/srv/irclogs.ubuntu.com/2016/06/30/#ubuntu-server.txt

torriztest00:07
sarnoldtorriz: test00:07
sarnoldtest torriz00:07
sarnoldtest00:07
genii!test | torriz00:09
ubottutorriz: Testing... Testing... 1. 2.. 3...00:09
torriz:D00:09
patdk-lapsarnold, you want to look at my interfaces file?00:14
patdk-lapdid several more tests00:14
sarnoldpatdk-lap: sure00:14
patdk-lapstill cannot get the gateway to turn up no matter what00:14
patdk-lapeven had a problem when it was a single static ip00:15
patdk-lapor interface00:15
patdk-laphttp://apaste.info/ePt00:16
patdk-lapit's worked on all ubuntu systems I have had, including 16.04, except on these two new ones I setup a few weeks ago00:17
patdk-lapI even attempted to add an, up route add default gw 10.1.0.100:18
patdk-lapbut that didn't work00:18
patdk-lapthough running it on the cli manually, works fine00:18
patdk-lapremoving everything except bond0, still doesn't make it work00:19
sarnoldpatdk-lap: "iface bond0 inet manual"  -- not 'static'? bond2 is set static00:20
patdk-lapit has no ip00:20
sarnoldahh00:20
patdk-lapatleast it works fine on my system here at home :)00:21
sarnoldpatdk-lap: no idea :/ it's the single most complicated one I've seen :) but it all makes sense.00:29
patdk-lapwell, even with just the two entries, to make bond0 and bond0.4 work, gateway won't come up00:30
patdk-lapthat is far from complicated :)00:31
* sarnold shudders00:31
patdk-lapthat is just a simple storage server :)00:31
patdk-lapyou should look at the router/firewall ones00:31
TarifaI just installed an Ubuntu server.  Was having a heckuva time getting ssh to reliably startup.  STarted to poke around, and find there are TWO init scripts -- /etc/init.d/ssh (for upstart) and /lib/systemd/system/ssh.service (for systemd).  I discovered that enabling one, seems to cause the other to try to start.  Screws up startup dependencies it seems.  As a test, I simply moved /etc/init.d/ssh out of the way, and restarted.02:06
TarifaEverything works perfectly.02:06
TarifaI'm not sure WHY there are both an upstart ssh and a systemd ssh installed; though Ubuntu had moved.02:07
TarifaANyway, is that the right way to DISABLE the /etc/init.d/ssh -- just move it?02:07
sarnoldhah, even better than that, there's _three_ different sshd init systems -- /etc/init.d/ssh /etc/init/ssh.conf and /lib/.../ssh.service. *sigh*02:09
RoyKTarifa: I stick to debian02:09
sarnoldif you move the initscript it'll probably come back on future package upgrades02:09
patdk-lapjust use the update-rc.d or whatever these days02:09
RoyKTarifa: for servers02:09
patdk-lapor systemctl disable/mask/...02:09
sarnoldTarifa: please file a bug, 'ubuntu-bug openssh' and tag it with systemd-boot, I think that's how it'll get to the right people02:10
patdk-lapsarnold, no, only if you use rpm :)02:10
sarnoldlol02:10
Tarifapatdk-lap: disable/mask to disable the init.d upstart service?02:10
patdk-lapwhy I left rhel, cause all the files kept coming back endlessly02:10
patdk-lapno to disable systemd service02:10
patdk-lapsystemd will normally filter it down to upstart if needed though02:10
Tarifapatdk-lap: Ah.  I want the "other way"  --  ENable systemd, but DISable upstart02:11
patdk-lapnormally systemd service just calls the upstart service02:11
patdk-lapsystemd likes to be in-charge02:11
Tarifapatdk-lap: yeah, and I'm happy with systemd.  Just trying to figure out how to get the upstart stuff to "leave me alone" ....02:12
=== steve____ is now known as stiv2k
Tarifasarnold: If I can figure it out enough to post more that "It don't work!", I will02:15
sarnoldTarifa: I've gotten the impression that one common reason why sshd doesn't come up reliably is that it binds to specific IP addresses withuot using IP_FREEBIND -- and that some network interfaces take too long to come up, so their addresses don't exist, so *boom*02:17
sarnoldTarifa: I wonder if fiddling with the sysv-init symlink causes it to run at a vastly different time..02:17
Tarifasarnold: Yeah, I set specific IP addresses to listen.  That's easily taken care of by adding network-online.target dependency to a Unit file override.   Eventually I'll also have ssh use an nfs-mounted /etc/ssh, and add a 'RequiresMountsFor=' clause in a Unit file.  That delays ssh nicely until the network's up.02:22
