[01:17] <ruben231> hi guys, i have installed apache2 on my ubuntu 10.04 LTS but problem when i reboot it wont start automatically i need to induce it with command before it starts /etc/init.d/apache2 restart <<<<<-------------------any idea how to solve this..?
[01:28] <sarnold> ruben231: try "chkconfig apache2 on" -- stolen from http://manpages.ubuntu.com/manpages/precise/man8/chkconfig.8.html   ...
[03:43] <rocket> Looking for a regular expression channel
[03:47] <patdk-lap> #regexp ?
[09:21] <jamespage> rbasak, I hit the rebuild button last night for mongodb and its now fine
[09:21] <jamespage> rbasak, probably something transitory I suspect
[09:22] <rbasak> jamespage: OK, great. I'll leave it then.
[09:22] <rbasak> I still wish I had a way to get sbuild to match the buildd exactly for build dep resolution :-/
[11:05] <Ng> is cloud-init supposed to execute all the cloud config files in /etc/cloud/cloud.cfg.d?
[11:06] <Ng> as in, always? I'm firing up a test instance in a local kvm without any cloud infrastructure and the config file I've added seems to be ignored
[11:12] <jamespage> adam_g, some comments on the rabbitmq-server HA merge
[11:12] <jamespage> but hacluster and mysql LGTM to merged
[11:19] <lynxman> Ng: as far as I remember you need to trigger cloud-init metadata execution for that to be read, but my information might be clearly outdated
[11:19] <Ng> hmm
[11:21] <lynxman> Ng: cloud-init reads the metadata instructions and stores them in /var/lib/cloud somewhere, then that is used to execute the rest, should need to look at the code to see if that has changed
[12:05] <xnox> roaksoax: jamespage: sorry for missing the HA session.
[12:06] <xnox> roaksoax: jamespage: debian release team are unhappy about the current state of redhat-cluster, corosync, clvm and friends. To the point where they are aggresively trying to remove those from debian archive now, on the basis that all of those packages have not been updated in a while.
[12:06] <xnox> Surprisingly clvm is more or less up to date, it's the others that are lagging.
[12:06] <xnox> but it's the clvm that was first to be removed from wheezy.
[12:07] <xnox> roaksoax: jamespage: is there an action plan to update / maintain this set of packages? if yes, are we willing to do so via Debian or not?
[12:14] <zorky> anyone here who have any knowledge on juju charm concerning zentyal?
[12:16] <jcastro> not offhand
[12:16] <jcastro> is it broken?
[12:16] <xnox> jamespage: thanks for taking over mongodb =)
[12:18] <zorky> jcastro, theres a problem for us. anyhow. we can't seem to get access beyond the zentyal web interface. The password is unknown. something to do with kerberos password
[12:18] <jcastro> have you set the password?
[12:19] <jcastro> juju set zentyal-single password="foo"
[12:19] <jcastro> according to the readme for a single node thing
[12:23] <Sparky-UK> Hi, does anyone know why (my) snmpd does not increment result for NIC's (particularly eth0) on ubuntu 12.04, counters update if i restart the service tho
[12:24] <zorky> jcastro, yes we tried to set the password to our own choosing. just tried setting the password from the command juju set zentyal-single password="foo" but still no access
[12:25] <jcastro> zorky: hmmm, well, all I can offer is the author's contact info
[12:25] <jcastro> https://launchpad.net/~christophe.sauthier
[12:25] <jcastro> He knows the zentyal guys very well, so hopefully that'll get you on the right track?
[12:30] <maveas> I'm about to write a upstart script for openerp, which relies on postgres, but I'm uncertain about upstart emits during sysvinit scripts. Does upstart emits event from a successfully started sysvinit script?
[12:31] <maveas> Which translates to.. can I use the old sysvinit script for postgres or should I write an upstart script?
[12:32] <maveas> for postgres too?
[12:32] <zorky> jcastro, Will do, thats for the info :)
[12:32] <zorky> jcastro, Thanks*
[12:37] <vezq> maveas: I find sysvinit works more reliably
[12:38] <maveas> vezq: Explain?
