[01:45] <dunnousernamefn> What are the *actual* memory requirements for Ubuntu Server on ARM? I see 3 different metrics on 2 different wiki pages. I want to install a drupal server with 256MB (the minimum requirement for drupal is 64MB) and I'm not sure what metric to look for
[01:47] <sarnold> I ran ubuntu on a pandaboard es for a while; that only had one gig. that was fine for a surprising amount of uses..
[01:48] <sarnold> 256 megs for an interpreted language runtime sounds tight to me
[01:48] <sarnold> but if drupal thinks they can pull it off...
[01:48] <dunnousernamefn> They said 64MBs
[01:48] <dunnousernamefn> But that doesn't consider the actual OS
[01:50] <dunnousernamefn> I see in different places on the wiki either 256MB, 384MB, or 512MB; which is kind of confusing
[01:51] <sarnold> if they said 64 megs for their own use, and you don't have unrealistic expectations, it sounds worth a try to me
[01:51] <sarnold> my little aws machine is only 512 megs and currently has >300 megs cached
[01:51] <sarnold> it's really only doing ssh, mosh, irssi, plus whatever it does by default
[01:52] <dunnousernamefn> Hmm, I found 31MB on https://help.ubuntu.com/16.04/installation-guide/arm64/ch02s05.html
[01:53] <sarnold> woah. I wonder how long it's been since someone tried that..
[01:53] <dunnousernamefn> It's 16.04 LTS...
[01:53] <dunnousernamefn> But that sounds pretty theoretical
[01:53] <dunnousernamefn> Maybe maintenance mode :P
[01:53] <sarnold> 31 megs sounds like kernel plus sshd plus busybox to me :)
[01:54] <dunnousernamefn> I could theoretically run it on a microcontroller given external RAM
[01:54] <dunnousernamefn> If you click the realistic figures link, it says 128MB without a desktop is minimum
[01:55] <dunnousernamefn> But I'm not sure if it is talking about Desktop or Server at this point
[01:55] <sarnold> they're close enough to identical if you don't actually *run* the desktop
[01:56] <dunnousernamefn> I had trouble with the Ubuntu Desktop installer... now I have Server on my laptop... it only breaks when I close the lid
[01:56] <dunnousernamefn> I only regret it a lot
[01:56]  * mwhudson looks at the arm64 in that url
[01:57] <sarnold> it's not like they're different operating systems, just default package selections; apt-get install network-manager unity   and you're 90% of the way to a tolerable desktop...
[01:57] <mwhudson> can you even get a system with an arm64 cpu and less that, say, 128 megs of ram?
[01:57] <sarnold> mwhudson: VMs?
[01:57] <mwhudson> sarnold: true
[01:57] <sarnold> granted .. the only arm64 VM provider I know seems to go for HUGE systems :)
[01:58] <dunnousernamefn> I found some cheap clone of the RPI and I'm trying to use that
[01:58] <mwhudson> sarnold: heh packet.net?
[01:58] <dunnousernamefn> Is there still a minimal ubuntu iso?
[01:58] <sarnold> mwhudson: yeah ... now that I'm at their website and trying to find it again, I just don't ...
[01:58] <mwhudson> sarnold: it's there, just hiding
[01:58] <mwhudson> or was last week anyway
[01:59] <dunnousernamefn> Also mwhudson https://github.com/Wunkolo/OakSim
[01:59] <sarnold> hehe
[01:59] <dunnousernamefn> Or https://github.com/atrosinenko/qemujs
[01:59] <sarnold> dunnousernamefn: maybe http://cdimages.ubuntu.com/netboot/xenial/ ?
[01:59] <dunnousernamefn> I think I can get everything I need if I just have apt and a dhcp client
[02:00] <dunnousernamefn> Is that provided?
[02:00] <dunnousernamefn> Oh, but I need openssh
[02:00] <mwhudson> oh scaleway have armv8 now too
[02:01] <sarnold> oo
[02:01] <sarnold> "Formerly the "Type 2a""
[02:01] <sarnold> sure enough they went to some effort to confuse me
[02:02] <dunnousernamefn> Packet.net says tiny was "as low as $0.00/hr in the last 7 days."
