/srv/irclogs.ubuntu.com/2015/08/24/#ubuntu-server.txt

=== Lcawte is now known as Lcawte|Away
=== txomon|home is now known as txomon|sauron
=== txomon|sauron is now known as txomon|home
=== markthomas|away is now known as markthomas
=== Joel is now known as Guest70422
=== markthomas is now known as markthomas|away
moneylotionhowdy, hey, anyone have an idea, how i can rsync "from" a read only file system (dvd)06:09
=== CiPi is now known as cipi
andolmoneylotion: The same way you would rsync from any other file system/path?06:10
moneylotion"operation not permitted"06:11
moneylotiona bunch of files have weird used, which cannot be changed06:12
moneylotionweird users06:12
andolAny chance that the problem is rather the destination, you are rsync:ing to?06:13
moneylotionhmmm.... i was copying to a network volume... seems to work when i go to my desktop06:14
lordievaderGood morning.06:40
andolmoneylotion: Well, if it was to a nework volume things makes a lot more sense, given that the mounting user might not have suffecient access to changer ownerships, or that the network filesystem handles users in a different manner.07:01
moneylotioni guess rsync can ignore permission :)07:01
andolmoneylotion: Yepp, just don't add flags telling rsync to care about permissions, ownerships, etc07:04
=== neurotus is now known as krsna
moneylotionany ways I can significantly compress backups?08:19
bearfacedepends if the data is significantly compressible08:20
=== Lcawte|Away is now known as Lcawte
purefanhello. I found a server who just rebooted and Im trying to find the cause, any advice?09:16
lordievaderpurefan: Look in the logs?09:16
purefanlordievader: I have looked at syslog, auth.log, lastlog, boot.log, apache error.log, apache access.log and wtmp, nothing seems unusual09:17
purefanlordievader: apache reports getting a sigterm but I cant find what triggered it09:17
lordievaderDoes the kernel log show anything?09:18
lordievaderOr does it just stop -> bios.09:18
purefanits actually missing the entries from around the server reboot09:24
purefanserver reboot happened around 2am, and the earliest entry is from 12:4009:24
purefanoh hold on09:25
lordievaderpurefan: That might indicate a failing disk. Bit difficult to write to a disk if it ain't there anymore.09:25
lordievaderCheck smart, I'd say.09:25
purefanlordievader: the earliest record in the previous kernel log is just for terminating ttys due to Sigterm09:26
lordievaderHmm, that sounds a bit like an actual reboot.09:27
purefanand thats weird because afaik no one was even awake at the time09:28
purefanlordievader: Do you think a faulty HD is a risk in a cloud environment? this server runs on AWS09:38
arcsky_Im going to install ubuntu on a server with RAID5. should i use the ubuntu installation on the entire drive or just a part of it?09:40
Walexarcsky_: depends on how you do it and what you want to achieve. Is that hw or sw RAID5? Do you want the system filetrees to be RAID'ed underneath? etc.09:43
Walexarcsky_: a typical installation profile is to have on each disk a '/' partition, a swap partition and an "other" partition, and if using sw RAID the "other" partitions are those that are RAID5-ed09:45
Walexarcsky_: traditionally the '/' partitions are RAID1-ed to make it easy to boot from them, and 'swap' does not need to be RAID'ed09:45
Walexusually at least, some people RAID1 the 'swap' too09:46
lordievaderpurefan: No, for a vm it is unlikely.09:52
purefanlordievader: Thanks09:52
lordievaderI guess Amazon has pretty good monitoring on that.09:52
purefanlordievader: I think so too, I would expect at least an alert09:53
lordievaderWalex: Raiding swap is usually a good idea. Prevents inconsistent swap when a disk fails.09:53
lordievaderpurefan: You won't receive it, their sys-admin team would get that alert.09:53
purefanlordievader: Ok, but they would let us know if the service was interrupted because of that, right?09:54
lordievaderI don'09:56
lordievaderdon't think their service will be interupted by a disk failure.09:56
purefanlordievader: Yeah I dont think it would either... I need to report this and Im looking at all possible angles, its not a good thing that a server reboots itself without an explanation09:58