[12:39] <vezq> maveas: maybe just more familiar with sysvinit
[12:39] <vezq> for example having different start and stop commands seem not to be trivial with upstart
[12:41] <maveas> vezq: How different? With variables or?
[12:43] <mkander> I want to pay someone to setup raid on my ubuntu server that runs on lvm (default). Anyone interested? Thanks for the help!
[12:44] <maveas> vezq: I am asking because I am still quite new with init (sysv+upstart)
[12:44] <vezq> maveas: just running any different command with upstart
[12:45] <vezq> I'm not that familiar with upstart either
[12:46] <vezq> mkander: RAID should be configured before LVM
[12:46] <vezq> mkander: I might be able to help if needed
[12:47] <mkander> vezq: It is already set up unfortunately
[12:47] <mkander> been running like this for a year, but now I want added security
[12:47] <vezq> mkander: then it's about the only way to recreate the server...
[12:47] <xnox> mkander: just follow https://help.ubuntu.com/12.04/serverguide/advanced-installation.html
[12:47] <maveas> It is possible to do scripts in upstart so I don't see any problems there. It is possible to do pre-start, post-start, pre-stop and post-stop script beside the start script
[12:48] <mkander> vezq: Reinstall everything? Hmm hope there is another way
[12:48] <vezq> mkander: yep I don't really see any other way
[12:49] <mkander> must be possible to install a raid 1 seperately and then move everything there or something
[12:50] <maveas> RAID or no RAID that is the question..
[12:50] <vezq> mkander: sure but might be easier to copy data to safety, delete current stuff, create raid etc.
[12:50] <xnox> mkander: if you have enought disks plugged in, you can configure & setup raid1, then lvm, and then move everything from the non-raid disk across, setup/update bootloader and reboot and hopefully be golden.
[12:58] <mkander> xnox: Yes must be something like that...
[13:11] <setra> hi
[13:13] <setra> Looking for an automation tool which runs on a ubuntu server to handle bash script commandline GUIs which can not be automated through parameter, rather than that need user interaction.
[13:16] <melmoth> setra, your question is unclear... The way i understand it, is "i m looking at a universal graphical interface wrapper for any exisiting shell script"
[13:16] <melmoth> wich i doubt exist
[13:32] <lynxman> jcastro: o/
[13:34] <jcastro> lynxman: yo yo!
[13:36] <lynxman> jcastro: very well done yesterday sir
[13:37] <jcastro> oh you were watching?
[13:44] <jamespage> xnox: np on mongodb - I'm adding some DEP-8 tests (and probably enabling SSL support) in the next few days as well
[13:44] <blazindrop> good day I have a problem with MAAS on ubuntu 12.10. MAAS status page continually says "No boot images have been imported yet...." I've ran maas-import-pxe-files multiple times but this still displays.
[13:45] <jamespage> xnox: re the cluster suite generally - I'd defer to roaksoax on that as he's much closer to me
[13:45] <jamespage> blazindrop, hmm - I've seen that as well - its lying
[13:45] <blazindrop> jamespage, I thought it may be but wasn't sure
[13:45] <blazindrop> jamespage, trying a new install from maas master in a few minutes so will find out the hard way :)
[13:45] <jamespage> blazindrop, well I guess there is the possibility that in your instance its not but I would suspect now
[13:45] <jamespage> now/not
[13:45] <blazindrop> jamespage, ok
[13:48] <jcastro> lynxman: I am looking forward to the webscale session soon
[13:48] <jcastro> \o/ come on nginx in main!
[13:49] <jamespage> jcastro, now you are making me feel guilty
[13:49] <jamespage> that BP has not had alot of love this cycle
[13:49] <jcastro> I know
[13:49] <jcastro> we got clinted.
[13:50] <jcastro> but it's ok, I always have hopes for you to pick up the slack!
[13:50]  * jcastro snickers
[13:50] <Daviey> jcastro: Did you just volunteer ?
[13:50] <lynxman> jcastro: I was! :) nginx in main is schweet
[13:50] <lynxman> Daviey: I think he did?
