/srv/irclogs.ubuntu.com/2014/09/03/#ubuntu-server.txt

jdstrandGuest79575: there were changes, but not something that you would need to do differently to have it enabled on boot00:09
jdstrandGuest79575: can you file a bug at https://bugs.launchpad.net/ubuntu/+source/ufw/+filebug ?00:10
Guest79575jdstrand: if I can narrow something down I will file a bug, am trying to reproduce locally with vagrant00:11
jdstrandok, thanks00:11
wmphello, i very need help with networking00:11
Guest79575jdstrand: it's wierd and may have to do with EC2 so I don't want to jump to conclusions00:11
wmpbecouse my server is donw00:11
wmphttp://pastebin.com/TPeqPELc - i thying to run this bridge00:11
wmpserer works good on normal eth0 inet static00:11
wmpbut when i trying make bridge, it down...00:12
jdstrandGuest79575: I was wondering if the security groups might be interfering (or some other software in the guest)00:12
Guest79575jdstrand: yea I wondered that as well - but they didn't change in the sequence of events I outlined before00:12
Guest79575i.e. they always allowed TCP port 22 from anywhere00:12
* jdstrand nods00:13
=== Lcawte is now known as Lcawte|Away
Guest79575jdstrand: FWIW I just reproduced the issue locally with vagrant VM00:19
jdstrandGuest79575: ok. can you file a bug with steps to reproduce?00:32
Guest79575jdstrand: I'm going to retry with Precise first but if that works then yes00:33
jdstrandthanks00:33
=== peter is now known as Guest71743
dustinspringmanserver 14.04... when a client disconnects from pptp on the server, cpu spikes to 100% for that task.... any thoughts on how to stop this?01:51
Patrickdkrestart pptp on 100% cpu usage?02:22
Patrickdk:) not sure, so haven't used pptp for so long02:22
dustinspringmanPatrickdk: all I have to do to remedy the issue is kill the pppd task in top... but whenever a client disconnects.. boom.. 99% load until its kilt02:30
Patrickdkheh? you said pptp would go 100%02:32
Patrickdknow it's pppd?02:32
Patrickdkthat sounds easy to fix02:32
dustinspringmanit's pppd in the task list on top.. but it's a pptp server i'm operating.. the client is a pptp client that disconnects and sends pppd into a spiral02:33
Patrickdkyes, sounds like a pppd problem02:33
Patrickdkso look for a pppd solution02:34
dustinspringmanbeen doing that... there's all sorts of random stuff on the web.. logs don't say much..02:34
Patrickdkwhat does strace say?02:34
dustinspringmanI've been chasing ghosts through the pptp.conf for a while.. got rid of half the problem.. (it was doing the same thing when users connected, but i disabled the logwtch in pptpd.conf and that went away)..02:35
dustinspringmanhmm.. strace looks interesting... i'll see what it has to offer02:36
dustinspringmanwow, I have no idea how to use this! XD02:37
Patrickdkyou wouldn't :)02:37
Patrickdka programmer would02:37
Patrickdkor atleast someone that understood programming02:37
Patrickdkbut if it's using 100% cpu, it should give a good hint02:37
dustinspringmanuh oh... i've gotten it stuck in some sort of gobbley gook and can't CTRL+C out of it.. XD02:42
Patrickdkheh?02:43
Patrickdkcontrol-c should work02:43
Patrickdkbut your terminal is probably screwed up02:43
Patrickdkafter control-c, try typing in reset, and press enter02:43
dustinspringman}'}"}(}"�}*~~�}#�!}!}-} }4}"}&} } } } }%}&�)}2}=}'}"}(}"�}*~~�}#�!}!}-} }4}"}&} } } } }%}&�)}2}=}'}"}(}"�}*~02:43
dustinspringmanthat's all I'm getting.. =/02:43
dustinspringmanevery now and again I get this: socket(PF_LOCAL, SOCK_DGRAM|SOCK_CLOEXEC, 0) = 702:44
dustinspringmanconnect(7, {sa_family=AF_LOCAL, sun_path="/dev/log"}, 110) = -1 ENOENT (No such file or directory)02:44
dustinspringmanclose(7)                                = 002:44
dustinspringmangettimeofday({1409712235, 73661}, NULL) = 002:44
