[11:49] <_ruben> rbasak: would you happen to have any clue as what could be the cause of bug #1008385, or in which direction to look?
[11:50] <_ruben> this bug report is pretty much the only info i could find on this matter
[11:52] <rbasak> _ruben: I'm not sure, sorry. I know that processes can modify their process name. Perhaps apache is doing it somewhere and getting it wrong for some reason? It's the sort of thing that probably needs a deep dive.
[11:53] <rbasak> And not being able to reliably reproduce it makes it difficult for anyone who isn't affected, of course.
[11:55] <_ruben> rbasak: that's about as much as i know as well ;)
[11:56] <_ruben> just ran a test on a bunch of servers i happened to be logged in on, at first sight more are affected than not
[11:56] <rbasak> _ruben: an explanation and patches will be greatfully accepted :)
[11:56] <rbasak> err, gratefully
[11:57] <rbasak> That word seems wrong to me this afternoon for some reason
[11:57] <_ruben> hehe
[12:11] <_ruben> wonder what will happen if i install -virtual kernel on a "broken" -generic vm .. will let you know in a bit ;)
[12:32] <rbasak> It might not be kernel related at all.
[12:32] <rbasak> That does seem less like to me, though I wouldn't rule it out.
[12:37] <_ruben> yeah .. finally managed to test booting -virtual .. no change
[12:39] <rbasak> Next I'd look for areas in the source that change the process name, and if there aren't any obvious bugs there, install debug symbols and get going with gdb.
[12:39] <rbasak> But there might be a better way of course
[12:48] <_ruben>  i'll be *peeeep* .. i did a strace on a single worker (apache2 -X) .. no references to /usr/sbin/apach .. turns out: process name is ok
[12:49] <_ruben> rbasak: interesting ... apache2 -k start: OK .. apache2ctl start: FAIL
[12:54] <zul> roaksoax:  ping
[13:01] <rbasak> strace is unlikely to catch it IMHO, if it's a problem with the process name being wrong. IIRC, changing process name doesn't involve a system call.l
[13:01] <rbasak> being set wrong, that is
[13:05] <_ruben> rbasak: the output of strace didn't help, but it did show me something else: apache2 (the binary) isn't responsible, not directly atleast
[13:05] <_ruben> disecting apache2ctl now
[13:05] <_ruben> (files are identical on both "kind" of servers unfortunately)
[13:06] <_ruben> the script is rather simple though
[13:07] <_ruben> ah! ... for some reason apache2 doesn't do proepr stipping for its process name
[13:07] <_ruben> apache2 -k start => apache2 ... /usr/sbin/apache2 -k start => /usr/sbin/apach
[13:10] <stgraber> rbasak: changing process name is done through the prctl syscall, so it should show up in strace
[13:10] <_ruben> stgraber: good to know
[13:10] <rbasak> stgraber: ah. I thought one could just change argv[0]?
[13:11] <_ruben> prctl(PR_SET_NAME, 0x7f0a15811b80, 0, 0, 0) = 0
[13:11] <_ruben> I'm guessing 0x7... is a memory location
[13:16] <rbasak> _ruben: prctl(2) says that PR_SET_NAME is limited to 16 bytes, and should be null-terminated if shorter. "/usr/sbin/apach" is 15 bytes, and the 16th byte could be a null terminator.
[13:16] <rbasak> OTOH, "/usr/sbin/apache" is exactly 16 bytes, so does not need a null terminator.
[13:16] <rbasak> Perhaps there's a code path which limits the name to 15 bytes?
[13:17] <rbasak> (in order to account for the null terminator that isn't actually necessary)?
[13:18] <_ruben> it shouldnt be trying to set it to /usr/sbin/... in the first place
[13:18] <_ruben> the code doesn't call prctl directly though (not for this at least)
[13:20] <rbasak> http://web.archiveorange.com/archive/v/1XS1vcyaGboxxVcKF2sG
[13:20] <rbasak> Looks like prctl is limited to 16 characters, and overwriting argv still works but is limited to length of the original cmdline.
[13:21] <_ruben>     process->argc = *argc;
[13:21] <_ruben>     process->argv = *argv;
[13:21] <_ruben>     process->short_name = apr_filepath_name_get((*argv)[0]);
[13:21] <_ruben> looks kinda promissing
[13:22] <rbasak> apr probably stands for Apache Portable Runtime. A portability abstraction, IIRC.
