[11:00] <pkunk> How do I configure netplan to work in "server mode". i.e interfaces are expected to be Always up if they are detected
[11:01] <pkunk> I would like to disable the link detection so that any configured interfaces are always assigned IP's. This is a server so I know how to handle the routing etc.
[11:01] <pkunk> Right now if I pull the network cable in my ubuntu server, all the IP's for that interface get removed
[12:14] <blackflow> pkunk: netplan is using the systemd-networkd backend by default and you can of course force specific interface name, match-by, and almost every other param, via netplan
[12:14] <blackflow> pkunk: examples: https://netplan.io/examples
[12:15] <blackflow> I wouldn't know what happens to ifaces when you unplug the cable, when networkd is in charge, but if they disappear, it's networkd doing it, not netplan. netplan is just configuration abstraction tool.
[12:42] <pkunk> blackflow: Thanks, so if I change the renderer to ifupdown then I can skip the "feature" of systemd ?
[12:43] <blackflow> pkunk: no. netplan only supports networkd or NetworkManager backends. if you want ifupdown, you need to install the pakcage, configure it, and remove any files under /etc/netplan/ so it doesn't mess up with your ifupdown stanzas
[12:44] <blackflow> I'm curious, though, what are you trying to achieve tho? netplan (And networkd) definitely do support static network config.
[12:44] <pkunk> Thanks, I'm passing the ball to #systemd now
[12:45] <blackflow> I mean when the cable is unplugged, there's no network over those ifaces, so what's the matter if the ifaces disappear?
[13:53] <tnewman> is there a way to install ubuntu-server without an internet connection?
[13:54] <tnewman> at least without an internet connection during install
[13:54] <tomreyn> tnewman: the alternative installer will do it. maybe also the 18.04.2 standard installer.
[13:55] <tnewman> hmmmmm
[13:57] <tnewman> could i get a link to the alternative installer?
[13:57] <tomreyn> looks like the 18.04.2 installer still has some issues with it if you a NIC with no cable connected.
[13:57] <tnewman> http://cdimage.ubuntu.com/releases/18.04.2/release/ubuntu-18.04.2-server-amd64.iso this one?
[13:57] <tomreyn> ubuntu.com/download
[13:58] <tomreyn> -> server -> see our alternative downloads
[13:58] <tomreyn> -> Alternative Ubuntu Server installer
[13:58] <tnewman> link me to an iso
[13:58] <tnewman> i'm looking at that webpage
[13:59] <tomreyn> http://cdimage.ubuntu.com/releases/18.04.2/release/ubuntu-18.04.2-server-amd64.iso
[13:59] <tnewman> and it just looks like a linkn to regular 18.04 server iso
[13:59] <tnewman> right
[13:59] <tnewman> is that just a different wizard on the normal server install iso?
[13:59] <tomreyn> yes thats the alternative installer
[13:59] <tomreyn> also known as debian-installer
[14:00] <tnewman> gotcha, i'll give that a whirl <3
[14:00] <tnewman> thanks for the help :)
[14:00] <tomreyn> you're welcome.
[14:43] <tomreyn> tnewman: fwiw, i just verified that you can not yet install 18.04.2 using the standard server installer if a NIC was detected but no cable connected / no pyhsical link detected.
[14:43] <tomreyn> so indeed you'd need to keep using the alternative installe rin this scenario
[14:44] <tnewman> thats a little poopy tomreyn
[14:45] <tomreyn> i'm sure we can all agree on this.
[14:46] <tnewman> wonder when internet connectivity started being a requirement for ubuntu server like that :<
[14:47] <rbasak> Soon
[14:47] <rbasak> It's bug 1750819
[14:47] <rbasak> I believe (but haven't verified) that this is fixed in the upcoming 19.04 release next week.
[14:47] <rbasak> The next point release for 18.04 should have it.