[13:50] <jcastro> Daviey: meh, it's in the charm store, that's just as good right? :)
[13:51] <setra> melmoth, I have a bash script I run on a server console which installs a lot of tools automated to fit my needs, but there are sometimes user interaction needed to move on. this I would like to eliminate
[13:51]  * Daviey marks blueprint complete, and cancels the vUDS session
[13:51] <setra> with an automatic tool like xdotool which does not work since there is no desktop/window anywhere
[13:52] <melmoth> setra, ok, i think i understand the use case.
[13:52] <melmoth> setra, what about using something like expect , or expect like (there s an expect like module for python) ?
[13:53] <melmoth> http://pexpect.sourceforge.net/pexpect.html
[13:53] <Jeeves__>  debconf-set-selections
[13:53] <setra> melmoth, yes expect is what I found, but I'm not sure how to incorporate it, since I did not find an example for an application like squirrelmail-configure...
[13:54] <setra> do I need to be a python expert?
[13:54] <Jeeves_> setra: debconf-set-selections allows you to set the answers dpkg wants to ask you in advance
[13:54] <melmoth> you ll need to be fluent in wichever tool you want to use. Expect requires tcl knowledge.
[13:54] <melmoth> the python module let you do the same with python (but...requires you to know python :) )
[13:55] <setra> melmoth, just got in touch with bash since I thought I can solve it all, but bummer...
[13:55] <melmoth> if your only problem is some pakcage reconfigure stuff, Jeeves_ is right, you want to have a look at preseed and debconf selection
[13:56]  * melmoth sing it s a long way to the top if you wanna rock n roll.
[13:57] <Jeeves_> setra: Correctly built packages should all use debconf, iirc.
[13:57] <Jeeves_> So you can just set stuff using debconf
[13:57] <Jeeves_> from bash
[13:57] <setra> Jeeves_, thx that solves 95% for me...
[13:57] <Jeeves_> echo "kerio-connect     kerio-connect/force-config      boolean false" | debconf-set-selections
[13:58] <jamespage> Daviey, which session are you running as we appear to have a conflict
[13:58] <Jeeves_> Quite easy, if you ask me.
[13:58] <setra> Jeeves_, straight forward
[13:59] <Jeeves_> `debconf-get-selections | grep 'packagename'` shows you which vars are available for setting
[13:59] <setra> melmoth, so now you know why I struggle a little, for another 5% putting a undefined amount of time in expect...btw, do you got an example for handling an application in a console from a bash script.
[14:00] <melmoth> there was an example using python expect module in th elink i posted above (a simple scp command line prompting for user/password)
[14:01] <setra> melmoth, thx
[14:04] <dv> Hi all. I have a question on MAAS / JuJu
[14:04] <dv> Suppose i said `deploy mysql` and it now serves 100 reqs per sec
[14:06] <Daviey> jamespage: sorry, i am FPI
[14:06] <dv> What will happen with my 4-node cluster after adding three new nodes and removal of three other?
[14:14] <roaksoax> xnox: hi! i thought madkiss was taking care of corosync pacemaer etc etc
[14:14] <roaksoax> xnox: our update plans is to basically drop redhat cluster
[14:14] <roaksoax> and clvm too
[14:15] <xnox> roaksoax: ok. drop clvm without replacement? just use resource agents / something else?
[14:17] <roaksoax> xnox: afaik upstream was dropping ra's and stuff
[14:18] <xnox> roaksoax: "upsream" as in Debian or RedHat? RedHat say it's all still supported, just unmaintained in Debian.
[14:19] <roaksoax> xnox: looing from where i read that
[14:21] <xnox> roaksoax: sorry, what does "looing" mean?
[14:21]  * xnox is not a native speaker.
[14:21] <roaksoax> xnox: if we upgrade, we would be doing option 3: http://theclusterguy.clusterlabs.org/post/34605496260/can-pacemaker-1-1-8-be-used-with
[14:21] <roaksoax> looking*
[14:21] <roaksoax> xnox: currently, both debian an ubuntu support option 2
[14:22] <xnox> ok.