[02:02] <dunnousernamefn> Cool
[02:02] <sarnold> three euro per month for an armv7 and two gigs ram, 50 gigs ssd. wow.
[02:02] <mwhudson> also thunderx it seems
[02:04] <dunnousernamefn> Woah I didn't know that there were legitimate relatively-public ARM cpus
[02:04] <dunnousernamefn> I thought it was all proprietary
[02:05] <dunnousernamefn> Anyways, I guess I should get 512MB
[02:05] <sarnold> I expect you'd be far happier  with 512 than 128
[02:06] <dunnousernamefn> No, 256
[02:06] <dunnousernamefn> Wait, did I say 128?
[02:06] <sarnold> you did say 256 ..
[02:31] <JanC> sarnold: we used to run interpreted language websites using CGI on 64MiB systems 18 years ago, why would that suddenly be a problem...  ;)
[02:32] <sarnold> JanC: because perl was small and fast and quick and python is none of these things :)
[02:32] <JanC> and that included MySQL, BIND, Apache, qmail, etc. all on the same server IIRC
[02:32] <JanC> we actually had Perl & PHP websites on that
[02:33] <JanC> to be fair, the machine was replaced with a 512 MiB machine the next year
[02:33] <sarnold> JanC: I still remember the docs saying "you can run linux in 4 megs of memory but if you want to use X11 you really do need 16M"
[02:33] <JanC> sorry, I forgot, we had Perl, PHP & ColdFusion sites on that same machine
[02:33] <sarnold> *coldfusion* too? wow ;)
[02:34] <JanC> I had to write sites in ColdFusion actually...
[02:34] <JanC> the Perl & PHP stuff was things we hosted
[02:35] <JanC> of course most people were still on dial-up back then, so slowness was easy to explain (except to the then small but growing number of those on cable)
[02:39] <JanC> but that 64 MiB linux machine didn't do too badly (it was about comparably with a 384MiB Windows NT machine with IIS, Exchange & MS SQL Server that my employer also had)
[02:39] <sarnold> let me guess, qmail handled *way* more mail than exchange did ..
[02:40] <JanC> I'm not sure which one handled more mail (the exchange was used for internal mail), qmail (and some POP server, I forgot which) was used for client's mail
[02:42] <sarnold> I wonder why we need machines with a thousand times the memory these days :(
[02:42] <JanC> it wasn't exactly a big company, and it did all sorts of things from selling computer parts to hosting sites...
[02:42] <JanC> in 2000 that still existed  :)
[02:43] <sarnold> haha
[02:43] <JanC> well, I guess Amazon still does all that too  :P
[02:43] <JanC> at a slightly different scale
[02:43] <sarnold> and back in 2000 one of the leading ISPs in the area ... was a book store. so.
[02:45] <JanC> about the extra memory: like I said, dial-up was still common then, and most people didn't have internet yet
[02:45] <JanC> so the load on a server and the expected speeds were much much lower
[02:46] <JanC> cable internet started a couple years before in some cities & became more generally available that year
[02:47] <JanC> and even DSL was often limited to 0.5-4Mbit/s
[02:48] <JanC> and all of it was horribly expensive (compared to now)
[02:49] <sarnold> my first isp warned me that he only had a 14.4 uplink to the internet before I joined; that sounded fine by me since I also had a 14.4. :) hehehe
[02:49] <sarnold> it was so much faster than the 2400 I had been using on BBSes..