SiilwynHi people, if I want to restrict a group of users to only have access to one directory through sftp (Match Group www-data), does it match all users in that group or only if it's their primary group?10:01
arcsky_Walex: its hw raid10:19
Walexarcsky_: then hw RAID is transparent to the OS and you need not worry.10:20
linociscohi all10:21
linociscono talloc stackframe at ../source3/param/loadparm.c:4864, leaking memory10:21
mnms_Guys it is reasonable to give 10GB for swap partition ?10:39
mnms_I have 5GB of memory10:39
mnms_it will be a mail server10:39
mnms_Somewhere I read it shoubd twice of RAM, but it is not too much ?10:40
mybalzitchits not much space10:43
mnms_mybalzitch: you mean 10GB for swap its not much space yes ?10:45
mybalzitchyes10:45
mnms_mybalzitch: I assigned 8GB finally, can I have problems ?10:46
mybalzitchif you do, add mroe ram10:46
ogra_well, mailservers are pretty I/O bound ... if you start swapping you should ASAP get more ram10:46
ogra_on a busy mail server, swap is rather a safety net than anything else10:47
mnms_I have 2xSSD disks and I want to put them in software RAID110:47
mybalzitchmdadm10:47
mnms_what initial values you would use for this configuration ?10:54
mnms_Yes, I dont want to use swap, I think 5GB of memory will be enough for my server and I dont want to hibernate my machine10:54
mnms_mybalzitch: You are not saying to not use swap at all ?10:55
mnms_I mean to not create swap partition10:55
mybalzitchno10:56
mybalzitch5-10 gb of swap is plenty, especially if it lives on an SSD10:56
lordievaderCreate some swap, as has been said before, as a safety net.10:56
ogra_create it, but turn down the value for swappiness10:56
mybalzitchhow much mail are you handling anyway10:57
ogra_so it doesnt get used by default, only for overflowing if required10:57
mnms_I assigned 8GB10:59
mnms_lordievader: 1000 mails per day10:59
* ogra_ doubts you will even use any ram with that, unless all 1000 hit the machine at the same time or some such :)11:02
mnms_swappiness to 0 ?11:02
ogra_not 0 ... take 1 or 5 ... just less than the default11:03
mnms_or 10 like it is on help.ubuntu11:03
mnms_ogra_: ok11:03
mnms_so 2GB would be ok for SWAP cause I dont expect to use it all, correct ?11:05
mnms_this would be enough for my "safety net"11:07
ogra_right, and worst case you can immediately create a swap file and add it to the swap space in case you really hit an emergency11:09
mnms_ogra_: btw 2GB of swap is not small, yes ?11:10
ogra_2G should really be fine11:10
=== MrPPS is now known as Guest62462
mnms_guys could you tell me how I can remove raid device which I created before, is says it is in use11:23
mnms_when I try to stop I get "Cannot get exclusive access"11:24
mnms_by mdadm --stop /dev/md011:24
mnms_Ok..solved it.11:32
SiilwynJust a quick question, can I chroot a sftp user to a non-home directory?11:46
SiilwynI got it working with a home directory but for some reason it doesn't work if I change it to /srv/userWebsite11:46
Siilwyn(and yes root does own srv and 'userWebsite'11:46
mnms_how do you backup your servers, with dd ?:)12:19
lordievaderDirvish here.12:19
mnms_lordievader: thanks12:25
jikuhey guys..14:13
jikuhaving a weird issue here..14:13
jikui just did an installation of php5-fpm with apache.14:14
jikuwas all fine until when i need to restart the service php5-fpm with /etc/init.d/php5-fpm or reload php5-fpm i get the error..14:15
jikuhttp://pastebin.com/D7QqMAV614:15
jikumany such lines like these.14:15
jikui have to kill the process with PID 2wice 3rice..14:16
jikuand then start php5-fpm14:16
jikuis it a bug or something which is per the design ?14:16
geniiI'm pretty sure it uses Upstart and not sysvinit14:18
=== frickler_ is now known as frickler
jikuyep it does.. but reload php5-fpm is giving the same erro..14:22
jikui believe start, reload, stop --> all are upstart related.14:22
patdk-wkservice php5-fpm restart :)14:27
F1skrIs it possible to get a ubuntu server root file system like what is available here for the cloudimages: https://cloud-images.ubuntu.com/vivid/current/ I'm looking for something without the cloud-init stuff because it doesn't work if I spawn it in a systemd-nspawn container14:38
catphishF1skr: personally i just do a quick install from dvd in kvm, then extract the filesystem with guestfish14:38
F1skrthat seems very complicated14:39