[14:51] <tomreyn> oh right there was this workaround, i had forgotten
[14:51] <tomreyn> comment #10
[14:52] <tomreyn> but then the boot process is also delayed because snapd can't reach the internet
[15:09] <tnewman> farts
[15:18] <pkunk> Is there any way for make netplan emit "ConfigureWithoutCarrier=yes" for a particular connection even though it isn't of type ND_VIRTUAL ?
[17:22] <Delerium> Good afternoon guys - hoping I have an easy one... I've just upgraded my home ubuntu server to Bionic Beaver but since the upgrade all programs will not accept incoming connections (apache, mysql.. even teamviewer). I have disabled ufw and this has no impact. I can actually ssh into the box without an issue. I've amended the application config files to have the correct ip address in as
[17:22] <Delerium> the upgrade reset them but still no luck. Really pulling my hair out at this stage - anyone have any pointers?
[17:25] <tomreyn> Delerium: can you locally connect to those tcp ports?
[17:26] <tomreyn> does     iptables -L    show remaining rules?
[17:27] <tomreyn> Delerium: and please don't cross post.
[17:28] <Delerium> tomreyn: oh sorry i was just following advice from the other guy to post in here
[17:29] <Delerium> tomreyn: i can locally connect to the apache and mysql server no problem at all
[17:29] <tomreyn> ok, i noticed you also asked and kept discussing it there, but notice now it's also the other volunteer who didn't let you go. ;)
[17:32] <Delerium> tomreyn: <tomreyn> does iptables -L show remaining rules? - Unsure what you mean here?
[17:34] <tomreyn> Delerium: 'iptables -L' lists all iptables firewall rules in all chains.
[17:35] <tomreyn> you said you disabled ufw. this probably means that most rules are gone, but some may still remain. either way it's best to check with iptables directly than with the ufw 'frontend'
[17:35] <tomreyn> just to rule out that disabling ufw did not work, or not entirely
[17:36] <Delerium> tomreyn: the iptables command comes up with a lot of entries - is it possible to reset it all?
[17:37] <tomreyn> you can flush all rules using -F. this *may* impact your existing ssh session.
[17:38] <tomreyn> so don't do it unless you have a way to physically access the system or have some form of out of band access
[17:38] <Delerium> its a local machine in my house :)
[17:39] <tomreyn> you said home server, you did not say where you are now, so i had to bring it up. ;)
[17:40] <Delerium> tomreyn: as if by magic everything is working again - thank you so much buddy - i've litterally spent about 5-6 hours today trying to fix this machine
[17:40] <tomreyn> interesting :-/
[17:41] <Delerium> all i can think of is some sort of corruption during the upgrade process
[17:41] <tomreyn> you should probably undo the fixed configuration son those daemon configuration files, though.
[17:41] <tomreyn> also have a look at the output of    ubuntu-support-status --show-unsupported
[17:41] <tomreyn> this can be useful especially after a release upgrade.
[17:42] <Delerium> thanks :) Ill take a look
[17:43] <tomreyn> ...and may hint on packages which are giving apt's package depednency resolver a difficult time.
[18:04] <Delerium> tomreyn: this is interesting....and frustrating.... after a short while after i've flushed those iptables the issue comes back
[18:06] <tomreyn> Delerium: a short while, in which you did nothing at all? or a short while in which you rebooted it, reconnected the ethernet wire, restarted or reconfigured networking?
[18:06] <tomreyn> i recommend having a look at the release notes regarding networking if you haven't done so, yet.
[18:07] <Delerium> the only thing i configured was the ufw to accept only local network connections and it was fine then.... pooof gone
[18:08] <Delerium> uf
[18:08] <Delerium> actually scratch that - in my latest test i did NOT even config/activate the ufw so literally did nothing
[18:09] <tomreyn> have a look at the systemd journal, see if you can correlate the time
[18:30] <Delerium> seems stable now
[18:31] <Delerium> problem 2 - looks like the update corrupted one of my mounts - the mount it corrupted is one that has a space in one of the directory names which the previous ubuntu version allowed by putting 040 as the spaced character. Any ideas whats changed?