[02:49] <JanC> until you found out the only time you could use all of that was at 5am?  :)
[02:52] <JanC> funny detail: my current ISP which I switched to half a year ago is the only one which still operates dial-up here in Belgium  :)
[02:52] <JanC> I wonder who actually still uses that  :P
[02:52] <JanC> IIRC they said they would operate it until all their dial-in equipment is broken  :)
[02:52] <sarnold> JanC: heh, he upgraded to a T1 at some point :D
[02:52] <sarnold> wow
[02:52] <sarnold> that's gotta be expensive
[02:53] <sarnold> time for me to bail :) have  agood ... morning? JanC :)
[02:53] <dpb1> :)
[02:54] <JanC> sarnold: it's not expensive for them, I guess
[02:55] <JanC> they just need one system with a couple modems you can call into; I doubt there are many users
[03:07] <dunnousernamefn> https://www.debian.org/releases/stable/arm64/ch03s04.html.en wow, debian has practically the same html documents as ubuntu
[06:26] <Sircle> How to make wildfly re run if it gets crashed (this is happening to me a lot I dont know why). Using Ubuntu
[08:42] <RoyK> Sircle: I guess someone in #java might know - I have no idea myself :)
[12:27] <Neo4> http://91.227.18.36
[12:27] <Neo4> I've set up on server LAMP
[12:27] <Neo4> http://91.227.18.36:10000
[12:28] <Neo4> webmin is not useful. It is good use when you know how application works and know how to customize using command line
[12:29] <Neo4> I can install and customize:
[12:29] <Neo4> 1. create new user, Set SSH key, and install firewal UFW
[12:29] <Neo4> LAMP + phpmyadmin
[12:30] <Neo4> it would be good install there webserver postfix, postfixadmin
[12:31] <Neo4> wait guys, we will together install webserver in realtime
[12:31] <Neo4> now put postfix, squrelmal and roundcobe on Apache and then will see. wait
[12:32] <Neo4> I think together we could do this
[12:33] <blackflow> Neo4: which ubuntu?
[12:33] <Neo4> 16.04
[12:34] <blackflow> don't install roundcube from packages. it's unmaintained with lots of security vulns unfixed since the version in the repo.
[12:39] <Neo4> blackflow: I'll install for test
[13:37] <Neo4> I've installed postfixadmin
[13:37] <Neo4> http://kselax.ru/postfixadmin-3.1/login.php
[13:37] <Neo4> postfix and dovecote
[13:37] <Neo4> opened ports, 25 and 143,
[13:37] <Neo4> telnet kselax.ru 143 shows dovecot
[13:37] <Neo4> and telent kselax.ru 25 seems doesn't work?
[13:38] <Neo4> when do from localhost it show there postfix
[13:38] <Neo4> telnet localhost 25
[13:38] <Neo4> go on, we'll have today mail server :)
[13:38] <sdeziel> Neo4: I can reach kselax.ru 25
[13:39] <Neo4> sdeziel: ok, good
[13:39] <sdeziel> Neo4: maybe your ISP blocks outbound SMTP connections, this is common here
[13:39] <Neo4> sdeziel: what does it means? I won't able to get mails? How you can reach and I not?
[13:39] <Neo4> ok, you use differ ISP and my host placed in other ISP
[13:40] <sdeziel> that ^
[13:40] <Neo4> my ISP that I use for go to internet block 25 port?
[13:40] <Neo4> why?
[13:41] <hateball> To reduce spam is usually the main reason
[13:41] <Neo4> if somebody will want to send mail?
[13:41] <sdeziel> but I was just guessing so you might want to test it properly
[13:41] <Neo4> I can send mails from thunderbird
[13:41] <sdeziel> Neo4: email is usually send on TCP/587 or TCP/465 for the first hop
[13:41] <sdeziel> s/send/sent/
[13:41] <Neo4> sdeziel: see I got this in local computer
[13:41] <Neo4> neo@neo3:~$ telnet kselax.ru 25
[13:41] <Neo4> Trying 91.227.18.36...
[13:45] <Neo4> Ok, port will break down later, I want to install roudcobe and squrelmail now
[13:45] <Neo4> And then maybe have to rebuild postfix with support mysql
[14:06] <gpiccoli> Hi folks, I'm facing an "issue" with libvirt and apparmor in Ubuntu. It might be silly, but I'd like to understand if this is my mistake or some bug
[14:06] <gpiccoli> I've added an image file to my guest through XML, but libvirt cannot start the guest - permission denied
[14:07] <gpiccoli> It's apparmor blocking it - so how can I circumvent this in a right way?