F1skrthen I could just get the cloud image and remove the stuff I don't need14:40
catphishnormally reasons for wanting a raw image *are* complicated :)14:40
F1skrI donøt know, I simply want to spawn a systemd-nspawn container with an ubuntu server instance14:43
ogra_F1skr, http://cdimage.ubuntu.com/ubuntu-core/trusty/daily/current/ (its a bit more cut down than an actual ubuntu-server though)14:44
catphishi've never seen those before, awesome14:45
ogra_dont mix them up with the daily-preinstalled images though ... that is ubuntu snappy core ... nothing you can use OOTB14:46
ogra_(and i guess the better place would actuallly be http://cdimage.ubuntu.com/ubuntu-core/releases/)14:47
F1skrwhat does OOTB stand for? :)14:47
ogra_out of the box14:47
F1skrthat seems perfect for what I'm doing! do you happen to know the default root password in those images?14:50
ogra_there is none14:50
ogra_it isnt set up at all14:50
ogra_it is pretty much mostly for using chroots14:50
ogra_the definition for the non..snappy ubuntu core images is: just enough OS to run apt14:51
ogra_(in a chroot that is)14:51
F1skror a container as I'm doing now :)14:52
ogra_:)14:52
ogra_note, you might need to copy resolv.conf in place14:52
F1skrsystemd handles that for me :)14:53
F1skrI also needed something where the network service was disabled (in order to let systemd handle it) this is perfect14:57
F1skrthanks ogra_ !14:57
ogra_good luck14:59
ostei am seeing something strange using nginx, when i change the server root symlink, i get “No input file specified” for a few minutes then the new changes are properly displayed15:26
ostei am using php fpm with fastcgi - and tried setting SCRIPT_FILENAME to a few different things with no luck15:26
tewardoste: there's a document for that somewhere...15:31
tewardoste: what version of nginx?15:31
tewardand what Ubuntu15:31
ostenginx/1.4.615:31
osteUbuntu 14.415:31
ostewonder if i should upgrade nginx15:32
osteupdate*15:32
ostePHP 5.5.9-1ubuntu4.1115:33
=== Tm_Tr is now known as Guest26968
=== markthomas|away is now known as markthomas
=== cipi is now known as CIPi
=== sarnold_ is now known as sarnold
jwitkoHey guys I have a server where my iscsi connections to my SAN stopped working.  I have identical setups in the same network where the iscsi connection is still just fine to the SAN.  Multipath -ll gives me the error "Error: : Inappropriate ioctl for device"   can anyone help ?16:40
=== wendar_ is now known as wendar
tewardrbasak: i plan to bash my head against nginx today/tomorrow, but FeatureFreeze is in, and nginx merging will have new features, do you just want me to wait to X series?16:47
rbasakteward: I can't approve a feature freeze exception if that's what you're asking?16:49
tewardrbasak: no, i'm asking if it's worth it - there's 3 new features in ntinx 1.9.416:49
tewardi know you can't approve ;P16:49
rbasakHmm16:49
tewardsubstantial bugfixes though16:49
tewardhttp://nginx.org/en/CHANGES16:49
tewardit'll be available in Debian tomorrow (uploaded only today, not sure if it's built there yet :P)16:50
rbasakI'm not sure if I had considered Wily feature freeze when we talked about plans for 1.916:50
tewardi don't think we had, but given FeatureFreeze has passed I'm OK leaving it alone16:50
tewardand saying "Wily people: use the ppa not the repos, we aren't updating because freeze"16:50
rbasakYeah I think you'll have to leave it alone, unless you want to do the work backporting bugfixes from upstream hg16:50
teward(the ppa gets a resync with debian tomorrow xD)16:51
tewardheheh, yeah, no thanks16:51
rbasakOK.16:51
rbasakI'm happy for you to leave Wily nginx alone then and focus on the PPA.16:51
tewardwheee rmadison lag16:51
tewardthat's the general idea now16:52
tewardsince college started back up and all :)16:52
tewardrbasak: PPAs get updated either way, so long as Debian's version builds, I just 'borrow' the package from there :P16:54
rbasakSounds good16:54
catphishis there something i can append to the installer kernel / initrd to give me a root shell?16:55
catphishinit=/bin/bash doesn't work because there doesn't seem to be a /bin/bash in the initrd16:55
rmc3Haven't been able to find an answer in the documentation about this: I see that the Ubuntu installer supports using Anaconda-style kickstarts. Is the kssendmac kernel parameter supported for sending the MAC of the interface that's in use to the server providing the kickstart as a parameter?17:02