[18:31] <Delerium> error is no such file or directory now
[18:44] <tomreyn> Delerium: maybe the locale changed from non-utf-8 to utf-8? but this should actually have happened before 16.04
[18:45] <tomreyn> use single quotes around the full absolute path and tab completion to have the shell help you enter the path in an acceptable way.
[18:46] <Delerium> tomreyn: "use single quotes around the full absolute path and tab completion to have the shell help you enter the path in an acceptable way." <--- Can you explain what you mean here - i have the entry in the fstab
[18:48] <tomreyn> type: ls /path/to/directory/i/can/still/enter/fine/
[18:48] <tomreyn> the directory name you cannot enter properly is below "fine/" in this example
[18:48] <tomreyn> then juyt go to the end of the line and double tap the TAB key.
[18:49] <tomreyn> this should print contained / sub directories
[18:49] <tomreyn> now type the first character of the otherwise unspellable subdirectory (the one with the special character in it)
[18:50] <tomreyn> then double-TAB again. add more single characters until it fully completes the name
[18:51] <tomreyn> this may actually be optional then:  once it fully completed the name, put a single quote to the beginning and end of the full path on this command line.
[18:51] <tomreyn> and press enter, you shoould now have a way to enter the full path.
[18:52] <Delerium> hmmm ok i think i tried this the first time round and fstab didnt like it
[18:52] <tomreyn> there is also    ls --escape
[18:54] <Delerium> ok so the method you have advised to create the path just added a \ before each space in the directory. Navigating to the directory that method works a treat but when trying to put his into the fstab for a mount command it gives a parse error and ignores the line :(
[18:55] <tomreyn> you need to escape blank spaces by backslashes on fstab
[18:55] <tomreyn> it may also be possible to enter those paths in quotes there, i never tried
[18:56] <Delerium> so as an example entry i have //192.168.0.13/test account/status
[18:57] <Delerium> ive tried //192.168.0.13/test\ account/status but its not worked - it worked prior to the upgrade with //192.168.0.13/test040account/status
[18:57] <tomreyn> so that's a network mount then? cifs?
[18:57] <Delerium> fstab comes back with a parse error if you try quotes
[18:57] <Delerium> yes its a cifs mount
[18:57] <tomreyn> 040 sounds like @
[18:58] <tomreyn> that'd be hex
[18:58] <tomreyn> since cifs mounts are specified as URI's, you'd probably need ot use URL escaping
[18:59] <tomreyn> so %40
[18:59] <Delerium> no dice
[18:59] <tomreyn> \@ might also work, or just @ (but i assume not)
[19:01] <Delerium> :( nada
[19:01] <tomreyn> "smbclient -L ip_of_net_interface -U your_user_name" should list available services
[19:02] <tomreyn> the unspellable thing seems to be a 'service' in the samba nomenclature.
[19:05] <Delerium> hmmm im tempted to rename the directory and then go through all of the scripts underneath to ensure ive not broken any - seems to be the path of least resistance at this point haha
[19:07] <tomreyn> oh i think you can specify a codepage as a mount option, too
[19:08] <Mead> so this has been bugging me since I installed ubuntu server in my lab, my startup hangs at this https://paste.ubuntu.com/p/VgGFZHsq2r/  how can I get it to NOT hang for a few minutes?
[19:08] <tomreyn> Delerium: here's the man page, in case you have some time ;-) http://manpages.ubuntu.com/manpages/bionic/en/man8/mount.cifs.8.html
[19:08] <Delerium> tomreyn: ill give it a whirl - thanks for all your help tom
[19:08] <cryptodan_mobile> Mead is it setup via dhcp
[19:09] <tomreyn> you're welcome, good luck.
[19:10] <Mead> cryptodan_mobile: you mean it is hanging to get a DHCP lease?
