Le mercredi 17 août 2016 à 18:46 +0200, Michael Scherer a écrit :
> Le lundi 08 août 2016 à 15:11 +0200, Michael Scherer a écrit :
> > Le mercredi 03 août 2016 à 14:43 +0200, Michael Scherer a écrit :
> > > Le jeudi 28 juillet 2016 à 10:21 -0700, Amye Scavarda a écrit :
> > > > On Thu, Jul 28, 2016 at 4:05 AM, Niels de Vos <[email protected]> wrote:
> > > > 
> > > > > On Thu, Jul 28, 2016 at 06:21:33AM -0400, Kaleb KEITHLEY wrote:
> > > > > > On 07/27/2016 05:18 PM, Amye Scavarda wrote:
> > > > > > >
> > > > > > >
> > > > > > > From the latest updates on the cage list, it looks like we're 
> > > > > > > tracking
> > > > > > > for these moves/downtime for August 8-9? That'll be 
> > > > > > > Monday-Tuesday.
> > > > > > > Any complaints about that schedule?
> > > > > >
> > > > > > 3.8.x releases are scheduled for the 10th of each month. In 
> > > > > > yesterday's
> > > > > > Community meeting Niels said that 3.8.2 is planned to be released on
> > > > > > schedule.
> > > > > >
> > > > > > If this goes as planned then it shouldn't be a problem, right?
> > > > >
> > > > > We just need to make sure that all patches for 3.8.2 have been tested 
> > > > > in
> > > > > the CI before the move. It shortens the development cycle a few days,
> > > > > and it would have been nice to know a little more in advance and
> > > > > mentioned in the community meeting.
> > > > >
> > > > > I do not think there are any critical patches for the release, so I do
> > > > > not expect any problems with the outage either. (Although it really
> > > > > isn't nice to treat 3.8 releases as guinea pig for infrastructure
> > > > > changes.)
> > > > >
> > > > > Niels
> > > > >
> > > > > I brought this up to the community cage group this morning, and we'll 
> > > > > look
> > > > to move the VMs after the 3.8.2 release.
> > > > I'll let Michael put in more details around exact timing.
> > > 
> > > So we still do not have the exact timing, that's waiting on IT to
> > > configure the network port (and then I have to copy data, and configure
> > > the server for new network, and admin cards, and various stuff). 
> > 
> > So After coming back from weekend, and dealing with my backlog of mail
> > and expenses, I just received a notification that IT did moved the
> > server (on friday evening) and it does even answer to ping on the admin
> > interface. So I will configure it for internet access later today or
> > tomorow (depending on my capacity to read all mails and doing meetings),
> > and will then plan to the test move once I am confident the server is
> > ok.
> > 
> 
> So, news about the server.
> 
> I did see there was some weird lvm corruption (that I didn't
> investigate), but couldn't find the exact fix. Turn out that this was
> just removing extranous PV from the VG and that's it. However, since we
> are speaking of moving production workload on it, I will need to
> reformat it to use hardware raid, and so doing that tonight. 

Ok so I did reformat and rename the server (with some pain, since the
idrac interface is a bit annoying, and I did hit a few roadblock with
java on linux, with Centos iso, with anaconda partitioning choices) Now,
that's 4 disk in hardware raid 5, and the name is
myrmicinae.rht.gluster.org (a type of ants).

I will finish the ansiblization later.

> Then i will start to test the copy of VM once that part is done.

On that front, seems doing any kind of disk snapshot requires some
downtime, because RHEL do remov^W differenciate features between qemu
for RHEL and qemu for RHEV, so Centos inherit the limitation. 

So I guess we might need a few reboot, which will also permit to get
kernel upgrade, etc. I have no idea of the needed downtime, but since
that's just a reboot, I hope I can do that outside offices hours.


-- 
Michael Scherer
Sysadmin, Community Infrastructure and Platform, OSAS


Attachment: signature.asc
Description: This is a digitally signed message part

_______________________________________________
Gluster-infra mailing list
[email protected]
http://www.gluster.org/mailman/listinfo/gluster-infra

Reply via email to