[14:22] <roaksoax> xnox: nd afaik the package versions are similar
[14:22] <roaksoax> and debian has newer of the stack in experimental
[14:22] <xnox> roaksoax: so do you want anything from the lvm2 package? ship  / don't ship clvm package / status quo?
[14:22] <xnox> (currently I kept clvm package around)
[14:22] <roaksoax> so the reason of not upgrading to latest software is a technical decision that would drop support for ofs2
[14:22] <roaksoax> and as i was pointed out  couple months ago
[14:23] <roaksoax> clvm ws to be dropped for no due to lack of support iirc
[14:23] <roaksoax> that's why i need to find where i read that frm
[14:24] <roaksoax> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=697676
[14:24] <roaksoax> which you are currently aware of
[14:27] <jcastro> smoser: we have like 3 hours from the end of UDS to the LUG talk, so we should be fine.
[14:28] <roaksoax> xnox: err i gave you a wrong link: http://theclusterguy.clusterlabs.org/post/34604901720/pacemaker-and-cluster-filesystems
[14:35] <dv> Bob_: Yes, i knew it, just wanted dv
[14:35] <dv> sry not relevant
[15:02] <blazindrop> jamespage, looks like that is just a false positive alert about pxe images not loaded. so far I have a MAAS slave booting fine
[16:03] <zul> jamespage:  can you have a look please? http://paste.ubuntu.com/5590733/
[16:07] <jamespage> zul: "This is a transitional dummy package> It can be safely removed."
[16:07] <jamespage> typo
[16:07] <jamespage> > -> .
[16:08] <zul> jamespage:  doh
[16:08] <jamespage> zul, typos:
[16:08] <jamespage> Replaces: pyhton-oslo-config (<< 2013.1~b5~)
[16:08] <jamespage> Breaks: pyhton-oslo-config (<< 2013.1~b5~)
[16:08] <jamespage> spelt wrong
[16:08] <zul> jamespage:  doh got it
[16:09] <jamespage> zul, I'd personally add a comment re the package renaming and transitional package in the changelog as well
[16:09] <jamespage> i.e. why thats happening
[16:09] <zul> jamespage:  done
[16:11] <jamespage> zul: please fix the watch file as well - it needs to take into account the package rename
[16:11] <zul> jamespage:  k
[16:12] <zul> jamespage:  this probably should be vcs controlled like the openstack packages me think
[16:12] <jamespage> zul: not quite sure what you mean?
[16:12] <zul> jamespage:  ie like python-glanceclient
[16:14] <jamespage> zul, you want to set it up in the CI?
[16:14] <jamespage> or just have a non-distro packaging branch?
[16:14] <zul> jamespage:  i think it should be in the ci so we can catch breakages
[16:39] <zul> jamespage:  lp:~openstack-ubuntu-testing/oslo/grizzl
[16:40] <zul> jamespage:  er lp:~openstack-ubuntu-testing/oslo/grizzly
[17:44] <jiboumans> jjohansen: are you around? I could use your input on a kernel/UDP issue i'm having.
[17:44] <jjohansen> jiboumans: what is the problem?
[17:46] <jiboumans> jjohansen: prefixing this with "I'm a total newbie C/socket programmer", I'm trying to figure out if I can make a write(udp_fd, str, len) call faster - it's taking 50 microsecs on my VM, and this is for a patch inside Redis (in memory store), where the actual processed commands take 40-100 microsecs in total.
[17:47] <jiboumans> this is to send stats off to Statsd on localhost, so I'm ok playing a bit fast & loose with losing a stat here & there if it gains some speed in the process.
[17:47] <jiboumans> in short, the stat collecting is slowing the rest down so much (relatively speaking) it's prohibitive to run it.
[17:48] <sarnold> jiboumans: 50 _microseconds_ or 50 _milliseconds_?
[17:48] <jiboumans> microseconds
[17:48] <sarnold> jiboumans: you may wish to try timing a tight loop of umask() or something similar to see what syscall overhead is expected to be. that sounds pretty snappy to me.
