Hello Everyone, On Thu, Apr 24, 2008 at 1:01 PM, Ola Lundqvist <[EMAIL PROTECTED]> wrote:
> Hi Christian > > Thanks for the clarification. I have marked this bugreport as wontfix > because of the vzmigrate issue, and because that this already exists > even though it is not document. > > However maybe it should be documented, but that is a separate issue. :) First, this must be documented, in 'man vzctl' (probably at files section, but I'd mention it in start/stop/mount/umount sections as well), and there ought to be example files provided (the ones Christian quoted are just fine; except exit values should be mentioned / demonstrated, like checking existence of the mounted-into directory and fail accordingly). Second, should I try to open a bug for it in Debian BTS and edit the manpage and provide a patch, or (preferably) any of you would love to get it done? :-) Third, vzmigrate manpage should mention this problem about in-mounted subtrees, and maybe check it even before progressing, but I cannot tell out of my brain whether there is a portable check for that. Still, I believe more people would mount large storage into VE than use vzmigrate. ;-) Thanks for your time & efforts, Peter ps: still gotta test whether .mount and .umount works ;) not that I don't trust you, erm... I mean... :-)) > > Best regards, > > // Ola > > On Thu, Apr 24, 2008 at 10:23:33AM +0200, Christian Hofstaedtler wrote: > > Hi! > > > > > > > Do we need a poststop as well? > > > > > > > > You're right, I happen to be lazy and usually let the system to umount > > > > all; but with high probability we should umount every manually mounted > > > > cruft before stop tries to unmount the whole VE, so a prestop should > > > > exist the same way prestart was implemented. > > > > > > And by all means feel free to post it upstream, you may be more > > > > familiar with the process than I am now. > > > > We are doing similar things with the /etc/vz/conf/$VEID.mount and > > /etc/vz/conf/$VEID.umount scripts; they are automatically handled by > > vzctl, but I'm not sure this is documented (except somewhere buried > > in old VZ pdfs). > > > > mount and umount obviously run at vzctl mount/umount time, which is > > pretty much pre-start and post-stop, and also does the "right thing" > > if manually starting/stopping a VE. > > These scripts can also prevent a VE from starting/stopping, if they > > return a non-zero value. > > > > Maybe you want to look at these first? > > > > Examples: > > cat /etc/vz/conf/201004.mount > > #!/bin/sh > > mount -onoatime /dev/raid10/ve201004-pg /var/lib/vz/root/201004/srv/db/pg > > > > cat /etc/vz/conf/201004.umount > > #!/bin/sh > > umount /dev/raid10/ve201004-pg > > exit 0 # ignore error from umount > > > > > > BTW, mounting filesystems into VEID/root probably breaks vzmigrate, > > so one needs to be careful :-) > > > > -Christian > > > > > > > > > > -- > --------------------- Ola Lundqvist --------------------------- > / [EMAIL PROTECTED] Annebergsslingan 37 \ > | [EMAIL PROTECTED] 654 65 KARLSTAD | > | http://inguza.com/ +46 (0)70-332 1551 | > \ gpg/f.p.: 7090 A92B 18FE 7994 0C36 4FE4 18A1 B1CF 0FE5 3DD9 / > --------------------------------------------------------------- > > -- byte-byte, grin -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]