[14:07] <gpiccoli> In the old times I've used selinux, and I was able to change the permission of the image file using selinux tools (i guess chcon)
[14:07] <Neo4> squirrelmail doesn't work, I dont know how to create there new user
[14:07] <Neo4> http://kselax.ru/squirrelmail-webmail-1.4.22/src/login.php
[14:07] <cpaelzer> gpiccoli: depends on the image and such, usually guests get a custom apparmor profile based on their description
[14:07] <gpiccoli> cpaelzer, I was informed that you may know the answer upfront heheh
[14:07] <cpaelzer> indeed
[14:08] <gpiccoli> wow, you're fast!
[14:08] <gpiccoli> I bet you have a highligt for the *virt*
[14:08] <gpiccoli> hehehe
[14:08] <cpaelzer> I have
[14:08] <gpiccoli> so, is there a way to change this profile cpaelzer ?
[14:08] <gpiccoli> it's a nvme image
[14:08] <cpaelzer> gpiccoli: well if not a super special awkward case first of all it should just work
[14:08] <gpiccoli> I'm adding through <qemu:command>
[14:08] <cpaelzer> aah
[14:08] <cpaelzer> here we go
[14:08] <gpiccoli> it is super awkard case!
[14:09] <cpaelzer> yeah qemu: namespace is invisible to libvirt and due to that not able to be handled by virt-aa-helper
[14:09] <cpaelzer> gpiccoli: but we can help it a bit :-)
[14:09] <gpiccoli> great cpaelzer =)
[14:09] <cpaelzer> gpiccoli: currently the only "problem" is that you have to apparmor-allow that path for all guests, you can not (yet) restrict it to just one
[14:09] <gpiccoli> it's totally fine by me!
[14:10] <gpiccoli> not so fine to people running vps services though heh
[14:10] <cpaelzer> gpiccoli: to do so go with editor of your choice to /etc/apparmor.d/abstractions/libvirt-qemu
[14:10] <cpaelzer> gpiccoli: we will get per guest includes, I just wait for an apparmor feature to land
[14:10] <gpiccoli> ok!
[14:10] <gpiccoli> cool =]
[14:10] <cpaelzer> in that file add your path with/without wildcards as you want
[14:11] <cpaelzer> actually
[14:11] <cpaelzer> which version of Ubuntu are you running?
[14:11] <gpiccoli> bionic cpaelzer
[14:12] <cpaelzer> ah we have no user-includes for the abstractions
[14:12] <cpaelzer> only for libvritd and virt-aa-helper
[14:12] <cpaelzer> go for the file I said
[14:12] <cpaelzer> for custom guest overrides it really will be the per guest include (one day)
[14:12] <gpiccoli> hehehe
[14:13] <gpiccoli> so cpaelzer, in that file, can I add /var/lib/libvirt/images/* X, where X should be...rw I guess?
[14:13] <cpaelzer> bug 1745114 for the per guest include btw
[14:13] <cpaelzer> gpiccoli: exactly
[14:13] <gpiccoli> cool cpaelzer, will be a great addition!
[14:13] <gpiccoli> cpaelzer, what is rwk ?
[14:15] <gpiccoli> cpaelzer, partially worked. Seems I don't have the permission issue aymore...or I might be
[14:15] <gpiccoli> 2018-04-17T14:13:59.811597Z qemu-system-x86_64: -drive file=/var/lib/libvirt/images/nvme0.img,if=none,id=nvme0: Failed to unlock byte 100
[14:15] <gpiccoli> Could be another type of issue...
[14:15] <gpiccoli> although the unlock keyword there might be related to permission again
[14:15] <cpaelzer> gpiccoli: http://manpages.ubuntu.com/manpages/artful/man5/apparmor.d.5.html
[14:16] <cpaelzer> TL;DR k = lock
[14:16] <cpaelzer> gpiccoli: so you need k
[14:16] <cpaelzer> as well
[14:16] <gpiccoli> hehehe
[14:16] <gpiccoli> cool, thanks a lot cpaelzer
[14:16] <cpaelzer> newer qemu locks all images to ensure it is used mutually exclusive
[14:16] <cpaelzer> have fun gpiccoli
[14:16] <gpiccoli> worked like a charm cpaelzer
[14:16] <gpiccoli> makes total sense!