[19:10] <cryptodan_mobile> Mead it is hanging because it cant configure networking
[19:17] <Mead> strange, cause networking works and configures via netplan config
[19:17] <tomreyn> Delerium: more text here, with some good hints https://www.kernel.org/doc/Documentation/filesystems/cifs/README
[19:18] <Delerium> ta
[19:25] <cryptodan_mobile> Mead but is there a net plan config file
[19:30] <Mead> cryptodan_mobile: yes, and I and I have even added lines to it.
[19:30] <Mead> err - one of those "and I"
[19:40] <cryptodan_mobile> When it was created did sudo netplan apply to see if it worked
[19:41] <Mead> yes, infact I just reapplied it and rebooted my server to make sure,
[19:41] <cryptodan_mobile> Is it set to render via networkd
[19:43] <Mead> yes
[19:43] <cryptodan_mobile> Odd
[19:45] <Mead> https://paste.ubuntu.com/p/W6XPTg63Qf/  this is my /etc/netplan/*.yaml config
[19:47] <tomreyn> what does     systemctl status systemd-networkd-wait-online.service    say about it?
[19:48] <cryptodan_mobile> Mead https://termbin.com/0vga
[19:50] <Mead> https://paste.ubuntu.com/p/JwYm8QTXh5/
[19:51] <Mead> cryptodan_mobile: what is up with that config?
[19:51] <tomreyn> hmm so maybe enp3s6 is failing to get a dhcp lease
[19:52] <tomreyn> if you look at those  "systemd-networkd-wait-online[696]: managing: enp3s6" messages on your journal, does it provide more contextual information?
[19:52] <Mead> well, nothing is plugged into it yet
[19:53] <cryptodan_mobile> Remove it then
[19:55] <tomreyn> https://askubuntu.com/questions/1046420/why-is-netplan-networkd-not-bringing-up-a-static-ethernet-interface#answers
[19:55] <tomreyn> ...may be related (i know, not static)
[19:56] <Mead> nope I was wrong, it is enp3s5 with nothing plugged into it, enp3s6 is plugged into the gateway and has a DHCP lease right now
[19:58] <cryptodan_mobile> Is the gateway also serving dhcp6
[19:58] <Mead> yeah, so the port with the error message in that pastebin is functioning
[20:00] <Mead> cryptodan_mobile: yes, I've got ipv6 service
[20:02] <Mead> hurm, strange.  I know my netplan yaml doesn't have the dhcp6: true line, but the "ip a" output shows an ipv6 address and gateway
[20:12] <Mead> welp adding the dhcp6: no to the interface didn't stop this.  Funny thing is that it was having this issue before I messed with the netplan yaml file and I was using whatever the default install configuration is
[20:17] <Mead> now this is even more strange,  after reboot and including the dhcp6: the "ip a" output still shows ipv6 address assigned
[20:19] <cryptodan_mobile> Mead might want to see if your cables are in the right ports
[20:20] <Mead> they are
[20:20] <cryptodan_mobile> Then I'm out of ideas
[20:21] <Mead> should I go ask the fine folks in ##linux or #netplan for the best results?
[20:22] <cryptodan_mobile> Netplan
[20:22] <Mead> thanks
[20:35] <tomreyn> i suggest you review the systemd-networkd configuration netplan wrote for you, too. just in case.
[22:26] <Mead> tomreyn:  you mean the ones in /run/systemd/network?
[22:35] <rockyfelle> Hey, im running a cloud service sync (megacmd) with a symlink there from www/html so apache2 can access it, ive given the megacmd folder the proper permissions so apache2 can use them, but every time a file is updated on behalf on megacmd the permissions are reset which makes apache2 unable use the files
[23:23] <tomreyn> Mead: yes
[23:26] <Mead> It looked fine to me, there was no differnece between enp3s5 and enp3s6 ports
[23:27] <tomreyn> ok
[23:27] <tomreyn> rockyfelle: either change the umask "megacmd" (i do not actually know what this is and what impact this change might have) operates with, or join www-data to megacmds' primary group (the previous note applies here, too).