Code review comment for ~redriver/cloud-init:frbsd-azure-branch

Revision history for this message
Hongjiang Zhang (redriver) wrote :

> > > Hi,
> > >
> > > I have been doing some testing on FreeBSD.
> > >
> > > The current growpart seems to work fine with FreeBSD 11.0 with GPT/UFS -
> the
> > > partition gets resized. The issue is with filesystem resize in resizefs.
> Its
> > a
> > > simple fix, before you can modify the MBR you must first set
> > > kern.geom.debugflags=16, then when calling growfs you must pass the '-y'
> > flag.
> > > You should then set debugflags back to it's original value.
> > >
> > > Basically this, I did a quick hack to test and seems ok:
> > >
> > > current_debugflags = sysctl -n kern.geom.debugflags
> > > sysctl kern.geom.debugflags=16
> > > growfs -y /dev/vtbd0p2
> > > sysctl kern.geom.debugflags=current_debugflags
> > My problem is growfs always reports error since the disk is already full
> > expanded.
> >
> > # growfs -y /dev/da0p2
> > growfs: requested size 28GB is not larger than the current filesystem size
> > 28GB
> > # echo $?
> > 1
> >
> > My disk is formatted to be UFS:
> > # gpart show
> > => 40 62914480 da0 GPT (30G)
> > 40 1024 1 freebsd-boot (512K)
> > 1064 58719232 2 freebsd-ufs (28G)
> > 58720296 3145728 3 freebsd-swap (1.5G)
> > 61866024 1048496 - free - (512M)
> >
> > => 63 146800577 da1 MBR (70G)
> > 63 1985 - free - (993K)
> > 2048 146796544 1 ntfs (70G)
> > 146798592 2048 - free - (1.0M)
>
> Not sure what the most elegant solution would be to avoid exit(1) when there
> is nothing to do...
>
> Just thinking out loud here:
>
> 1.
> Can resizefs be called to ignore exit status 1? Could be dangerous?
>
> 2.
> Is there a way to pass back from growpart to see if gpart size did anything,
> if so then run growfs?
>
> 3.
> You could do a dry run first and check the exit status before continuing:
> # growfs -N -y /dev/da0p2
>
> 4.
> Could we look at the current size of the file system and compare to gpart,
> (there is probably a better way to get the fs size):
>
> # dumpfs -m /
> # newfs command for / (/dev/da0p2)
> newfs -O 2 -U -a 4 -b 32768 -d 32768 -e 4096 -f 4096 -g 16384 -h 64 -i 8192 -j
> -k 5240 -m 8 -o time -s 83885872 /dev/da0p2
>
> Grab -s <size> round it and compare against
>
> # gpart show
> => 40 83886000 da0 MBR (466G)
> 40 128 1 freebsd-boot (64K)
> 168 83886000 2 freebsd-ufs (40.0G)
Firstly, I appreciate your help to resolve this issue. I prefer the 4th solution. Thanks.

« Back to merge proposal