[14:16] <cpaelzer> might I ask who blackened me that I'd know these things?
[14:17] <gpiccoli> cpaelzer, some bird...in some place...
[14:17] <gpiccoli> hahaha
[14:17] <gpiccoli> kidding,
[14:18] <gpiccoli> it's jdstrand!
[14:18] <gpiccoli> from apparmor =)
[14:18] <cpaelzer> thanks
[14:19] <cpaelzer> he is "allowed" to point to me :-) I tihnk I still owe a few favors for all the help I got :-)
[14:19] <gpiccoli> heheh
[14:20] <ahasenack> cpaelzer: I'm looking at debian policy 5.6.30 Testsuite, where it says that source package control files may have that field "if needed in other situations". No clue what "other situations" might be
[14:20] <ahasenack> cpaelzer: I grepped existing packages I had lying around, and found XS-Testsuite instead of Testsuite
[14:20] <ahasenack> do you know what's the difference?
[14:21] <cpaelzer> I remember discussing that once
[14:21]  * cpaelzer starts the page in of old memory
[14:21] <ahasenack> fwiw, I got "Testsuite: autopkgtest" automatically included in the .dsc file
[14:25] <cpaelzer> ahasenack: XS- ... is old style is what my memory tried to give me
[14:25] <cpaelzer> does that make any sense?
[14:25] <cpaelzer> with what you see?
[14:25] <ahasenack> my sample *only* had xs-
[14:26] <ahasenack> but it does sound like a prefix to be used with non-official fields
[14:26] <ahasenack> like in email headers
[14:26] <ahasenack> I searched for "xs" in the debian policy, found nothing
[14:26] <cpaelzer> ahasenack: https://webcache.googleusercontent.com/search?q=cache:rTRO6caAUjkJ:https://lintian.debian.org/tags/xs-testsuite-header-in-debian-control.html+&cd=1&hl=en&ct=clnk&gl=de
[14:27] <ahasenack> my local sample: https://pastebin.ubuntu.com/p/9vFfQ52nFt/
[14:27] <ahasenack> ah, ok, so it became an official field
[14:27] <cpaelzer> hmm this page of lintian is gone
[14:27] <cpaelzer> I think this just is from the far past
[14:27] <ahasenack> I'll add Testsuite then
[14:27] <cpaelzer> yes I thnk XS- is hwo it started changing later
[14:27] <ahasenack> just like it is in the .dsc
[14:28] <cpaelzer> too much posts exception on this thread https://lists.debian.org/debian-devel/2014/01/msg00040.html
[14:28] <cpaelzer> if you want to read a lot feel free
[14:28] <ahasenack> not even vim's syntax highlighting recognizes "Testsuite"
[14:38] <coreycb> jamespage: can you remind me how to get around the unexpected upstream changes when building ceph?
[14:38] <jamespage> coreycb: its todo with the checkout in git which translates to local system line-endings
[14:38] <jamespage> if you patch -R the delta file, and the git add the diff it will sort things out
[14:40] <coreycb> jamespage: ack thanks
[14:43] <popey> hello! who 'owns' the ubuntu server armhf images?
[14:43] <popey> a valiant community member has done some testing...
[14:43] <popey> http://iso.qa.ubuntu.com/qatracker/milestones/384/builds/169986/testcases/1464/results/
[14:46] <powersj> popey: those bugs look like linux-* issues
[14:46] <powersj> I would expect the kernel team to respond on making the proposed changes
[14:47] <powersj> I know that dannf does a number of arm testing as well
[14:48] <popey> ok, thanks.
[14:48] <dannf> powersj: yeah, lots of arm64, but don't have any armhf gear
[14:58] <ahasenack> cpaelzer: meh, ok, so even Testsuite is a lintian warning now
[14:58] <ahasenack> N:    You do not need to specify a Testsuite: autopkgtest field if a
[14:58] <ahasenack> N:    debian/tests/control file exists. It is automatically added by
[14:58] <ahasenack> N:    dpkg-source(1) since dpkg 1.17.1.
