[Rdo-list] Image generation of current stable hangs

[Steven Hardy]

On Wed, Dec 16, 2015 at 05:52:30PM +0800, 韩潮 wrote:
>    Hi All,
> 
>    Currently I am experiencing issues with the installation of RDO/TripleO. I
>    followed the installation guide [1], but it seems the process for image
>    generation is stuck each time on the same spot.
> 
>    At the preparation I pulled the current Delorean repo file:
> 
>  sudo curl -o /etc/yum.repos.d/delorean-liberty.repo https://trunk.rdoproject.org/centos7-liberty/current/delorean.repo
> 
>    as of now it refers to:
> 
>  baseurl=http://trunk.rdoproject.org/centos7-liberty/fc/09/fc098c4bad45ff168fe33b5fc7d11c3c45686bce_dd6915d9
> 
>    When I perform the overcloud image generation:
> 
>  openstack overcloud image build --all
> 
> 
>    It never gets out of the profiling step
> 
>    + echo '--------------------- END PROFILING ---------------------'
>    --------------------- END PROFILING ---------------------
> 
>    meta-data=/dev/loop0             isize=256    agcount=4, agsize=282584
>    blks
>             =                       sectsz=512   attr=2, projid32bit=1
>             =                       crc=0        finobt=0
>    data     =                       bsize=4096   blocks=1130336, imaxpct=25
>             =                       sunit=0      swidth=0 blks
>    naming   =version 2              bsize=4096   ascii-ci=0 ftype=0
>    log      =internal log           bsize=4096   blocks=2560, version=2
>             =                       sectsz=512   sunit=0 blks, lazy-count=1
>    realtime =none                   extsz=4096   blocks=0, rtextents=0
> 
>    The machine seems to hang. I have performed a virsh destroy and start of
>    the instack machine at least twice, with the same result. And have also
>    regenerated the whole deployment performing the same steps to see if this
>    was reproducible. Unfortunately it seems to be the case.
> 
>    We have been able to succesfully deploy and install the overcloud before
>    with a previous stable release.
> 
>    What could cause this?

This can happen if you have a machine without enough RAM, see this patch
for details:

https://review.openstack.org/#/c/234860/

It looks like we don't have that patch on stable/liberty, so I just
proposed it as a backport:

https://review.openstack.org/#/c/258397/

The workaround is either ensuring you have an undercloud with RAM of at
least 12G or do export DIB_NO_TMPFS=1 before the image build.

Steve