(replies inline) On 02/03/2017 04:00 AM, Philip Hands wrote: > To be useful one also needs to change choose_partition/iscsi/do_option to call > iscsi_start() rather than iscsi-start, otherwise you're defining a > function that's never called. Also, one needs to define > partman-iscsi/initiatorname in the template. See: > > https://patches.ubuntu.com/p/partman-iscsi/partman-iscsi_40ubuntu3.patch >
Ah yes, I should have diffed the whole tree looking for deltas. I got a little too laser focused there. > BTW Is there a reason not to also pick up the removal of '-o update' > From the Ubuntu patch while we're about it? We seem to have as recent > open-iscsi as ubuntu, so presumably also don't need that option. It's been awhile since I've done those ops by hand, but I have used the Ubuntu installer and it does to the right thing, so those changes can be safely included. > I'd use a here document (<<"!EOF!" ...) rather than all the echos. > > Also, the -n is superfluos, and is not in line with the [ "$RET" ] usage > in the rest of the file. Their patch does seem inconsistent with the upstream style (eg: I see here documents in other files). I imagine if we get this fixed up they could drop their patch. > Does the /etc/iscsi/initiatorname.iscsi actually need to be only > readable by root? Seems needlessly restrictive to me, but I don't use > iscsi. In the grand scheme I don't think it matters, but there may be security implications of which I'm not aware. Should I work up another patch or would you like to take a pass at massaging the changes from Ubuntu? > > Cheers, Phil. > Thanks, Kevin
