[00:03] <cengique> Sarnold We got it by running microstack v 218 there is now a working control/compute node setup
[00:05] <sarnold> cengique: oh sweet :D now you get to work on whatever it was you were interested in doing in the first place, if you can still remember it :)
[00:22] <tds> TJ-: do you have v6 forwarding enabled?
[00:25] <TJ-> tds: yeah --- this is the weird part; all the usual bits are there, but the host doesn't do anything with the neighbour discovery broadcasts
[00:27] <TJ-> tds: just to be sure I'm not going loopey - proxy_ndp =1 on the 'upstream' interface (in my case wlp4s0) is the correct i/d to enable it on?
[00:27] <TJ-> this is the interface receiving the NS broadcasts
[00:27] <tds> yes
[00:27] <tds> you'll need to make sure you have the weird forwarding sysctl enabled there as well
[00:28] <TJ-> tds: yes: /proc/sys/net/ipv6/conf/wlp4s0/forwarding:1
[00:28] <TJ-> tds: could this need mc_forwarding too?
[00:28] <tds> i don't think so
[00:29] <tds> you're not throwing away the NSes with nft/ip6tables or similar?
[00:29] <TJ-> tds: no - nothing in ip6tables except 'ACCEPT' inc. policies on the chains
[00:33] <TJ-> tds: in case it gives a clue - I noticed the entry in the proxy table is removed by something a while after I add it (maybe 5-10 minutes) - maybe that is normal
[00:40] <tds> TJ-: might be interesting to try testing inside the machine, or something like that
[00:40] <tds> eg bring up a veth pair between that and another netns, and try proxy ndp over that
[00:40] <TJ-> tds: hmmm, good idea
[00:41] <tds> (fwiw, testing on a 20.04 box here, proxy ndp works ok between two netnses)
[01:07] <TJ-> tds: hmmm, maybe I'm setting it up incorrectly. Just tried creating 'test1' adding global IPv6 addresses in a /64 from our prefix. Can ping in and out host<>netns (as with the libvirt guest) but remote (gateway router) NS's not getting a reply from my laptop
[01:08] <tds> so what if you set the host to proxy ndp an extra address in that /64, does test1 get replies to NSes?
[01:09] <TJ-> oh hang on, I forgot on this one to add a specific proxy-ed address
[01:13] <TJ-> tds: ahhh, got it for the netns. Had to add the proxy address and recall to re-add the host's default route to the gateway
[01:16] <TJ-> tds: and now it works!!
[01:17] <TJ-> tds: I think I was being messed about by doing this with NetworkManager managing the wlp4s0 interface and removing settings behind my back, such as the ipv6 default route
[05:37] <benl90> Hello, I need some help. I use netplan on 18.04.5 and for about 4 month no problem with resolving the DNS, but suddenly yesterday the same config won't run and nslookup won't work, cause massive queue of email, for now I only can solve domain via overwriting /etc/resolve.conf which always change everytime, I never encounter this and this's really strange. any help is appriciated. Thanks
[06:24] <lordievader> Good morning
[06:25] <lordievader> benl90: What issues are you seeing exactly?
[06:25] <lordievader> Timeouts, actual errors, something else?
[11:30] <benl90> lordievader, nah It's timeout. But somewhat strange using systemd/resolve.conf it works,using netplan it don't
[11:31] <lordievader> Is netplan writing different resolvers to /etc/resolv.conf?
[11:41] <mgedmin> netplan is not touching /etc/resolv.conf; netplan generates configuration files either for NetworkManager or for systemd-networkd
[13:35] <benl90> mgedmin, nah That's what I know, but when I sudo systemd-resolve --status it doesn't have any DNS record. It's strange. But if I put "DNS=1.1.1.1 8.8.8.8" at /etc/systemd/resolve.conf it works.
[14:17] <lordievader> benl90: What are the contents of `/etc/resolv.conf` when you get the timeouts?
[14:30] <robertparkerx> how can I install webp ?
[15:09] <robertparkerx> nevermind
[15:09] <robertparkerx> apt-get install webp
[15:09] <robertparkerx> ty
[22:11] <BC64> Is this an ok place to ask questions about microk8s, MaaS, Juju, etc?
[22:13] <powersj> BC64, there are separate channels for maas and juju, but you might get what you are looking for here
[22:17] <BC64> I'm trying to decide between microk8s and k3s for use in a small production cluster. I want to deploy Rancher on this cluster to deploy Kubernetes on a separate OpenStack cluster deployed using MaaS and Juju.
[22:18] <BC64> As I typed that, I'm trying to figure out why I wouldn't just use Juju to deploy the Kubernetes cluster on OpenStack...
[22:19] <BC64> I really like Rancher's GUI for the kube-illiterate