=== strive is now known as paragon === paragon is now known as strive === strive is now known as parag0n === parag0n is now known as strive === exxxit is now known as doublehelix === strive is now known as par4g0n === robot is now known as uptime === exxxit is now known as doublehelix [06:57] soahccc: dd /dev/zero is more then enough. [06:57] soahccc: There is even papers on it [06:57] soahccc: or /dev/urandom if you want to wait longer [07:01] I have an issue, that Japanese and Chinese are not displayed in the installer of ubuntu-server (bionic) daily build March 6th iso image. And, should I where to report? thank you for your read. [07:01] Good morning === tobasco|away is now known as tobasco [09:14] rharper, for the dhcpclient pivot to root, would it be sufficient to continue running dhcpclint from the initramfs, post pivot? or for example continue running dhcplient post-pivot, in addition to / in competition to netwokrd? [09:15] rharper, is there a MAAS bug you mentioned, w.r.t. recovery on bionic? [09:15] smoser: ^ [09:15] things will not be killed post pivot if their argv[0] is @ [09:15] * xnox <3 hacks & cookies [09:16] dhcp in the initramfs is klibc [09:16] which does nothing but one shot [09:16] if you wanted to run dhclient, that might be sane [09:16] but you'd have to deal with moving its open file handles and such over [09:16] ugh [09:16] you can probalby run it in the initramfs in a way that it writes its leases to /run/ [09:17] but that doesnt happen now. [09:17] ack [09:17] or just generate /run/systemd/network/*.network file which basically says to dhcp [09:17] but name it zzzz-something.network [09:17] such that if and when nplan/real networking config is generate it takes priority [09:18] as _only_ the first matching .network file is in effect [09:18] (alphanumerical ordering as per systemd(tm) algorithm - patent pending) [09:19] that seemed like it could work. [09:28] smoser, rharper - cause when maas recovery / emergency was mentioned.... was that far enough in the boot, that e.g. networkd was actually available? [15:29] is there a bug that i cant add secondary dns to interfaces? im using dns-nameservers 8.8.8.8 8.8.4.4 and when i restart network or ivn reboot in resolv.conf only shows first dns? [15:29] ubuntu 16.04 server [15:32] saban: where are you adding that option? [15:32] ./etc/network/interfaces [15:37] you've definitely got an 's' on the end of dns-nameservers? no dns-nameserver ? [15:40] yes dns-nameservers 127.0.0.1 10.0.0.230 [15:42] i tride ivn this add nameserver 127.0.0.1 nameserver 8.8.8.8 in file /etc/resolvconf/resolv.conf.d/base [15:43] and secondary gets ignored [15:43] saban: as I recall, ifupdown has scripts hooked by resolvconf via /etc/resolvconf/update.d/ or related directories which handle it [17:02] Hello, Is ubuntu can be used as L1 (First level of VM) in a nested virtualization setup ? I'm having hard time to enable it to spin up another KVM instance ("Hardware not found") [17:05] run a vm inside a vm guest? [17:07] adylas: should work. What do you get when running "egrep -c '(vmx|svm)' /proc/cpuinfo" inside the VM? [17:31] sdeziel, Hi, nothing. The L1 (Ubuntu 16.04) is not unaware. [17:32] *is _not_ aware [17:33] adylas: if your host has an intel CPU, make sure it has the nested param enable (it does by default) [17:33] adylas: also check that your VM's CPU allow passing the required flags (vmx or svm) [17:34] adylas: "" should do it [17:34] Yeah, its ovirt, its supported in there and my hypervisor has kvm-intel.nested=1 [17:34] sdeziel, Is this in the config of the L1 guest ? Or the L2 guest ? [17:36] adylas: the nested param needs to be enabled in the host. The first layer of virtualization (L1) needs the right CPU flag. [17:36] sdeziel, interesting [17:36] adylas: the L2 shouldn't need anything special AFAIK [18:05] sdeziel, The hook did the trick. This one : https://github.com/oVirt/vdsm/blob/master/vdsm_hooks/nestedvt/before_vm_start.py [18:05] Thanks for your help !! [18:06] adylas: great