cannot cover everything, but you can cover enough things to give a statistically meaningful result. I've always held that given the choice between two possible objectives - one somewhat common and one very much less so - the weight should fall to the common one. LPI has always worked hard to avoid being seen as an exam that has many obscure questions in it. I feel that using lsattr is one of those somewhat obscure things - fine if you need it, but I would never expect a sysadmin to be familiar with it. I would expect him or her to have the entire owner/group/perms model in their head though > >> Now that one goes right to the heart of how Unix works, and >> a very worthy exam question. > > I thought LPI was about Linux, not UNIX. Should we drop iptables > because other UNIXes don't have it? That comes across as facetious. Linux with GNU userland is at heart a Unix and for the most part sticks to proven Unix principles > >> [1] I've used attrs twice in my life, both times to immute valuable >> files, and both times I've undone it months later after hours of >> frustrating debugging. Nowadays I use RCS instead. > > RCS is not installed by default by *any* distribution that I know of. > Plus, RCS is designed to other things compared to chattr. Filesystem > extended attributes add file properties beyond the traditional UNIX > ones, while RCS "manages multiple revisions of files". They are two > different tools designed for different purposes. > > Do you really think RCS should be in LPIC-1 or -2? > Do more people use RCS than chattr? No, I never said that. I'm quite baffled how you concluded I might have meant that. Perhaps you parsed my words completely literally; I did not intend that -- Alan McKinnon [email protected] _______________________________________________ lpi-examdev mailing list [email protected] http://list.lpi.org/cgi-bin/mailman/listinfo/lpi-examdev