Tarifasystemd makes it easy.  Trick *here* is to get Ubuntu to stop futzing around with upstart ssh.  Really, no clue why the /etc/init.d/ssh is still even installed.02:23
sarnoldTarifa: I think it was just a matter of running out of time to remove all the system upstart configs before release02:29
sarnoldto do it Right would probably mean submitting patches to debian for the packages there with upstart configs, etc., and all that takes time.02:29
Tarifasarnold: So even the Ubu init system is heavily dependent on Debian?  As you can guess, I'm new to this env.  systemd on Opensuse is very clean and pretty much legacy free at this point.02:31
sarnoldTarifa: yes, they've had several more years to clean it up, and only the one distribution to worry about :)02:31
sarnoldTarifa: funny thing is, debian didn't get rid of their N different inits.. it just switched The Default. we may not be able to get rid of the upstart configuration files, we may be stuck with keeping around config files for all three major inits :(02:33
TarifaWell, more than one.  Suse != Opensuse.  And now Opensuse Leap != Opensuse Tumbleweed.02:33
TarifaThat's part of the problem I'm seeing these days.02:33
Tarifasarnold: Hm.  Hadn't realized that.  Rolling these out in this 'shape' is gonna cause some problems it appears.02:34
sarnoldTarifa: well, that's what made me suggest the bug report -- this really should Just Work, even if it is bloody annoying on sysadmins to figure out which tool to use to disable a service. But sshd should start everywhere without trouble, and it mostly does (modulo those non-existant interfaces I mentioned a little while ago)02:35
Tarifaagree on the JustWork(tm)02:36
TarifaIn my experience so far, the distros don't do a particularly good job of either SSH or NFS startup.  Lots of sloppy, poorly thought out Unit files out there.02:37
Tarifaaaaaaand, of course, getting the maintainers to actually respond is like pulling teeth.02:38
* Tarifa sighs02:38
sarnoldI haven't noticed any issues with my NFS setup -- what went wrong for you there?02:38
Tarifasarnold: It's working now.  But lots of mess with waiting for network interfaces to come up, drives to spin up, and static port deployments.  To 'amplify' the problem, try a big server with a dozen interfaces and 40+ drives in raid arrays.02:41
TarifaLots of fun.02:41
sarnoldTarifa: aha :) I've only got the one interface and ~dozen drives. heh.02:42
TarifaThank the stars for systemd's flexibility, imo!02:42
TarifaThrow a KVM or Xen server in there, and watch the sparks fly differently for each distro.02:42
Tarifasarnold: Then file a bug, and get back "Can't reproduce here, therefore not a bug".02:43
TarifaSome of those folks need to upgrade their RPi's if you ask me ;-)02:44
patdk-lapjust mark it invalid or won'tfix :)02:44
patdk-lapcan't upgrade02:44
patdk-laplinux kernel has bugs, that make it unusable for me02:44
sarnoldhaha02:44
Tarifaheh02:44
patdk-lapthey added in new network code for min packet header length02:45
patdk-lapbut the min length is set to the max size02:45
patdk-lapand for variable sized header protocols, well, that doesn't work out well02:45
patdk-lapfirst people to start complaining where the rpi users02:46
TarifaAh, just got a ping from a friend.  Secret sauce seems to be: "systemctl disable ssh; update-rc.d ssh disable; systemctl enable ssh".  Apparently all 3 are needed.  Then add both Wants=network-online.target & Requires=network-online.target in an ssh service override.02:48
TarifaJust tried it -- works a charm.02:48
Tarifai really wish the devs would stop & RTFM on systemd network dependencies.  Specifically 'network' vs 'network-online'.02:49
sarnoldTarifa: please do file a bug.02:50
patdk-lapyou just want to mark it, won't fix02:50
sarnoldpatdk-lap: not me, pitti :)02:50
sarnoldI'm still new enuogh with systemd that I'm basically fish out of water .. he knows what's going on and is in a much better position to figure out what needs to be done about it :)02:51
Tarifafwiw, @ https://www.freedesktop.org/wiki/Software/systemd/NetworkTarget/: "network.target has very little meaning during start-up" & "network-online.target is a target that actively waits until the nework is "up""02:51