sarnoldcatphish: /bin/sh ought to be a symlink to /bin/dash17:06
catphishsarnold: thanks, i just actually just this second tried that and it worked :)17:06
sarnold\o/17:06
catphishis there a way to run the disk detection procedure from there?17:07
sarnold"disk detection procedure"?17:07
sarnolddo you mean udevadm trigger or whatever that is?17:07
catphishthe installer does something that loads modules for detected hardware17:08
catphishi just want to load the modules for my raid array so i can mount something17:08
catphishi can't run the normal recovery, because its network detection routine breaks my IPMI connection :)17:10
catphishi suppose a simpler question is: where are the .ko files in the installer?17:10
jwitkoHey guys I have a server where my iscsi connections to my SAN stopped working.  I have identical setups in the same network where the iscsi connection is still just fine to the SAN.  Multipath -ll gives me the error "Error: : Inappropriate ioctl for device"   can anyone help me trouble-shoot iscsi issues?17:15
jwitkodoing a scan results in the error "no session found"17:15
md50Does anyone know if there is an ubuntu server installer that includes the sfc (Solarflare) nic drivers?  I tried 15.10 and 14.04.3, but no luck.17:57
md50only mention of sfc I saw was this bug, but I'm not sure if this was iso or pxe specific.  I tried the isos https://bugs.launchpad.net/ubuntu/+source/linux/+bug/148149018:01
ubottuLaunchpad bug 1481490 in linux (Ubuntu Vivid) "Add sfc to nic-modules udeb" [Undecided,Fix committed]18:01
=== Luke_ is now known as Luke
sarnoldmd50: how are you determining if the driver is supported?18:57
sarnoldmd50: 'apt-file search sfc.ko' on my 14.04 LTS system shows kernel modules for it for both the 3.13 and the 3.16 based kernel lines18:57
md50The daily iso builds of the installer don't recognize the cards, and sfc.ko doesn't appear to be on the media.  However, after installing without a network, the kernel that boots contains the sfc module.  So it's a crappy installer, but things eventually work19:02
md50I won't be able to mass install anything automatically without the driver (e.g. pxe install) but I can hack around it for now19:03
TJ-if you're doing PXE based installs you can slipstream the driver into the initrd that is served to the host via TFTP19:13
md50yeah, I guess I'll have to add it as part of my PXE boot, but I was hoping it would just work.19:19
TJ-md50: I usually have 2 PXE entries per ISO... 1 standard and one unpacked on the server file-system so it can be customised easily19:36
=== JanC_ is now known as JanC
eatingthenightim on ubuntu server 14.04. When i type hostname i get the hostname as expected. When i type hostname --help it shows all the flags. One of them being -i to get the IP address. when i type hostname -i it says name or service unknown.22:05
eatingthenightanyone ever ran into this? it's driving me crazy. It seems to work on ubuntu 14.04 02 but not on the original 14.04 release/22:06
eatingthenightI can't find any bugs about it as well.22:06
eatingthenightBut i have tested it on multiple servers all with the same results.22:06
capricorn_oneyour RR in DNS is missing22:06
eatingthenightsorry what is RR? capricorn_one22:07
capricorn_oneReverse Record22:07
capricorn_oneman apge says22:07
capricorn_one       -i, --ip-address22:07
capricorn_one              Display the network address(es) of the host name. Note that this works only if the host name can be resolved. Avoid  using22:07
capricorn_one              this option; use hostname --all-ip-addresses instead.22:07
eatingthenighto i see. I was using -i because -I adds an extra space at the end of the output22:08
eatingthenighti guess i'll just strip it out since that is more reliable22:08
eatingthenightcapricorn_one: Thank you very much22:08
capricorn_onewhen in doubt use man pages22:09
eatingthenightcapricorn_one: Yeah i need to get better about that one.22:10
sarnoldeatingthenight: why are you looking for the local IP? perhaps there's a better way to do what you're doing22:43
=== Lcawte is now known as Lcawte|Away
=== CIPi is now known as cipi

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