[14:59] <ahasenack> vim was right, as usual :)
[16:11] <teward> rbasak: nacc: dpb1: oooh guess what came out today xD  NGINX 1.14.0
[16:11] <dpb1> teward: and what I want to hear next.... "teward: I've already packaged it and tested it"
[16:11] <dpb1> teward: :) :)
[16:12] <teward> dpb1: waiting for them to spin the tarballs
[16:12] <teward> it's tagged in git.
[16:12] <dpb1> teward: that's cool
[16:12] <teward> or rather hg.
[16:12] <dpb1> right
[16:12] <dpb1> ya, I found that out when I was digging into your request from a few weeks back
[16:13] <teward> dpb1: my guess is by EOD it'll be available.
[16:14] <teward> yep and there's the announcement.
[16:14] <dpb1> teward: crunch time.
[16:14] <teward> the tarball should be up soon
[16:14] <teward> yep I know but hey at least we won't need a post-release update :P
[16:14] <teward> and i have the base packaging I did yesterday and uploaded so that's pretty much ready to go :P
[16:14] <teward> just need the **tarball**
[16:15] <teward> yay it's a no-changes thing too :D
[16:25] <teward> dpb1: local test builds worked fine, and I got the defaults working.  It's also sitting in the approval queue now, so if all goes well then we won't be scrambling for a post-18.04-release changeset to switch the version numbers over.
[16:25] <teward> (for once)
[16:26] <dpb1> teward: well, I'll be.
[16:26] <dpb1> teward: nice.
[16:28] <teward> now where's my lunch...
[16:30] <runelind_q> I have an install of 16.04 on an mdadm mirror as a boot disk and a ZoL pool as LXD storage.  I'm wondering what my chances of a relatively pain free migration of the disks to an entirely different system would be?
[16:36] <nacc> teward: nice!
[16:38] <sdeziel> runelind_q: what you could do is set the new system up (maybe skip mdadm if you use zfs?) and then to "lxc move" or "lxc copy" of your containers
[16:39] <runelind_q> sdeziel: yeah, I was just hoping to re-use the disks that I was using for ZoL
[16:40] <blackflow> runelind_q: keep in mind that different systems might have issues with different ZFS pool versions. eg, pools created on 0.7.x ZoL are read-only on 0.6.x ZoL systems
[16:40] <sdeziel> runelind_q: ah, in that case, you could break your mirror and set the new box with half a mirror. It's a small gamble but should work
[16:40] <blackflow> runelind_q: it's actually about features enabled on the pools, so i'm talking about default features with respective versions.
[16:41] <sdeziel> runelind_q: with 18.04 almost out, you may also want to use it for your new bo
[16:41] <runelind_q> blackflow: yeah, but I was going to move the mdadm boot mirror as well.  Complete forklift of the system onto new HW
[16:41] <runelind_q> yeah, that is a consideration.
[16:41] <sdeziel> blackflow: that's a good point. If lxc move/copy is used, this should be transparent though
[16:42] <blackflow> there's also that bug about zfs send-receive between different pool versions, in 0.6.x. it also affects FreeBSD iirc.
[16:43] <sdeziel> blackflow: but that only affected a release that was never imported in Ubuntu, AFAIK
[16:43] <sdeziel> err, nvm, I'm referring to something affecting 0.7.7
[16:44] <blackflow> sdeziel: no it currently affects pre-bionic ubuntus.
[16:44] <blackflow> yeah, you probably mean the data corruption bug found the other day in 0.7.7
[16:44] <sdeziel> yup
[16:45] <blackflow> I think this is the LP issue  https://bugs.launchpad.net/ubuntu/+source/zfs-linux/+bug/1733230
[16:46] <blackflow> it's not just about FreeBSD. I had that issue when sending from a pool on Gentoo (0.7.x) to Ubuntu Artful (0.6.5)
[16:47] <blackflow> otherwise I run ZoL on root in 17.10 (with separate ext4 /boot, but due to LUKS'd volume for ZFS), works like a charm.
