=== nOgAnOo_ is now known as noganoo [05:11] Anyone have a good document for nameservers with bind? I'm having trouble getting my nameservers set up and resolving... and I'm not sure now what is wrong at this point. [06:01] Good morning [06:01] xase: Is bind a must? Else I'd recommend using PowerDNS, much easier to set up. [06:06] good morning lordievader [06:06] hope your week started well [06:06] Hey cpaelzer [06:06] It did, busy but good. [06:06] How about you? [06:28] lordievader: about the same :-) [06:32] Any plans for today? [06:34] The days I could have ONE plane for a day are long over :-) [06:34] but for like "now" ppc64el build errors between stdbool/altivec headers [06:35] well I also never had a plane, but I meant "plan" obviously [06:37] 😉 === michal_f is now known as michal_f_2 [09:51] fejwf [09:51] hey all, I had a webapp crash... I logged in and I was getting "out of space errors" I ran a df -h and got back /dev/xvda1 299G 287G 0 100% / where did those 12gb go?! [09:51] hello all [09:52] I went to use a service I run for myself tonight and it wouldn't load, I ran a df -h and got back /dev/xvda1 299G 287G 0 100% / [09:52] where sis those 12gb go?! [09:52] did* [09:53] ext filesystems reserve a bit of space for emergency logins. [09:53] This can be configured and was per default 5 percent IIRC. [09:55] The tool `dumpe2fs` can show these configurations. [09:59] that seems... extreme for so much disk space... especially since it doesn't reserve the space for non-root purposes [10:01] sudo tune2fs -m 2 /dev/sdb1 would set it to a reasonable 6gb for a 300gb disk, yes? [10:03] yeah it's a default from a different era really [10:03] in fact i thought it had changed but maybe not... [10:03] Apeparently not. >.> [10:04] I was not pleased to discover that 300gb had 12gb reserved XD [10:07] in the era multi-terrabyte disks... 5% of a disk is an unreasonably large percentage to reserve >.> [10:08] 1-2% is reasonable.. but still a significant portion to set asie [10:08] aside* [10:58] cpaelzer: hi, how would I best come up with a url that identifies that my patch was merged upstream, given https://github.com/apache/libcloud/commits/trunk shows that my change was rebased on top of master, instead of merged? [10:59] the PR link which is closed and says "fix committed"? That's equal to "Forwarded:" [10:59] (context: dep3 headers) [10:59] I could construct a url in github that shows the diff on master between two commits perhaps [10:59] or just not add the applied-upstream header bit [11:01] hi ahasenack [11:02] I think "yes" would suffice for this case, but the dep3 header template says it must be either a url, version, or commit [11:03] ahasenack: do I understand it correctly that you'D want one (=1) link that you can use as "origin:" for all the commits you had here? [11:03] cpaelzer: not origin, but "Applied-Upstream" [11:03] ah ok [11:03] I tihnk "yes" will do in this case [11:04] you can add in the description the link you just shared [11:17] ahasenack: to some extend (but that is opinion) we never should be overzealous on the last bits of interpreting these tags - after all they are meant to make things understandable and to be able to follow the path of a change [11:17] sure [11:17] IMHO if the metadata you add will allow one to find the right spots, then it doesn't matter too much if you write yes/yez/y or similar [11:17] I know tools will have to rely on a certain format [11:18] but just saying, I'd want not to spend "too much" on trying to be too accurate [11:22] just checking if there was precedence [11:26] cpaelzer: ok, I made libcloud green (https://bileto.ubuntu.com/excuses/3384/cosmic.html) [11:26] including a few "always failed" cases [11:27] I know why the i386/armhf ones are failing, but I don't know why they are *not* failing in the other arches [11:27] it's caused by a change in python-requests: https://github.com/requests/requests/issues/3477 [11:27] an old-ish one even [11:27] I checked versions and i386 and amd64 have the same ones [11:28] I'll investigate a bit more, but we are already in a better place [11:29] thank you ahasenack [11:29] for not giving up! [11:31] and the ppc64 build issue fixed (again) [11:31] ready for lunch ... [12:15] cpaelzer: https://code.launchpad.net/~ahasenack/ubuntu/+source/libcloud/+git/libcloud/+merge/353856 [12:15] I'm counting on that one unblocking paramiko and strongswan from