[13:23] <JanC> yes, some other applications use APR too
[13:24] <_ruben> but still .. all this doesn't explain how same package version on different systems could behave diffferently .. if it were "a simple" bug that is ;)
[13:25] <rbasak> _ruben: so it depends on how you start the service? I usually use "service apache2 start" rather than calling apachectl or apache2 directly.
[13:25] <_ruben> rbasak: any service wrapper is likely to use apache2ctl
[13:25] <rbasak> _ruben: how many reporters are doing something different that causes it to start differently?
[13:26] <rbasak> _ruben: the only wrapper I'm aware of is /etc/init.d/apache2.
[13:26] <_ruben> rbasak: none probably .. i know i aint
[13:27] <rbasak> Could there be two different code paths in the init script?
[13:29] <_ruben> my curent suspect: a envvar called APACHE_HTTPD .. apache2ctl initializes it to /usr/sbin/apache2, but if you override it to apache2, it works as expected .. let me check on a "good" server
[13:32] <JanC> so, maybe a default config differs between some older and current package versions?
[13:33] <rbasak> How about the upgrade path? Is everyone installing from fresh, or have they upgraded from a previous version?
[13:34] <JanC> rbasak: from the bug report it seems the working ones probably upgraded from 12.04.1, the broken ones installed 12.04.2 directly
[13:34] <_ruben> upgrade or fresh appears to be the very difference between good and bad
[13:36]  * rbasak looks up the diff
[13:36] <_ruben> nope, the envvar isn't it .. changing that also changes the output in `top` .. `top` should list /usr/sbin/apache2 (that is, it does list it like that on both good and bad)
[13:37] <_ruben> which leads back to a bug in apache2's dirname stripping again .. which again would be rather unlikely
[13:37] <_ruben> i hate nasty bugs like this! :P
[13:38] <rbasak> I found them intruiging :)
[13:39] <_ruben> up to a point, sure ;)
[13:39] <rbasak> apachectl was modified in 2.2.22-1ubuntu1.3 though I can't immediately see that it caused this
[13:39] <rbasak> http://pastebin.ubuntu.com/5932802/
[13:39] <_ruben> the change isnt in apache2ctl directly .. i compared those between a good and bad server
[13:40] <_ruben> i dont have any gdb skills .. or i'd trace the stuff upto the prctl call
[13:41] <_ruben> heck, even that apr_ function goes beyond my C knowledge, and its pretty simple function
[13:43] <roaksoax> zul pong
[13:44] <rbasak> I don't see anything else relevant in the update diffs
[13:44] <zul> roaksoax:  available to do some reviews today?
[13:44] <rbasak> Most of the udpates have been in moduels
[13:45] <_ruben> wonder if it's some obscure libc related thing or something
[13:45] <roaksoax> zul nah :p
[13:46] <zul> roaksoax:  ok then
[13:46] <roaksoax> zul lol i am
[13:48] <rbasak> _ruben: but why would it be random? Or if it is only affecting upgraded systems, what is different about them to cause that to happen?
[13:48] <rbasak> _ruben: anything different in /etc/default/* or /etc/apache2/* between the two systems?
[13:50] <roaksoax> zul just send them my way and ill review them as soon as my network behaves right
[13:50] <zul> roaksoax:  ack
[13:58] <_ruben> rbasak: i dont think its random, just a change that only affects new installs, and not upgrades .. as for changes in those locations: haven't found anything obvious, yet
[14:44] <LLckfan> Does any1 know how to stop Shockwave flash from crashing? I have uninstalled both Flash and my browser (Chrome), installed both from a fresh download, and scanned my computer (come up clean). Everything is updated
[14:45] <^^rcaskey> where does $INTERFACES get set out in /etc/init.d/dhcp3-server?
[15:04] <Techdude1011> Is it possible to customize received snmp traps? I am using traptoemail
[15:28] <zul> Daviey:  fyi https://bugs.launchpad.net/ubuntu/+source/ceilometer/+bug/1206967
[15:29] <mikeey> does samba4 require any additional configuration except of the share definitions?
[15:29] <mikeey> if it does, have anyone got a tutorial or any info regarding it?
[15:33] <Daviey> zul: the declared universe deps, are they covered by MIR?
[15:33] <zul> Daviey:  working on the universe deps MIR right now
[15:33] <Daviey> coolio
[15:47] <jkitchen> question: multipath doesn't seem to be reading the "defaults" stanza of my config file, anyone experienced this issue?
[15:48] <jkitchen> I'm trying to set the default path_grouping_policy to multibus
[15:49] <jkitchen> because I would really rather not have to manually put all of my wwids into the multipath.conf since they'll be pretty much constantly changing (virtualization environment)
[16:06] <chmac> I've got a machine on which mysql / apache do not start at boot. The /etc/init/mysql.conf contains a 'start on runlevel [2345]' line. How else can I debug the issue?