[16:48] <blackflow> eh... root on ZoL :)
[16:48] <sdeziel> that's something I don't have the balls to try just yet
[16:48] <blackflow> I have experience from FreeBSD, but was pleasantly surprised how everything just works. Then again I set up the pool manually and use debootstrap for installation.
[16:50] <ScottE> Yeah, I've been running with ZFS for everything (except /boot, sometimes - for the same LUKS reason). It's great! A bit more work to install via debootstrap, but worth it in my opinion.
[16:53] <blackflow> I don't know if grub is capable of mounting ZFS pools yet, I know it wasn't on 17.10, so a separate /boot is required even without LUKS.
[16:54] <sdeziel> what worried me is the rescuability of this root on ZFS
[16:54] <sdeziel> Ubuntu live CD didn't have the zfsutils-linux package installed last I checked
[16:54] <dpb1> there is one issue
[16:54] <ScottE> It is capable - I've done it without a separate /boot on xenial and jessie but it's a bit more finicky about grub version and configuration
[16:55] <dpb1> the on disk format version.
[16:55] <blackflow> sdeziel: yeah you have to install it manually on live cd, but if it can mount root, then all the tools are in initramfs. it's basically just two commands, zpool and zfs, with zpool doing all the important work. dunno if zdb is present, not sure if it's needed though. I actually never had to use it.
[16:57] <blackflow> dpb1: that's the part about different versions I mentioned first. a warning about using systems with ZoL 0.7.x and systems with ZoL 0.6.x, the former pools being only read-only on 0.6.x
[16:58] <dpb1> ya... you need to set '28' to be compatible with oracle zfs
[16:58] <sdeziel> blackflow: I want to be able to rescue my box offline so installing packages in the live env is tricky.
[16:58] <dpb1> or something like that
[16:58] <dpb1> it's one considering we are thinking about before enabling this as a default option in Ubuntu
[16:58] <dpb1> *consideration
[16:58] <blackflow> sdeziel: I've installed ubuntu w/ root on ZFS both from the 17.10 live env, and from debian stretch, in both cases using debootstrap.
[16:59] <blackflow> sdeziel: in both cases I had to install zfsutils-linux, which in addition had to compile the DKMS on debian (not needed on Ubuntu, it's part of the kernel package)
[17:00] <sdeziel> blackflow: sure but my main concert is the offline rescue environment so I cannot apt-get install zfsutils-linux so even if the zfs.ko is available, it's useless to me
[17:00] <blackflow> sdeziel: why can't you use apt?
[17:00] <blackflow> (unless there's no networking of course)
[17:01] <teward> blackflow: key words in sdeziel's message: "offline rescue environment"
[17:01] <teward> no network there
[17:01] <sdeziel> blackflow: I want offline rescue capabilities
[17:01] <blackflow> ah, yeah.
[17:01] <blackflow> I guess you can always prepare a bootable USB stick with all tools needed? I had one for Gentoo.
[17:02] <sdeziel> or I could fill a bug to have zfsutils-linux added to the live env
[17:02] <blackflow> or... that, yeah :)  that'd be better.
[17:04]  * sdeziel doesn't know which package to report the bug to
[17:04] <nacc> sdeziel: i'd file it on zfsutils-linux first
[17:05] <sdeziel> nacc: on my way
[17:05] <nacc> sdeziel: as it sounds like you want it seeded?
[17:05] <runelind_q> is ubuntu a rolling release?  Like if I download the latest nightly of 18.04, apt will keep me in sync once the final version comes out?
[17:06] <nacc> runelind_q: that's not what a rolling release is :)
[17:06] <runelind_q> oh
[17:06] <nacc> runelind_q: but apt will keep you up to date with 18.04 yes
[17:06] <runelind_q> I don't need to do dist-upgrade or anything like that?