cosmic-proposed [12:15] I might have to ask for some sort of in-step migration, though [12:39] ahasenack: checking the MP now ... [12:49] ahasenack: #placeholder [12:49] did I miss something why this is needed in your MP? [12:49] yo don't like it? :) [12:49] I might like it once I get what it is about [12:50] git-ubuntu build is failing miserably in packages that do not have debian/patches [12:50] so I'm trying to get used to it [12:50] I've seen "#placeholder" for d/p/series in some packages that do not have any patches [12:50] didn't consciously see that before, but I see why it is there now [12:50] gmm [12:50] hmm [12:50] so I started by adding it, when there was no patch [12:50] and just added patches on top [12:50] but since you now have patches it looks odd from an empty file to [12:50] but of course it's not needed [12:51] #plaeholder + 2 changes [12:51] ok, I can drop it [12:51] it's really a g-u bug [12:51] but will probably be a long while until it's addressed [12:51] yes [12:51] can the file be just empty? [12:52] that I don't know [12:52] haven't seen it [12:52] just like you don't use g-u submit I gave up on g-u build-source [12:52] hehe [12:52] I could not go on day-by-day without the review/backport features [12:53] but it seems the non core features are unrelibale (as we know) [12:53] ahasenack: once I push the ta after you dropped the placeholder - do you need sponsoring on this? [12:54] I see you have a PR for the i386/arm tests [12:54] next question: do you want to wait and bundle those once accepted? [12:55] or get it fixed soon to unblock migrations? [12:59] morning! [13:00] good morning! === Sven_vB_ is now known as Sven_vB [13:19] hi sdeziel and kstenerud [13:19] ahasenack: I see you dropped the #placeholder [13:19] ahasenack: what about my question on the i386 test fix [13:19] do you want to do all that in one upload? [13:19] or get things up to unblock now [13:20] and if you want things now, do you need more than tags being pushed? [13:20] since it is not imported you don't even need that actually [13:23] lordievader, I followed a server setup setup for ISPConfig... Will powerDNS drop in as a replacement? [13:24] Err, no idea. However, PowerDNS is a lot easier to set up than bind. [13:25] harr - we are more upstream than upstream on the openvpn service file and that breaks us [13:25] It sounds easier. With a name like powerDNS [13:26] andol: kstenerud: their .deb has older .service than their actual git repo - it is a trap if the former works and the latter which we follow doesn't ? [13:38] ahasenack: Do I need to use the web interface for this ipa testing env? What else needs to be set up? [13:38] there is a command-line tool, but I used the web interface in the past [13:38] I think it demands kerberos authentication now. So you would have to install krb5-user on your machine, set it to the example.com realm, and tell it that the kdc is the vm where you installed freeipa [13:39] then run "kinit administrator" and get a ticket for the admin user (or kinit admin, I don't remember the exact name) [13:40] OK, I typed that and it didn't complain. Where would the ticket be? [13:41] run klist [13:42] OK. what does the ticket do? [13:42] it's an authentication token, valid for X hours. It can allow you access to other kerberized services without having to type in a password again [13:42] kstenerud: before going further, [13:42] kstenerud: this setup used your bind9 package with that linker patch, right? [13:43] I have the ppa version installed [13:43] kstenerud: ok, so I think you can propose that fix for cosmic [13:43] since you just validated it yourself :) [13:43] by installing it? [13:43] by completing the freeipa installation [13:44] oh right :) [13:44] you can further check /var/log/syslog to see if there are bind9 crashes in there [13:44] named-pkcs11 iirc is the binary name [13:45] yeah no crashes [13:48] so, MP it away [13:48] we can still upload fixes [13:48] but after a while, even those need an exception [13:49] keep an eye on the calendar: https://wiki.ubuntu.com/CosmicCuttlefish/ReleaseSchedule === xase_ is now known as Guest42923 === Guest42923 is now known as Xase [16:32] ahasenack: I'm getting weird behavior with uvt-simplestreams-libvirt. When I sync a release, the command returns with no error, but the image doesn't actually download [16:48] kstenerud: does it delete the images you already had in the query command? [16:48] kstenerud: you have been having weird behavior with that tool since the beginning, something is going on [16:48] maybe purge everything and start over (libvirt, uvt) [16:52] cpaelzer: after the libcloud upload, paramiko and strongswan migrated right away \o/ [16:52] and look at that green wave: http://autopkgtest.ubuntu.com/packages/libcloud [16:52] it's just starting [17:52] Any recommended cluster (distributed) file servers on ubuntu? [18:06] "Windows, i.e., NetBIOS, name resolution is on Linux provided for by the "winbindd" daemon" uhh? [18:07] "One does not need either of the other two Samba daemons "smbd" and "nmbd" when NetBIOS name resolution is all you need" [18:07] * ahasenack wonders what he things nmbd does [18:20] Hello, Would anyone here be able to answer a question i have about ceph? [18:22] best just to ask the question, RandomTech :) [18:27] I was just wondering if it is still recomended to have my journal on a seperate ssd (I have 2 3TB HHD and 1 250GB SSD) and if so how should i calculate the space needed [18:27] For ceph of course [18:54] cpaelzer: Can you help me with a virtual bridge problem? === arka500_ is now known as arka500 [19:02] !ask | kstenerud [19:02] kstenerud: Please don't ask to ask a question, simply ask the question (all on ONE line and in the channel, so that others can read and follow it easily). If anyone knows the answer they will most likely reply. :-) See also !patience [19:39] kstenerud: virtual bridge problem with ? [19:40] ahasenack: libcloud -> paramiko -> strongswan and co - \o/ [19:41] yeah, that was nice [19:41] cpaelzer: It was the virtual bridge you'd helped me set up yesterday. For some reason now when I try to uvt-kvm create with the template, it fails: [19:41] and quick [19:41] uvt-kvm: error: libvirt: Cannot get interface MTU on 'br0': No such device [19:42] Something's changed on my system because it used to work until a few hours ago [19:42] and I can't for the life of me figure out what [19:42] hmm [19:42] one moment [19:43] virsh net-list shows br0 [19:43] active [19:43] does it exist with brctl show ? [19:44] and ifconfig br0 is the device there [19:44] or did lxd take it away for some reason? [19:44] hmm nope none of that [19:44] kstenerud: so it doesn't exist with brctl show ? [19:44] correct [19:44] arr [19:44] what does lxc profile edit default tell you [19:44] what is the bridge [19:45] to be used in lxd [19:45] ? [19:45] it shows eth0 bridged to br0 [19:45] name: eth0 [19:45] hmm, maybe it is rather dynamic [19:45] nictype: bridged [19:45] parent: br0 [19:45] are any lxd containers up that would use the bridge? [19:46] if not it might have removed the bridge [19:46] nope. no containers [19:46] yeah , there you go [19:46] we created a frankenbridge intentionally [19:46] you need to have once [19:46] one container most likely to ahve it existing [19:46] sorry for the typos [19:47] check my telegram pic for the reason [19:47] :-) [19:47] lol [19:47] OK so basically as soon as I do the net-define thing I need to create some minimal container to use it? [19:48] yeah, and never remove the last container or the bridge will go away [19:48] you could follow the netplan.io example I linked [19:48] to create a fully separate bridge [19:48] you'd need to make sure there is some dhcp server on it external or no [19:48] not [19:48] but then the lifecycle of that bridge would neither depend on libvirt nor lxd [19:49] kstenerud: I'd leave for today, is that enough to get it working? [19:49] so if I did a full on netplan bridge, that would do it? [19:50] I think so [19:50] yeah that bridge would exist all the time [19:50] just as I said, someone should provide dhcp on it [19:50] but that might as well be your router or whatever is on the bridged ethdev# [19:52] cpaelzer: This is all going to be local to the VM I'm running this whole shebang on [19:53] I'm doing a cloud-side version of this tester setup [19:53] I just use the libvirt-created bridge [19:53] virbr0 [19:53] that's where I 'hang' my vms [19:53] fine for me if that is what you want [19:53] but can lxd and vms talk to each other on virbr0? [19:53] I hang containers in another bridge, the one created by lxc init [19:54] they get another network [19:54] <_KaszpiR_> should be doable [19:54] but can talk to each other just fine [19:54] just not dns, usually [19:54] the host routes [19:54] ahasenack: he had it on the one by lxc init [19:54] dns got messy after systemd-resolved [19:54] he jsut ran into that bridge being removed when the last container goes away [19:54] I had it working so nicely in xenial, with the containers resolveable by name automatically [19:55] kstenerud: being on the same bridge with an IP of the same subnet will work [19:55] it really behaves like a switch most of the time :-) [20:02] cool :) [20:56] * jhebden is back from [afk] - 426524h:56m:12s away [21:05] 48 years? [21:08] think about it for a sec, it was a very magical time for *NIX [21:25] Ah, yes. [21:30] * RoyK guesses a raspberry pi with no rtc or ntp [21:30] is NFS installed by default on sever? [21:31] the kernel modules, yes, but you need nfs-common to use it [21:31] (iirc) [21:34] so here is my situation. i setup(partially, not fully configured to my liking yet) samba on ubuntu server to share a zfs mirror array to my network. i have mapped that share on win 10 as a drive and now im getting my ubuntu desktop setup. im not sure if i should use NFS or just use the the samba share? [21:34] should work if the smb share is mounted [21:35] just install nfs-common first and configure /etc/exports - exportfs -a and test it [21:36] jhebden: just wait until you see what's been done with unix, you'll /away again in a real hurry.. [21:36] sarnold: ? [21:37] what do you mean should work if the smb share is mounted? [21:37] RoyK: I suspect polkit could make anyone cry.. [21:37] sarnold: I'm just now reading about javascript and am digging a bunker brb [21:37] jhebden: lol [21:37] * RoyK installs windows 95 on sarnold's machine [21:37] madLyfe: ubuntu can mount smb/cifs shares just fine [21:38] RoyK: hehe, in 1995 that actually seemed pretty neat. :) [21:38] for a short while [21:38] ya im just asking if i should stick with the smb share i have setup or setup NFS as well? transition to only using NFS for nix and win both? [21:38] RoyK: I mean, I had plug-n-play with linux earlier, and 32 bit protected mode, and better screensavers, and web browser, and web servers, and so on. [21:38] RoyK: but as far as windows went it was neat ;) [21:38] on windoze it was plug-n-pray, remember? [21:40] madLyfe: I'd stick with samba if you have it working already.. unless you want to learn nfs [21:40] madLyfe: I don't know how well win10 supports nfs mounts [21:40] i will just do that for now. do they play well running together? [21:41] windows has supported nfs since nt 3.51 or something, but it has always sucked - don't use nfs with windows [21:42] is it possible to make the share act like a drive? and auto mounted? [21:43] !autofs [21:43] Automount is the modern way to mount directories over a network. It is much easier to manage and more economic in bandwidth than static mounts via fstab. For more info - https://help.ubuntu.com/community/Autofs [21:43] well do i need cifs-util? [21:43] to mount cifs, yes [21:44] yeah I tried NFS on windows. do not recommend. stick to smb/cifs stuff for windows. [21:45] but since this is a samaba server hosted by ubuntu server i dont need that? [21:45] madLyfe: this is needed client-side only [21:45] the autofs is only needed you mean, sdeziel? [21:46] madLyfe: no I meant cifs-utils [21:46] cifs-utils is the client side to mount cifs, samba is the server-side. autofs is the client thing to automount things [21:50] so the samba server is using cifs standard? [21:50] samba is using SMB3, also called CIFS [21:51] the worst name of a network filesystem ever - "common internet filesystem" [21:51] ok cuz i read that cifs was IMB or MS? implementation of the samaba protocol? [21:51] and SMB really sucks over a slow link [21:51] i have cat5 [21:52] so i need the cifs-utils and the autofs? [21:52] madLyfe: cat5 can be used for a lot of things, including strangling people - the question is the data rate of your network [21:52] (don't strangle people) [21:53] IIRC cat5 can be used for gigabit up to 50m or so, cat5e up to 100m, cat6 can do 10Gbps over 50m, cat6a 10Gbps over 100m [21:54] so what is the difference between just right clicking and adding the share to the file manager(nautilus or dolphin) and do the cifs-utils/auto mount? [22:05] *crickets* :P [22:07] the former does, i think, not automount. the formeruses udisks2 to make it happen. [22:08] did you mean the latter on the last part of your statement? [22:08] you may not get great results asking desktop tools questions