[16:07] <ikonia> chmac: try to start them manually - confirm there is no problem
[16:07] <chmac> ikonia: Yeah, they both start ok
[16:07] <ikonia> that seems odd then
[16:08] <ikonia> chmac: maybe worth putting a touch /tmp/testfile.mysql(or apache) at the start of the script and see if the scripts even get executed
[16:08] <chmac> ikonia: Ok, interesting, to see if maybe upstart is not getting started or something.
[16:09] <chmac> I'm a little out of my depth on this stuff. Thankfully ssh does start!
[16:09] <ikonia> chmac: yes, simple test to see if the scripts even get excuted or not
[16:09] <ikonia> chmac: then you know the problem is not the scripts, but that they aren't getting run
[16:09] <ikonia> chmac: or you know there is a problem with the scripts and you can take it from there
[16:10] <chmac> ikonia: The scripts are the stock, whatever was installed with the system
[16:10] <chmac> ikonia: I'll try that now and reboot. Thanks for the suggestion.
[16:10] <ikonia> chmac: yes, but that doesn't mean there i a problem at boot time
[16:10] <chmac> ikonia: In the pre-start section?
[16:11] <chmac> Ok, so apache is in /etc/init.d/apache2 not /etc/init/
[16:13] <zul> roaksoax:  https://code.launchpad.net/~zulcss/ceilometer/refresh/+merge/177898
[16:19] <chmac_> This bouncer is on that server, so I lost my connection!
[16:19] <chmac_> ikonia: No files were touched :-(
[16:19] <chmac_> dpkg-reconfigure upstart ?
[16:19] <chmac_> I'm guessing!
[16:25] <ikonia> chmac: interesting so it didn't run
[16:25] <ikonia> chmac: so check what those scripts depend on
[16:25] <ikonia> chmac: or make sure those scripts actually do run when called directory
[16:27] <chmac> ikonia: The /etc/init.d/apache2 script I can execute directly like `/etc/init.d/apache2 status` and I get the expected output
[16:27] <chmac> ikonia: But /etc/init/mysql.conf is not an executable file...
[16:35] <chmac> ikonia: I gotta run, thanks for the help, I've made some progress in understanding. I'll get into it again another time. :-)
[17:17] <zul> adam_g/roaksoax: https://code.launchpad.net/~zulcss/neutron/refresh/+merge/177910
[17:48] <Shadowandlight> semi off topic.... what should the privileges look like for a mysql user who only needs to read/write access to there specific database? - i know this is too much (i think?) http://i.imgur.com/CBLdvMA.png
[17:48] <Shadowandlight> but im not sure what to pare it down to
[17:48] <Shadowandlight> do i uncheck everything in "administration"?
[17:52] <ikonia> chmac: you still there ?
[17:55] <roaksoax> Madkiss: howdy! so just wanted to let you know that I have a few patches for your packaging. I've also packaged 'dlm' and dropped redhat-cluster from Ubuntu
[17:55] <roaksoax> Madkiss: so this means we now fully support pacemaker + corosync 2.0 and bye bye cman
[17:56] <Madkiss> cool.
[17:57] <Madkiss> i'm hunting a bug right now that affects pacemaker.
[17:57] <roaksoax> Madkiss: i'll also be forwarding dlm packaging to you by EOW
[17:57] <Madkiss> wonderful, thanks
[18:11] <vedic> I have installed pgbouncer using dpkg. When I try to start it (/etc/init.d/pgbouncer start) there is no output. Even I don't see the log file created into /var/log/postgres/pgbouncer.log . How to know what is the problem?
[18:57] <chmac> ikonia: Was offline, but my bouncer was on, so I'm seeing your message now… :-)
[19:02] <Peryton272> I'm having a problem installing Ubuntu Server to my Dell PowerEdge 2600. Is this the right place to be?
[19:04] <sarnold> Peryton272: sure
[19:04] <Peryton272> so I boot from the disk that has the .iso on it, but when it boots my server no longer recognizes my USB keyboard so I cant even select a language or anything.
[19:08] <Peryton272> the keyboard works fine in the Dell bios but as soon as I try to boot from the disk that has Ubuntu on it, the keyboard stops working
[19:09] <sarnold> Peryton272: some BIOSes have a setting for how to handle usb keyboards and mice; it's intended to help along OSes without USB support, like many win95, win2k releases, but if you fiddle with it, you might get lucky?