[17:06] <sdeziel> nacc: yes, exactly
[17:06] <nacc> runelind_q: well, you should (in general) be using full-upgrade (IMO) until release
[17:06] <nacc> and even after, probably
[17:07] <nacc> sdeziel: it's rather late for seed changes, fyi
[17:07] <nacc> sdeziel: you may want to file the bug and then bring it up in #ubuntu-release
[17:07] <sdeziel> nacc: I know but I'm in no rush
[17:07] <nacc> sdeziel: ok :)
[17:32] <ahasenack> what is "/usr/lib/triplet" (sic), in the context of the debian policy for shared libraries?
[20:06] <ddstreet> smoser fyi, remember that isc-dhcp 'wait for DAD' bug (and related follow-on bugs for it)?  It looks like upstream isc-dhcp has "fixed" that by adding a --dad-wait-time param that of course defaults to 0 (and requires the OS script to support it too).  just fyi.  https://bugs.isc.org/Public/Bug/Display.html?id=36169
[20:13] <smoser> ddstreet: nice. that http redirects you to a https that mozilla doesnt like
[20:14] <Sircle> How to make wildfly re run if it gets crashed (this is happening to me a lot I dont know why). Using Ubuntu
[20:14] <smoser> ddstreet: you should link to that in our bug.
[20:19] <ddstreet> smoser yeah i think their (newly public, as far as i know) bug tracker is not direct-link friendly
[20:20] <ddstreet> i'll put a note in the bug(s)...and also fyi there's yet another offshoot of this same bug, for 'stateless' dhcp, lp #1764478
[20:20] <ddstreet> maybe we can move off isc-dhcp to something more...maintained...
[20:22] <ddstreet> especially since ISC appears to be abandoning their isc-dhcp-server in favor of their 'kea' server http://kea.isc.org/wiki
[20:25] <smoser> ddstreet: well, we kind of have moved off isc-dhcp
[20:25] <smoser> in fact it is no longer part of ubuntu-minimal in bionic
[20:25] <ddstreet> oh good, what's the current recommendation?
[20:26] <smoser> now its replacement ..... systemd-networkd i suspect has its share of issues.
[20:26] <smoser> https://bugs.launchpad.net/ubuntu/+source/maas/+bug/1717983
[20:26] <ddstreet> ah, systemd...one ring to rule them all... ;-)
[20:27] <ddstreet> smoser i haven't looked, do you know if networkd includes a dhcp server as well, or only client?
[20:28] <smoser> ddstreet: dont worry, I have a feeling you'll have plenty of opportunity to become more familiar with systemd-networkd's dhcp server via bugs in the coming  months :)
[20:28] <smoser> ddstreet just a client
[20:28] <ddstreet> oh, i'm *sure* of that ;-)
[20:28] <ddstreet> any 'official' recommendation for dhcp server?
[20:29] <ddstreet> dnsmasq?
[20:29] <smoser> isc-dhcp-server is still in main as is dnsmasq.
[20:29] <sdeziel> systemd-networkd supports being a DHCP server (DHCPServer=)
[20:30] <smoser> wow.
[20:31] <ddstreet> soon, it will be a battle between emacs and systemd to see who can become the next OS-in-application
[20:47] <ddstreet> smoser wow i didn't realize you'd already opened a 'stateless' dhcp bug, lp #1633562
[20:47] <ddstreet> that's definitely not invalid...oh well i'll work it in my newer bug :)
[21:58] <MASM> .
[22:24] <Sircle> My fail2ban bans ip ranges with /32. I want it /24 at least. How can I do it?
[22:28] <MASM> Sircle i found an example
[22:28] <MASM> https://github.com/fail2ban/fail2ban/issues/927
[22:29] <MASM> "subnet.blacklist" > "198.27.100.224/29"
[22:29] <MASM> or
[22:29] <MASM> "ip.blacklist" > "198.27.100.224 - 198.27.100.231"
[22:30] <MASM> or check this
[22:30] <MASM> https://github.com/XaF/fail2ban-subnets
[22:30] <Sircle> MASM,  I mean when f2b auto bans anything
[22:30] <Sircle> not blacklist