in a server channel, hehe [22:08] true [22:08] I've never seen nautilus or dolphin, can't tell you what they do :) [22:09] * tomreyn was referring to the graphical file manager in both cases [22:10] oh ok. checks out. :P [22:11] i know the system has a GUI settings area to auto mount things like the internal drive. [22:12] so is that what i need the cidfs-utils for? to make the samba share 'look' like a drive? [22:22] so if im going to mount the network share, 'sudo mkdir /media/servershare' is their example, how do you make the dir a drive and not a dir on the OS drive? [22:23] i obv dont understand fully how that works but im just trying to formulate the question as best i can. [22:23] the only way to access files on linux is to mount the drive into the filesystem. [22:25] sarnold: i think you're confusing yourself about filesystems and disks [22:25] so like all my internal drives are already auto mounted and are listed under devices. i right clicked and added the network share under places called 'kinghat-server': https://usercontent.irccloud-cdn.com/file/1SqiE56E/image.png [22:25] nacc: unlikely. :D [22:25] my goal is to make the share show up like a disk under devices. [22:25] sarnold: bah, sorry, meant madLyfe! [22:25] nacc: I'm doing my best to respond to madLyfe's questions .. hehe. [22:26] and had already started typing your nick about something previously :) [22:26] aha! :D [22:26] * nacc goes back to other things [22:29] is there a proper place in root where i should be making the dir at? [22:30] lordievader, you're familiar with PDNS, did you get a failure on your first startup? and if so how did you fix it? [22:30] The error is very vague. [22:30] Xase: what's the error? [22:31] Process: 10232 ExecStart=/usr/sbin/pdns_server --guardian=no --daemon=no --disable-syslog --log-timestamp=no --write-pid=no (code=exited, status=1 [22:31] Oh something else [22:31] Unit start.service could not be found. [22:33] Xase: that feels likely to be a misuse of the systemctl command [22:34] what exactly did you type to get that error? [22:34] service pdns.service start [22:34] err service pdns start [22:34] not .service [22:34] try systemctl start pdns [22:36] Same... reading journalctl gave me a better error. [22:36] Hold while I pastebin it. [22:37] ooooo https://usercontent.irccloud-cdn.com/file/DnEyLZmU/image.png [22:37] https://pastebin.com/puLf9yK1 sarnold [22:37] I stopped bind before starting it... [22:38] Disable bind from starting and reboot server? [22:38] Xase: okay, something is already trying to do dns. what's binding that port? you can check with netstat -ulnp | grep :53 [22:39] madLyfe: what tool is that? [22:39] udp 21504 0 127.0.0.53:53 0.0.0.0:* 802/systemd-resolve [22:39] aha [22:39] you mean that is displaying that info? [22:40] madLyfe: yeah [22:40] dolphin file manager if thats what you mean: https://usercontent.irccloud-cdn.com/file/FWpFzG9C/image.png [22:41] Xase: try systemctl mask systemd-resolved.service [22:41] aha :) so that's dophin.. [22:42] Created symlink /etc/systemd/system/systemd-resolved.service → /dev/null. [22:42] Should I reboot my server or stop resolved? [22:43] also should I mask Bind9 so it doesn't interfere? I don't want to remove it, incase pdns doesn't work out. [22:43] Xase: that's prolbably a good idea [22:43] But that systemd-resolved might've been stopping bind from working too right? [22:44] I don't understand exactly when a disabled service would be re-enabled but I understand some standard action does that... but I don't think any automatic tool will undo the 'masking' operation [22:44] yes [22:45] alright this time, no errors starting pdns. [22:46] Sweet. On the right track. [22:47] sarnold, thanks :) [22:48] Xase: all sorted? woo [22:49] Seems to be... now I just need to figure out how to configure powerdns :D [22:59] Good lord -_- [23:03] well since adding the fstab entry my computer no longer shuts down. it shuts down and then turns right back on. lol [23:04] its a dual boot machine and windows shuts down fine [23:15] hmm its like it got confused and mixed suspend and shutdown.. [23:15] anyways, not server related. [23:15] thanks for the help guys! [23:57] I have another small issue after setting up PowerDNS via a tutorial for ubuntu. [23:57] When I dig @localhost. the status is SERVFAIL and it says (2 servers found) [23:58] can you pastebin the whole thinig? [23:59] Of course sarnold