dustinspringmanfcntl(6, F_SETLKW, {type=F_WRLCK, whence=SEEK_SET, start=316, len=1}) = 002:44
Patrickdkhow exactly did you run strace?02:44
Patrickdkyou didn't strace -p *the pid of the pppd that is using 100% cpu*02:44
dustinspringmansudo strace pppd02:45
Patrickdkthat is not going work02:46
Patrickdkthat is just running pppd on the console02:46
Patrickdkthat is suppost to happen02:46
dustinspringmanoh,. so, how do I get out of this?02:46
Patrickdkwell, the problem is you need to get out of pppd :)02:46
Patrickdkprobably control-d02:46
dustinspringmanctrl+d /anything does nothing..02:47
dustinspringmanopen another term and kill something?02:47
Patrickdkcontrol-z ?02:47
Patrickdksure :)02:47
Patrickdkonly option now, that I can think of02:47
dustinspringmanhah.. so... now what should I kill!? I don't see strace anywhere... or pppd...02:48
Patrickdkps ax02:48
dustinspringmanthat got it!02:50
dustinspringmanwhew!02:50
pmatulisdustinspringman: still banging away at it?03:12
dustinspringmanpmatulis: yea man.... I've isolated the issue.. just can't seem to find the smoking gun...03:13
dustinspringmanpmatulis: I got rid of the connection logging that was eating CPU, now I just have the "disconnect of pptp client causes pppd task to spike CPU" issue...03:14
dustinspringmanpmatulis: I've read every thread on the web I can find...03:14
dustinspringmanpmatulis: some talked of "address loops" and such, but I've checked on that, doesn't appear to be the issue... its got to be something in the pptpd.conf file that is trying to "do something" when the pptp connection to a client drops.. .but what that "doing" is, is eluding me..03:15
Patrickdkwhat that is, is what strace will show you03:21
Patrickdkstart a connection03:21
Patrickdkattach strace to it03:21
Patrickdkthen disconnect03:21
Patrickdkand strace will show you what happens03:22
Patrickdkthen you just have to figure out the cause vs the effect03:22
dustinspringmanPatrickdk: i'll dig into that a bit more and see what I can come up with.. The syntax for strace is cray cray03:24
pmatulisdustinspringman: 'strace -f -o <output_file.strace> -p <pid of your process>'03:38
Patrickdkmaybe, but probably not needed yet, -s 4096, or so03:38
dustinspringmanpmatulis: Patrickdk: I'll give that a shot in a bit.03:46
pmatulisdustinspringman: the perf utility will probably help you more in your situation04:06
pmatulissudo apt-get install linux-tools-generi; <start your process>; sudo perf record -g -p <pid>; reproduce problem; <stop perf>; sudo perf report -g04:10
pmatulislinux-tools-generic*04:10
sarnoldpmatulis: ooh. thanks for quick tutorial :)04:14
pmatulissarnold: lol04:14
sarnoldpmatulis: no, seriously, "learn about perf" has been on my todo list for bloody ever04:15
sarnoldnow I have it up and running and am impressed :)04:15
pmatulissarnold: well now you know :)04:15
pmatulisobviously it generates a default named output file (perf.data) that can be specified with '-o'04:17
pmatulisthere's a lot more but that is the basic stuff.  and i'm no expert04:18
sarnoldpmatulis :D04:20
Patrickdkhaven't heard of or used perf before04:21
Patrickdkbut since I know the kernel, and programming good, strace makes perfect sense to me :)04:21
sarnoldPatrickdk: strace is awesome but there's a lot it misses.. it doesn't show functions inside a program, which can be really helpful, and it doesn't show functions inside the kernel, which can be helpful, and it is a bit .. blunt .. when trying to find program hotspots04:24
Patrickdkah ya, it's not a profiler04:25
Patrickdkit's a hmm, what is the word04:25
Patrickdksyscall dumper04:26
user123321Please help, my machine again crashed at 10.10 display time with no response, recovered with ctrl+alt+prtsc+R+I+U+S+B. Kernel log around that time: http://pastebin.com/dkS92jui04:50
sarnolduser123321: anything before that?04:51
Valduarehi guys i just upgraded from 12.04 to 14.04 and the system is bad….. seems to be issue with perl. I cant install anything complains of perl stuff… looking around there are no files in /usr/lib/perl04:52
Valduarehow can I fix this04:52
user123321sarnold,  09:05:00 <---- it's the previous one before the first log entry I posted.04:52
user123321seems one hour before.04:52
sarnolduser123321: wait, is that "ubuntu 10.10" aka "maverick" aka "end of lifed over two years ago"?  Or is that just saying at 10:10 it hung....04:53
sarnolduser123321: ah, okay04:54
user123321sarnold, Oh I mean, it hanged at 10.10 display time :D04:54
sarnolduser123321: okay, good... I just googled for that memory address you found and people running 10.10 were complaining about it... I got worried that you were two years behind on security updates :)04:54
user123321oh lol04:55
sarnoldinteresting if true https://bbs.archlinux.org/viewtopic.php?pid=135083604:56
user123321interesting indeed! hmm.04:58
sarnoldand definitely interesting:  https://bugzilla.kernel.org/show_bug.cgi?id=16661  :)04:58
uvirtbotsarnold: Error: Could not parse XML returned by bugzilla.kernel.org: HTTP Error 404: Not Found04:58
sarnolduvirtbot: the downside is... the kernel.org bugzilla gives me the feeling that this message is probably unrelated to your actual lockup, or perhaps a symptom of a different underlying problem..04:59
uvirtbotsarnold: Error: "the" is not a valid command.04:59
sarnoldsigh you stupid virtbot04:59
user123321sarnold,  It be same in Ubuntu and LUbuntu right? Because I'm in LUbuntu now.05:00
user123321It should be*05:00
sarnolduser123321: so, here's a long-shot theory. it's not great. perhaps there's some electrical issue with your NIC; the connection bounced at about the same time, perhaps a lose wire wiggled or the machine moved around slightly causing electrical connections to come and go and wedge the machine05:01
sarnolduser123321: (My dad's dell had this problem, the gigabit card we bought wouldn't work, he's stuck with onboard 100mbit ethernet as a result..)05:01
user123321sarnold, Oh the ethernet, I restarted my router a while before the crash. That's why eth0 restart is displayed in logs.05:02
sarnolddang :)05:02
Valduarehow can I install POSIX.pm via command line05:03
sarnoldwell, that's probably good. it'd be replace motherboard or something if it were broken..05:03
sarnoldValduare: could be horrible. try "apt-get install perl-base" and see if you get lucky...05:04
sarnolduser123321: you're probably going to have to try to collect more data next time it happens. have a ps auxw running or use a serial console or something similar...05:04
sarnoldwell, you did get the sysrq messages, it was alive enough for that..05:04
Valduareapt dosnt want to work without perl fully functional05:04
eagles0513875hey guys is there a way i can see all the apache redirects i have on my system05:04
user123321sarnold, Yeah, how to get more data? Are there any logs I can check now? :D05:05
sarnoldValduare: dang. okay, look in /var/cache/apt/archives for perl* debs05:05
ValduareCan't locate POSIX.pm in @INC05:05
sarnolduser123321: maybe next time it hangs, use sysrq to dump task info and stack traces05:05
Valduareya I have the deb in the cache05:06
sarnoldValduare: sweet; try do dpkg --install the thing05:06
user123321sarnold, any idea of the exact buttons? https://www.kernel.org/doc/Documentation/sysrq.txt <--- this shows a lot I bet05:07
sarnolduser123321: https://www.kernel.org/doc/Documentation/sysrq.txt05:07
sarnoldhahaha05:07
user123321lol05:07
sarnolduser123321: l m d w   probably the starting point05:07
Valduareah progress05:09
user123321sarnold, cool05:09
Valduaresarnold next roadblock Perl API version v5.14.0 of Socket does not match v5.18.0 at /usr/share/perl/5.18/XSLoader.pm line 92.05:09
sarnoldValduare: dang, that one's harder. Socket.pm from perl-base should report v5.18.0 .. and Socket.pm from e.g. libsocket-perl isn't installed on my laptop, seems unlikely to be the package that you need to upgrade..05:13
Valduarehmm05:14
sarnoldValduare: hrm, now that I look into a bit .. why 5.18 and not 5.18.2??05:16
sarnoldValduare: ... did you say "yes" to CPAN's offer to install a new perl interpreter or something similar? :)05:17
Valduarehavnt used cpan05:17
sarnoldgood good, that always drove me up the wall05:17
sarnoldbut still, that's the craziest..05:18
Valduarethis is a 12.04 system freshly upgraded to 14.04 if that helps05:18
Valduarehad the grub issue on first boot05:19
Valduarehad to use boot-rescue05:19
sarnoldyou may be able to better understand my confusion .. well, eventually .. if you take a look at the publishing history here: https://launchpad.net/ubuntu/+source/perl/+publishinghistory05:21
sarnoldI just don't see a 5.18 or 5.18.0 package in the list.05:21
Valduarebtw I wonder how the MaaS guys are doing05:25
Valduarehear of any progress with maas and arm devices?05:25
sarnoldValduare: hmm, no, I haven't, all my maas use has been x86 (well, virtualized x86 at that..)05:26
ikoniaeagles0513875: look in the config file - that will show you all the redirects05:26
Valduareah how does it work for virtualized stuff05:26
ikoniaeagles0513875: the ones that won't be shown are ones that are handled by the application, not the webserver05:26
sarnoldValduare: they've got #maas -- it might be worth poking your head in if you're curious :)05:26
eagles0513875you mean the global apache configuration ikonia05:27
sarnoldValduare: well, it worked well enough for my testing -- it might not be something you'd actually want to -use- of course05:27
ValduareI have a cobbled together setup here lol05:27
Valduarefreenas box serving iscsi to a gigabyte gae350n mobo with 16 gigis of ram running esxi atm and another host box an old dell optiplex 745 running esxi as well  all with ubuntu vm’s05:28
sarnoldooh fun :)05:28
sarnolddoes the esxi do all the iscsi itself or do your VMs do the iscsi initiating?05:29
Valduareit handles it05:30
Valduarevm’s just act like vms if they were on virtualbox even05:31
sarnoldI've heard linux's iscsi target is .. not great. It might be best to have esxi do it :)05:32
ValduareI am interested in other solutions eventually05:34
Valduareso far this setup has been fairly thoughtless05:34
ValduareI use smoothwall 3   as a virtual router05:34
user123321sarnold, where are the logs from " l m d w" going to save?05:34
Valduareso make a vm, install smoothwall 3 on there just 128 megs of ram or so. then give it two virtual nics one for the public ip and one for internal traffic05:35
eagles0513875ikonia: reason i asked that I have round cube installed and something has changed where its redirecting to the main website instead of roundcube login page05:35
eagles0513875i followed https://help.ubuntu.com/community/Roundcube05:35
sarnolduser123321: syslog -- you may need to change the "log level" if they don't get saved there..05:35
eagles0513875to set it up and it worked05:35
user123321cool05:36
sarnoldValduare: ha :)05:36
Valduarethen I pile as many vm’s as I want behind each smoothwall router lol05:37
=== liam_ is now known as Guest81233
ikoniaeagles0513875: it's possible your upgrade to 14.04 and thus apache 2.4 has changed the way it's redirection works06:34
ikoniaeagles0513875: first thing is the redirection apache or the application06:34
eagles0513875wont be suprised there.06:34
eagles0513875i think there as well were config issues with round cube after the upgrade im working around them bit by bit06:35
ikoniais the redirect the webserver or the application ?06:37
eagles0513875application06:42
ikoniaeagles0513875: then it's unlilkley the webserver has anything to do with it06:43
eagles0513875im looking in htaccess06:43
ikoniadoes htaccess do redirects for you ?06:43
eagles0513875in this case it did ikonia :) now just have to fix imap issue06:51
lordievaderGood morning.07:36
salih-emingood morning07:36
=== wendar_ is now known as wendar
cfhowlettlordievader, ehhh, no.  afternoon in Beijing.  How about "greetings!"07:37
lordievaderHey salih-emin, how are you?07:37
lordievadercfhowlett: Good afternoon to you then ;) How are you doing?07:37
salih-eminfine thanks !07:37
cfhowlettlordievader, no complaints07:37
lordievadersalih-emin, cfhowlett: Good to hear :)07:40
eagles0513875hey lordievader07:42
eagles0513875hows it going07:42
lordievadereagles0513875: Pretty good here ;)07:46
lordievaderHow are you eagles0513875 ?07:46
eagles0513875not bad chugging along getting things donw07:46
eagles0513875done07:46
=== darkness is now known as Guest8986
=== a1berto_ is now known as a1berto
user123321Machine got frozen again during boot with nothing on screen. Kernel log from the start of boot to the emergency restart: http://pastebin.com/A5V8yWjS10:42
user123321Any ideas? Thanks :)10:42
lordievaderuser123321: Run memtest: Corrupted low memory at ffff880000006598 (6598 phys)10:45
user123321lordievader, shall I just run memtest in terminal?10:46
lordievaderuser123321: No, reboot and select memtest from the grub menu.10:48
user123321Ah ok,10:50
user123321brb10:50
user123321lordievader, machine froze two consecutive times when trying recovery mode. I'm back with normal boot now *sigh*10:58
salih-emindo you have a bootable media ?10:59
salih-eminyou can try memtest from a live media10:59
user123321salih-emin, I could prepare a bootable media. Just to avoid confusion, is memtest supposed to be used for testing systems that cannot be booted at all? Or can it identify rare crashes that has happened to me? For example, I'm logged in now with the machine.11:00
salih-eminuser123321, memtest is for testing RAM for corruptions (hardware issues)11:01
salih-eminnot the actual OS11:01
user123321oh I see11:02
user123321I'd try it sometime, coz I need to prepare a bootable media.11:02
salih-eminyes if you encounter errors during the memtest you should consider replasing your RAM memory11:03
user123321salih-emin, But i don't get crashes when I use Windows 7 :|11:04
user123321not like this fatal :/11:05
=== hxm is now known as Guest47546
Andy80hi, I'm trying to make mcrypt php extension work on my Ubuntu 14.04. I'm aware of this bug https://bugs.launchpad.net/ubuntu/+source/php-mcrypt/+bug/1318021 but even after installing the plugin and checking that is enabled with php5-fpm -m | grep mcrypt still one of my wordpress plugin can't see it. Please not I'm using php5-fpm. Any idea?11:10
uvirtbotLaunchpad bug 1318021 in php-mcrypt "Not Find mcrypt.ini" [Undecided,Confirmed]11:10
lordievaderuser123321: It could be that the memory corruption is in a place that Windows does not address quickly.11:10
Andy80please also note that there is no /etc/php5/conf.d/ directory... there is /etc/php5/fpm/conf.d/11:11
lordievaderuser123321: Also, that the machine freezes upon booting the recovery mode is not a good sign.11:12
user123321lordievader, This link is surprising, and what do you think about its response?: https://bbs.archlinux.org/viewtopic.php?id=16344011:13
user123321He has the same laptop as mine, DV611:13
lordievaderuser123321: Ewaller probably knows what he is talking about ;)11:14
user123321haha11:15
=== Guest47546 is now known as HackeMate
cariboujamespage: has anything (i.e. SRU like) has been planned to take charm fixes from ./next into the existing stable release ?12:19
cariboujamespage: your document states the reason why it should be done, then stops there12:20
jamespagecaribou, nope - each fix would be assessed on a requirement and impact basis via a bug12:22
jamespageif you have fixes you want to see in the stable charms, they feel free to nominate stuff12:22
cariboujamespage: what if the fix is already coming from a bug ?12:22
jamespagecaribou, not sure I understand your question then12:23
cariboujamespage: I'm thinking of the known_hosts/authorized_keys fix I did for nova-cloud-controller/nova-compute charms12:23
jamespagecaribou, the change needs to be merged into /next and then proposed/merged into the trunk branch12:23
cariboujamespage: this fix is already into ./next (nova_cc one)12:23
jamespageok12:23
cariboujamespage: TL;DR : what do I need to do to get those fixes which are in ./next in the official charm :-)12:24
jamespagecaribou, cherry pick the commit from the next branch into the stable branch and propose it12:25
cariboujamespage: k12:25
abhishekhow can i scan new hard-drive without rebooting ?12:28
ikoniaabhishek: did it get detected as a udev event ?12:28
abhishekudev event ? u r talking about udev at startup12:29
ikoniaudev will create the device node used to access it12:29
ikoniaif you don't see a udev event (eg: your hardware does not support hot plug) you can't access it12:29
ikoniayou could try triggering udev, but it's unlikley to do anything if you don't have hotplug support12:30
abhishekhow can i do this ?12:30
ikoniadoes your hardware have hotplug support ?12:30
abhishekmay be I don't know ? I just created a lun in storage want to add in ubuntu-serrver12:31
abhishekI cant reboot this server this time12:31
ikonialun in storage ?12:31
ikoniaso this is a fibre array ?12:31
abhishekyes12:32
ikoniaok, you need to walk the fibre12:32
ikoniaand it shoul trigger a udev event12:32
dasjoeabhishek: find out how many scsi hosts your system has, then something like this: for x in $(seq 0 7); do echo "- - - " > /sys/class/scsi_host/host$x/scan; done12:32
ikoniayou send a "1" to an option in proc12:32
abhisheklet me try this12:33
abhishekI have four scsci host12:34
abhishekthis works thank u dasjoe and ikonia :)12:34
abhishekthank u very much12:34
caribousorry for the dumb question, but where do the stable openstack charms live (for precise) these days ?13:18
caribouI used to pick them at lp:charms/{nova-cloud-controller|nova-compute} a while back13:18
jrwrenhow do I use uvt-kvm to test utopic? uvt-simplestreams-libvirt query release=utopic returns nothing13:24
jamespagecaribou, either lp:charms/<charm-name> or lp:charms/trusty/<charm-name>13:26
jamespagewe keep the branche sin sync13:26
cariboujamespage: good, those are the ones I was expecting13:26
cariboujamespage: thanks13:26
rbasakjrwren: uvt-simplestreams-libvirt sync --source http://cloud-images.ubuntu.com/daily release=utopic arch=amd6413:29
rbasakjrwren: daily images don't get published in the released streams, except for alpha/beta etc.13:29
jrwrenthanks rbasak13:29
zulhallyn: http://people.canonical.com/~chucks/libvirt/13:59
hallynzul: ok, will get to that today14:06
zulhallyn: ack14:06
user123321sorry my internet was reset, did anyone say anything related to my system crash after I logged out?14:45
=== TDog_ is now known as TDog
Guest47177jdstrand: I was here asking about UFW changes between Precise and Trusty: turns out it was not related to firewall at all.  two other issues conspired to make us think thep problem was firewall related15:25
Guest47177first issue: we were rebooting like this: reboot now15:25
Guest47177second issue: NFS mount was hanging boot15:25
Guest47177this explains our first issue: http://askubuntu.com/questions/483670/what-causes-ssh-problems-after-rebooting-a-14-04-server15:26
user123321Is upgrading the linux kernel to a newer version than the official ubuntu released advisable?15:26
user123321release*15:26
jdstrandGuest47177: oh, interesting15:27
lordievaderuser123321: No.15:27
Guest47177so the machines weren't actually rebooting, and when we rebooted them via the EC2 console they'd hang on boot waiting for NFS mount15:27
jdstrandah15:27
cfhowlettuser123321, for the vast majority - inadvisable. BUT if you have the expertise ....15:27
jdstrandvery interesting15:27
jdstrandGuest47177: thanks for the update! :)15:28
Guest47177yep, didn't want to let that hang out there w/o explanation15:28
user123321Oh ok, because I got an advice from #linux to upgrade the kernel to check if it might solve my crash issues :D (crashes that result from possibly firmware bugs in my DV6 laptop)15:28
lordievaderuser123321: Have you already checked the memory?15:30
user123321lordievader, no bootable media yet :/15:31
lordievaderuser123321: Check that first. A new kernel will not fix hardware issues.15:31
user123321Oh, I thought a new kernel could "ignore" hardware issues ;)15:32
user123321like Windows does. :|15:32
=== keithzg_ is now known as keithzg
=== Lcawte|Away is now known as Lcawte
=== Adri2000_ is now known as Guest60909
=== SierraAR is now known as Sierra
=== danley_ is now known as danley
=== ming is now known as Guest50512
=== psivaa_ is now known as psivaa
=== broder_ is now known as broder
=== hachre_ is now known as hachre
=== specialkevin_ is now known as specialkevin
=== SP33D_ is now known as SP33D
=== kees_ is now known as kees
=== andol_ is now known as andol
=== jeremy_carroll__ is now known as jeremy_carroll
bananapieif I type 'cat /dev/zero > /dev/null & ' in bash, how many file descriptors did I just open ?17:53
bananapieit seems to me it would open 5, stdin, stdout, stderr as well as /dev/zero and /dev/null17:53
sarnoldbananapie: iirc bash hands out more filedescriptors to processes; best check /proc/pid/fd/17:54
bananapieSo, if I do a while [ true ] ; loop with the above code, I should get errors about to many processes open pretty quickly ?17:54
sarnoldbananapie: check ulimit -a to see your limits; my max user processes limit is 125759 -- enough to make the system mighty unhappy17:57
bananapieulimit -a says 1024, but I have a  process open with 1500 descriptors. it's not getting any errors.17:58
bananapieIt's running as root, is it possible that root is not ulimited ?17:58
sarnoldbananapie: the nproc limit is applied per-user... the filedescriptor limit is applied per-process. none of your processes will have more than a handful of descriptors open.17:59
bananapieok18:00
bananapieis a process and a thread different ?18:00
sarnoldand yeah root isn't subject to the rlimits in the first place -- that's CAP_SYS_ADMIN and CAP_SYS_RESOURCE18:00
sarnoldbananapie: that's complicated :)18:01
bananapie:P18:01
bananapieok. But if a process is running as root, it can open 10,000 files and ulimit doesn't care. Only the sysctl file limit right ?18:01
sarnoldcap_sys_admin means even /proc/sys/fs/file-max can be ignored18:02
bananapieCool.18:03
bananapiealso, brb. I have to go uncrash a server18:03
sarnoldhappy uncrashing :)18:03
bananapieI think I'll just reboot it. The servers is on it's knees right now. I looped 1000 times only. It's not a very powerful machine.18:05
sarnoldhahahaha18:06
sarnoldgood luck with the reboot :)18:06
bananapiepkill cat :)18:11
=== daker_ is now known as daker

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