Bug#248884: Some comments regarding last patch/diff
09:26 < finnarne> it doesn't work with lvm1 anymor
09:26 < finnarne> it could, if you load lvm-mod as well as dm-mod
09:27 < pere> the postinst change kill support for lvm1, yes. why not
install both, or select which module/package to install depending on the
kernel version?
09:27 < finnarne> if lvm2 is availible on the cd, it would be installed
anyway
09:27 < pere> what is the story with sending stderr to /dev/null when
calling vgdisplay?
09:27 < finnarne> and it works fine with 2.4-kernels
09:28 < finnarne> you get a bunch of "File descriptor [3,4,5,6,7] left open"
09:28 < pere> the missing pclose() call seem to be a pure bug.
09:28 < finnarne> so you wont be able to read the vg- size
09:28 < finnarne> yes
09:29 < finnarne> I hoped that the missing pclose was the one that cause
the "file descriptor...", but it wasn't
09:29 < pere> right. that deserves a comment in front of the call. :)
09:30 < finnarne> yes - I can maybe attach the logfiles, if I still got them
09:30 < finnarne> There is also maybe another but introduced (or left)
09:30 < pere> finnarne: great work. we should get the change working
both with lvm 1 and 2, to avoid too big surprises while we test.
09:31 < finnarne> there is a test for existing "/proc/lvm/VGs"
09:31 < finnarne> that will of course not work.
09:32 < pere> right. that is the code detecting existing volume groups?
09:32 < pere> is there a problem with instaling both the lvm10 and lvm2
packages?
09:32 < finnarne> Not sure what will happen if you somehow try to create
a vg that exist.
09:32 < pere> it will fail.
09:32 < finnarne> with a sane error message ?
09:32 < pere> not sure.
09:33 < pere> vgcreate will fail.
09:33 < finnarne> So I would guess.
09:33 < finnarne> Not sure how to test - create a table with two vg's
with the same name ?
09:34 < finnarne> no problem with both lvm10 and lvm2, but lvm2 will be used
--
Finn-Arne Johansen
faj@bzz.no
http://bzz.no/
Reply to: