Well, the patch was simply to add a mkdir -p $DATASTORE_LOCATION to the monitor_ds.sh script at the top. Can you try it and let us know if that works for you?
On Fri, Apr 4, 2014 at 3:50 PM, Hyun Woo Kim <[email protected]> wrote: > Hi Jaime and Carlos, > > Thanks very much for the explanations. > Until you release a new version that deals with this bug, > we can manually mkdir /var/lib/one/datastores in each of our VMHosts to > avoid this issue. > > Thanks again! > HyunWoo KIM > FermiCloud > > > From: Jaime Melis <[email protected]> > Date: Friday, April 4, 2014 6:53 AM > To: Hyunwoo Kim <[email protected]> > Cc: users <[email protected]>, Carlos Martín Sánchez < > [email protected]> > > Subject: Re: [one-users] data store not created in VMHosts > > Hi Hyun, > > I've looked at this issue and you're completely right. If > /var/lib/one/datastores does not exist in the VMHost it will report that it > doesn't have free space and therefore it will never deploy the VM, which > means that the directory will never be created, which is a Catch 22. > > We haven't noticed this earlier because usually people share > /var/lib/one/datastores and therefore you don't hit this issue. And people > using SSH and upgrading from earlier version will not hit it because in > previous pervious the disk capacity wasn't checked so it always created the > directory. > > Long story short, bug confirmed, applying patch to monitor_ds.sh > > Thanks for the heads up. > > Cheers, > Jaime > > > On Mon, Mar 31, 2014 at 11:25 AM, Carlos Martín Sánchez < > [email protected]> wrote: > >> Hi, >> >> On Fri, Mar 28, 2014 at 7:40 PM, Hyun Woo Kim <[email protected]> wrote: >> >>> If HostXML::test_ds_capacity only creates $DATASTORE_LOCATION/<id> >>> assuming $DATASTORE_LOCATION itself exists, >>> my next question here might be, >>> which code creates $DATASTORE_LOCATION itself and when? >>> >>> I don't think I missed an instruction to create $DATASTORE_LOCATION >>> manually in the ON documentation.. >>> i.e. $DATASTORE_LOCATION must be created by ON codes (maybe) during the >>> first VM deployment. >>> >>> But like I described in the original email, our test ON4.4 fails at >>> this. >>> >>> Hyunwoo Kim >>> FermiCloud >>> >> >> The $DATASTORE_LOCATION/<id> dir is created by the tm drivers on the >> first deployment, using the ssh_make_path function (mkdir -p). As I said, I >> think the problem is not the dir creation, is with the mechanism to decide >> the first deployment on a new host. >> >> Regards >> >> -- >> Carlos Martín, MSc >> Project Engineer >> OpenNebula - Flexible Enterprise Cloud Made Simple >> www.OpenNebula.org <http://www.opennebula.org/> | [email protected] >> | @OpenNebula <http://twitter.com/opennebula> >> >> _______________________________________________ >> Users mailing list >> [email protected] >> http://lists.opennebula.org/listinfo.cgi/users-opennebula.org >> >> > > > -- > Jaime Melis > Project Engineer > OpenNebula - Flexible Enterprise Cloud Made Simple > www.OpenNebula.org | [email protected] > -- Jaime Melis Project Engineer OpenNebula - Flexible Enterprise Cloud Made Simple www.OpenNebula.org | [email protected]
_______________________________________________ Users mailing list [email protected] http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