TarifaGuess which one is used in Ubu's ssh.service?02:51
patdk-laphaven't looked :)02:52
patdk-lapwonder what is in my ssh package02:52
Tarifapatdk-lap: Don't.  Be Happy.02:52
TarifaWow. The kitchen sink's in here.  ssh@.service & ssh.socket too.02:53
TarifaAnyway. TT4N.02:59
k2gremlinHello all, I was wondering the best approach to setup 2 hard drive for my media server. The OS drive I would like to have say 25GB. The "Storage" drive would be on the same datastore and maybe around 400gb.04:14
sarnoldk2gremlin: i've got to run, here's a nice series of blog post about zfs that may interest you :)  https://pthree.org/2012/12/04/zfs-administration-part-i-vdevs/04:16
k2gremlinThe reason for this is if I ever mess up my OS.. I would like to rebuilt it and add the storage vmdk as a secondary04:16
sarnold(fwiw I think it's still too much trouble to use zfs on root; simple ext4 or something is a lot easier there)04:17
patdk-lapbtrfs for root?04:18
sarnoldwell if you want reasons to rebuild it.. hehe04:19
patdk-lapweekly?04:19
sarnoldhey, the front page of their wiki no longer says "don't use this in production". it might be worth a look soon then.04:21
patdk-lapdoubtful04:21
sarnoldtrue, I'd rather have easy zfs on root (where easy means not having to read a howto :)04:22
patdk-lapit's always a nightmare for me to do encrypted root04:23
patdk-lapthis last time to go encrypted, grub encrypted + EFI grub, took awhile04:23
sarnoldhmm I thuoght the installer would do that if you wanted.04:25
* sarnold blames patdk-lap and runs for dinner04:25
patdk-lapit doesn't really like encrypted /boot04:26
sarnoldahhhhh. /boot is getting harder to do these days.04:26
sarnoldI gave in, I think none of my systems have a separate /boot any longer04:27
patdk-lapbut as I learned about EFI04:27
patdk-lapheh, that is just a whole vaun point04:27
patdk-lapas I see it04:27
stephanbuyshi all, anyone here from canonical to chat about commercial support/troubleshooting of a failing system? (tried the web form a couple of weeks ago but got no reply)05:34
M4R4B4!help05:55
ubottuPlease 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 !patience05:55
M4R4B4I need a scan like this someone has ??? http://prnt.sc/bmwh0005:55
=== ksx4system_ is now known as ksx4system
=== chmurifr- is now known as chmurifree
=== Agusanz_ is now known as Agusanz
=== profall_ is now known as profall
=== _degorenko|afk is now known as degorenko
=== jamespag` is now known as jamespge
=== jamespge is now known as jamespage
=== PaulW2U_ is now known as PaulW2U
=== josepht` is now known as josepht
EmilienM§b rdo-ci13:18
cpaelzernacc: the step of "usd-importer/usd-merge reconstruct old/ubuntu" usually fails for me13:42
cpaelzernacc: reason is that it has no local representation of debian/sid at this stage13:43
cpaelzernacc: usd-merge: debian/sid is not a defined object in this git repository.13:43
cpaelzernacc: a simple "git checkout debian/sid" gets it going again13:43
rbasakcpaelzer: try "git branch debian/sid origin/debian/sid", assuming your remote is "origin".13:43
cpaelzerrbasak: yeah the branch achieves likely the same as my checkout13:44
rbasakcpaelzer: this is bug 159529813:44
ubottubug 1595298 in usd-importer "usd-merge reconstruct [<onto>] defaults to debian/sid instead of origin/debian/sid" [Undecided,New] https://launchpad.net/bugs/159529813:44
cpaelzerthanks for the pointer rbasak13:44
pulsar12when i create a file on a file-ACL enabled system, the mask applied is rw-. but i have the group set to rwx. i was expecting to see mask=rwx, since documentation states that ACL mask is the union of all group and user permissions14:28
=== `jpg_ is now known as `jpg
=== caribou_ is now known as caribou
nacccpaelzer: yes, or do `usd-merge reconstruct old/ubuntu origin/debian/sid`15:27
nacccpaelzer: will fix on the bug15:27
cpaelzernacc: now we have three workarounds :-)15:33
cpaelzernacc: thanks for fixing it somewhen15:33
nacccpaelzer: either today or tmrw, i hope15:35
=== tarpman_ is now known as tarpman
=== degorenko is now known as _degorenko|afk
=== sarnold_ is now known as sarnold
=== crazybluek is now known as Blueking
=== ivoks_ is now known as ivoks
=== not_phunyguy is now known as phunyguy

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!