[pve-devel] quorum using Rasberry Pi as node

[Wolfgang Bumiller]

> On May 22, 2016 at 11:37 PM Thomas Lamprecht <t.lamprecht at proxmox.com> wrote:
> 
> 
> 
> 
> On 22.05.2016 22:48, Michael Rasmussen wrote:
> > On Sun, 22 May 2016 16:44:48 +0200 (CEST)
> > Dietmar Maurer <dietmar at proxmox.com> wrote:
> >
> >>>> This would be an interesting task, but I would be more interested
> >>>> in pve-cluster (to manage corosync.conf).
> >>>>
> > I have looked into this and this is what I have found out:
> >
> > $ file /usr/bin/create_pmxcfs_db
> > /usr/bin/create_pmxcfs_db: ELF 64-bit LSB executable, x86-64, version 1 (SYSV), dynamically linked, interpreter /lib64/ld-linux-x86-64.so.2, for GNU/Linux 2.6.32, BuildID[sha1]=160c6b57da02b9b0e86be3b8995d7097ec8f7558, stripped
> > $ file /usr/bin/pmxcfs
> > /usr/bin/pmxcfs: ELF 64-bit LSB executable, x86-64, version 1 (SYSV), dynamically linked, interpreter /lib64/ld-linux-x86-64.so.2, for GNU/Linux 2.6.32, BuildID[sha1]=dd96ad8a35fee8a9013efcc17dd20aa0c69231bb, stripped
> >
> > Dependencies:
> > libcorosync4-pve
> > corosync-pve
> > libpve-common-perl
> >
> > libcorosync4-pve and corosync-pve could be replaced with the versions from jessie-backports (2.3.5-3~bpo8+1)
> >
> > 1) To have this working a porting of create_pmxcfs_db and pmxcfs will
> > have to be done to armhf.
> > 2) pve-manager will need to be aware of ignore_nodes: <list_of_nodes>
> > in datacenter.cfg. Possibly configurable from gui.
> >
> > Is there something I386 specific in create_pmxcfs_db and pmxcfs?
> >
> 
> Raspberry 1 here with Raspian Jessie, did not care for build times and its good to test low performance ;)
> 
> I built packages for libqb, corosync-pve  and pve-common, no big problems, only two qb test fail but only for test timeout reasons.

Did you verify the syscall numbers in pve-common?

> I'm onto pve-cluster, there are some portability issues, mostly things which would fixed by setting W-no-format.
> It's stuff like size_t variable but %lu format and uint64_t but %zX, ... nothing all to problematic.

For 'size' there's %z, but for fixed-size types like uint64_t you'll probably want to use the PRI* macros or cast the parameter (which is easier to read).