[19:09] <Peryton272> alright I'll see if I can find something
[19:17] <Peryton272> sarnold: I got it to work. Thank you!
[19:21] <sarnold> Peryton272: cool, what was it?
[19:21] <Peryton272> just what you said. I had to turn on the bios support for USB devices
[19:22] <sarnold> Peryton272: crazy. that shouldn't have been needed since 2.0 or 2.2 kernels, I forget which it's been long enough... anyway. hooray. :)
[19:26] <Peryton272> sarnold: now it's saying "This kernel requires an x86-64 CPU, but only detected an i686 CPU."
[19:27] <Peryton272> sarnold: What does that mean? lol I haven't done anything with servers before
[19:27] <sarnold> Peryton272: there's two images for x86 systems, i386 and amd64. the amd64 ones will require the x86-64 CPUs, but it sounds like you've got a machine that'll require the i386 image instead
[19:29] <Peryton272> sarnold: oh ok. Could you point me in the right direction to get that?
[19:30] <sarnold> Peryton272: http://www.ubuntu.com/download/server   note the 'choose your flavour' drop down, you'll want the 32 bit option
[19:31] <Peryton272> sarnold: awesome thank you
[19:33] <ikonia> chmac: there ?
[19:37] <caraconan> Hi there. How can disable this apparmor filter? http://paste.org/66396 It's a VM inside a KVM hypervisor. Thanks
[19:40] <caraconan> Just removing /etc/apparmor.d/libvirt/libvirt-VM-UUID files ?
[19:41] <caraconan> Looks like is performing a: sudo /etc/init.d/apparmor teardown
[19:43] <sarnold> wow, running without apparmor? brave man
[20:03] <patdk-wk> sarnold, apt-get remove apparmor
[20:04] <goddard> i am trying to change apache document route for default to /home/goddard/Web/
[20:04] <goddard> but i am getting 403 error
[20:06] <patdk-wk> goddard, heh?
[20:07] <goddard> patdk-wk: i am trying to change the default apache document route from /var/www/
[20:07] <goddard> to /home/goddard/Web/
[20:08] <patdk-wk> goddard, ya, but apparmor doesn't block that
[20:08] <patdk-wk> what version of ubuntu?
[20:08] <goddard> 13.04
[20:09] <patdk-wk> maybe that is why, I'm in 12.04 as my servers stick to lts
[20:10] <patdk-wk> and there is no policy to block that
[20:10] <genii> Why not use userdir with UserDir Web instead of UserDir public_html, and Directory /home/*/Web>  instead of  Directory /home/*/public_html>   ... in the userdir.conf file
[20:10] <jdstrand> apache is not confined by apparmor by default
[20:10] <patdk-wk> genii doesn't matter
[20:11] <goddard> just a test server
[20:11]  * genii sips and ponders
[20:11] <goddard> i didnt wanna do anything complicated
[20:24] <jose> Hey, guys! I don't konw if you can give me a hand with my DNS server in Ubuntu, which appearently doesn't work (using bind9)
[20:41] <jkitchen> I figured it out. apparently the 'device' defaults override my explicit defaults in multipath.conf, so I just added a devices/device stanza and set what I wanted. works great!
[20:45] <jose> oh, nvm. Looks like I was missing some reverse record. thanks anyways!
[21:27] <LLckfan> Does any1 know how to stop Shockwave flash from crashing? I have uninstalled both Flash and my browser (Chrome), installed both from a fresh download, and scanned my computer (come up clean). Everything is updated
[21:28] <sarnold> LLckfan: file a bug report with adobe?
[21:30] <LLckfan> sarnold I have
[21:31] <LLckfan> Never got an answer
[21:33] <Patrickdk> weeee, watchs debian 3.0 install to an ssd
[22:00] <tux050> My postfix server keeps sending Error 501s.
[22:01] <tux050> I don't have domain, but I have the external ip address in the config file.
[22:01] <tux050> *an MX domain
[22:02] <ikonia> how are you sending mail to it though ?
[22:02] <tux050> from my gmail account
[22:02] <ikonia> how though if you don't have a domain ?
[22:02] <tux050> to root@my.ip.address
[22:03] <ikonia> is it getting there ?
[22:03] <tux050> no
[22:03] <tux050> but there are error logs
[22:03] <ikonia> so it is hitting the server but getting rejected ?
[22:03] <tux050> 501/invalid address from mail.blahblah.gmail.com
[22:03] <tux050> yes
[22:03] <tux050> What would I need to change in my config?
