[10:47] Good afternoon. [10:49] G'Day folks [11:24] Hi again [17:42] has anyone seen network interfaces named `ens##` in virtual environments instead of `eth##` for Wily? [17:42] (whereas installer disks see `eth##`) [17:43] ens? Not enp? [17:44] mhm [17:45] also, installer set up the interfaces file with eth0, eth1, so networking didn't come up (this is the server iso but still) [17:45] (so I had to write /etc/udev/rules.d/70-persistent-net.rules to assign the names) [17:46] lordievader: it threw me for a loop too [17:46] and there's a discrepancy between the installer ISO and what's actually installed [17:46] so IDK where to report such issues [17:46] or what to file against [17:46] and it wasn't part of the ISO testing, i needed the VM set up for package change verify-it-works ness :P [17:47] Hmm, well we should be moving to bios names sometime soon. But ens is new for me, perhaps because it is virtual, but nic's are named as enp with me. [17:48] lordievader: that still doesn't explain why /etc/network/interfaces was populated with eth#, nor why the installer ISO read the interfaces as eth# [17:48] True, true. [17:48] and that shows the bigger problem of if the ISO doesn't correctly configure the /etc/network/interfaces on install, you get a nuked network routing table and connectivity [17:49] the naming is done by systemd now, isn't it? [17:49] and it's a damn good thing i keep a 70-persistent-net.rules file template lying around [17:49] TJ-: if it is, does that explain the discrepancy between installer naming and system naming? [17:49] and does that explain the broken /etc/network/interfaces upon install? [17:49] Still udev right? Though udev and systemd share a lot. [17:50] http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames/ [17:50] Hmm, perhaps it is systemd.. [17:50] lordievader: i had to put a udev override [17:51] but there's still that ISO/installed discrepancy [17:51] that will make people be "WTH Y DOESNT IT WORK" [17:51] (and this was the daily iso from yesterday) [17:51] "Names incorporating Firmware/BIOS provided PCI Express hotplug slot index numbers (example: ens1)" <-- explains the ens## [17:51] does NOT explain why only 2 cards got ens33 and ens35 [17:52] (at least from the guest os's perspective) [17:52] I am much more concerned about the ISO seeing the kernel style naming versus the installed version using systemd and the predictable method in that link [17:54] Ah, thanks for clearing that up :) [17:55] what init system does the installer use? [17:59] can't tell i can only get a busybox prompt o.O [17:59] unpack the squashfs image, if that's what it is [18:00] No ps in busybox? [18:03] /bin/busybox init <-- that [18:03] Sounds like sysv/upstart. [18:04] so what do i file the bug agains [18:04] because this is a HUGE issue [18:04] teward: I'm pulling in the ISO to try in a VM [18:05] TJ-: i have two interfaces attached, one NAT, one host only [18:05] i use VMware [18:05] VBox might be different [18:06] I use neither :) [18:06] KVM? [18:06] but of course :) [18:07] I have several servers here with multiple interfaces so I can test it on those [18:07] Whoop whoop [18:19] teward: is it booting UEFI or BIOS? [18:25] TJ-: I think BIOS, within VMware [18:25] any way to find out from the shell? [18:32] "ls /sys/firmware/efi/vars/" [18:33] no such dir [18:33] no efi dir either [18:33] so probably BIOS [18:33] OK, so BIOS [18:40] I've found another bug... missing kernel package! [18:44] o.o [18:52] TJ-: did you try a BIOS boot yet, and what did you find? I see the +UEFI boot issue in -devel you just said [18:59] I'm about to try it... but it wont make any difference. The detection happens from within the /target/ chroot and the device names will already have been set