[22:04] <ikonia> invalid address suggests that server is not aware of that user/domain combination
[22:04] <tux050> ok.  how do I fix that?
[22:05] <ikonia> do a "HELO" test locally to root@ip
[22:05] <ikonia> see if it works
[22:06] <tux050> it seems to
[22:06] <tux050> it gives a 250
[22:06] <LLckfan> Does any1 know how to stop Shockwave flash from crashing? I have uninstalled both Flash and my browser (Chrome), installed both from a fresh download, and scanned my computer (come up clean). Everything is updated
[22:06] <ikonia> tux050: ahh are you using a nat ?
[22:07] <tux050> yeah
[22:07] <ikonia> I wonder if that's the issue
[22:07] <tux050> I was giving it it's external ip address though.
[22:07] <ikonia> eg: the server is listening on 192.168.24.10 - but when it gets rom external.ip it doesn't know about it
[22:07] <tux050> maybe
[22:09] <Patrickdk> this is all pointless
[22:09] <Patrickdk> you need to pastebin 'postconf -n' and the logs
[22:10] <ikonia> yup
[22:18] <tux050> would "myhostname" have anything to do with this?
[22:18] <tux050> should that be set to the external IP?
[22:19] <ikonia> that's what it sets up what it's listening on
[22:21] <tux050> also, what are aliases?
[22:22] <tux050> never mind, I got it
[22:38] <joshu> hi can someone please have a look at my upstart script because i'm not confident it's working as it should https://gist.github.com/anonymous/ed8b9b4e54bc66e079af
[22:39] <LLckfan> Hello
[22:39] <LLckfan> Does any1 know how to stop Shockwave flash from crashing? I have uninstalled both Flash and my browser (Chrome), installed both from a fresh download, and scanned my computer (come up clean). Everything is updated
[22:42] <Patrickdk> LLckfan, ubuntu-server doesn't have a gui, so flach and crome are unsupported in this channel
[22:44] <sarnold> joshu: you might want to use 'setuid test' rather than su -c .. test
[22:44] <sarnold> joshu: http://upstart.ubuntu.com/cookbook/#setuid
[22:44] <joshu> hi sarnold initially I tried that from the exact link, but it wouldn't work
[22:45] <joshu> don't know why
[22:45] <sarnold> joshu: did you get any error messages in the logs or on the screen?
[22:46] <joshu> not on the screen, but when I tried to scan the brscan-skey wasn't producing a file as it does with su -c or if I start it manually as user test with brscan-skey.
[22:46] <joshu> I can try again now
[22:47] <joshu> testing now will report back after I reboot and try to scan
[22:49] <joshu> sarnold ok just tested with setuid test, scanned fine, but no document was created in /home/test/brscan as it should
[22:51] <sarnold> joshu: any error messages in log files?
[22:51] <joshu> sarnold /ar/log/syslog ?
[22:52] <sarnold> joshu: check also /var/log/upstart/brscan-skey-daemon
[22:53] <sarnold> joshu: if there's nothing obvious there, try adding a chdir /home/test/brscan/  to the config: http://upstart.ubuntu.com/cookbook/#chdir
[22:53] <joshu> sarnold https://gist.github.com/anonymous/fc59b7641918b9f516c4
[22:54] <joshu> it's trying to write to /brscan and not /home/test/brscan
[22:54] <sarnold> joshu: cool, that might be fixed by the 'chdir' ..
[22:55] <joshu> ok going to test
[22:57] <sarnold> joshu: note that you don't have to reboot to test that -- you can just service brscan-skey-daemon restart
[22:59] <joshu> sarnold oh ok tried with chdir but it is still writing to /brscan
[22:59] <sarnold> joshu: what's the home directory for the test user?
[23:00] <joshu> "/home/test/brscan"
[23:00] <joshu> I tried chdir /home/test/brscan
[23:01] <sarnold> joshu: try just chdir /home/test, and try changing the home directory in /etc/passwd to just /home/test as well
[23:03] <joshu> ok let me try one sec
[23:03] <joshu> still same thing
[23:04] <joshu> tiff2pdf: Can't open output file /brscan/20130801010332_27436.pdf for writing.
[23:04] <joshu> I don't know what su -c does instead of setuid that the former works
[23:06] <sarnold> joshu: hrm, me neither. you could go back to su -c ..
[23:07] <joshu> ok is the script otherwise written ok? something I found odd is that if i kill the process using sudo kill <number> and then do sudo status brscan…. the process number is the same. Doesn't seem right to be if it respawns?