[17:49] <jjohansen> jiboumans: there isn't a lot you can do to optimize at the syscall level. What networking does your VM use
[17:50] <jiboumans> sarnold: in absolute numbers it's pretty snappy indeed, but relatively speaking it throttles the redis througput to about a third. Obviously the call is doing more than just copying the string into kernel memory, and if I could get it to return when that was done and handle the sending 'later' at the risk of losing some error handling, i'd be fine with that.
[17:51] <jiboumans> jjohansen: this is to 127.0.0.1, so it doesn't go off box. I'm assuming it'd use loopback for that.
[17:53] <jjohansen> jiboumans: you can try doing async io, http://www.ibm.com/developerworks/library/l-async/ has a decent introduction to it
[17:53] <jiboumans> jjohansen: here's the code https://github.com/jib/redis/compare/2.6...statsd-support-2.6#L4R57 (linked to connect). The send is on line 162.
[17:54] <jiboumans> jjohansen: hmm, that looks different than the O_ASYNC/O_NONBLOCKING flags, right?
[17:55] <jjohansen> jiboumans: yes
[17:57] <jiboumans> jjohansen: i'll give that a try. is this linux kernel only thing?
[18:00] <sarnold> I think there's a POSIX spec for aio but I wouldn't be surprised if implementers are few..
[18:00] <jjohansen> jiboumans: I am not sure what the status of aio is on other kernels but it is a posix spec
[18:01] <jiboumans> jjohansen: will have to dig into this then.. pretty sure redis folks wnat to keep building on BSD etc too.
[18:01] <jiboumans> but first, let's test that it goes faster indeed. thanks for the pointer!
[18:26] <jiboumans> jjohansen: i may be doing something very silly, but I'm getting this error when calling aio_write: redis/src/statsd.c:191: undefined reference to `aio_write64'. Looking in /usr/local/include, I can't find any definitions for aio_write64 indeed. Does this come from an extra kernel module or so? I have libaio-dev & libaio1 installed.
[18:27] <jiboumans> manpage suggests I only need <aio.h>, and that does provide the struct needed, but no definition of aio_write.. bit puzzled.
[18:28] <sarnold> jiboumans: my aio_write64 is protected by #ifdef __USE_LARGEFILE64
[18:30] <jjohansen> jiboumans: its in the aio.h include here as sarnold says its protected by an ifdef
[18:30] <jjohansen> jiboumans: aio.h is installed by libc6-dev:amd64:
[18:31] <jiboumans> hmm, i dont even see it in the /source/ though. Never mind the #ifdef not exposing it.
[18:31]  * jiboumans checks versions
[18:31] <jiboumans> ii  linux-libc-dev 2.6.32-41.89 Linux Kernel Headers for development
[18:32] <jiboumans> (this is on lucid btw, I should have probably mentioned this earlier)
[18:34] <jiboumans> hmm, looking at the right aio.h may help.. there are many.
[18:35] <jiboumans> sarnold, jjohansen: so, what should i be doing at this point? I'm a tad lost.
[18:37] <SpamapS> jiboumans: o/
[18:37] <jiboumans> SpamapS: howdy!
[18:37] <SpamapS> jiboumans: still plugging away on ye olde lucid eh?
[18:38] <SpamapS> haven't you heard, the cool kids just run raring now
[18:38] <SpamapS> forever
[18:38] <jiboumans> on daily snapshots right?
[18:38] <sarnold> jiboumans: heh, sorry, haven't used it beyond toys in ages :/
[18:38] <jcastro> oh hey jiboumans!
[18:38] <jiboumans> damn i wish i were cool
[18:38] <SpamapS> jiboumans: snapshots are for pussies.
[18:38] <SpamapS> jiboumans: while apt-get update ; apt-get dist-upgrade -y ; done
[18:39] <SpamapS> NO WAITING
[18:39] <jiboumans> jcastro: howdy sir!
[18:41] <jiboumans> sarnold: thanks for trying. I'd expected to at least have a working aio_write definition in there though :)
[18:41] <sarnold> jiboumans: indeed :)
[18:50] <jjohansen> jiboumans: sorry I haven't played with it either
[18:51] <jiboumans> jjohansen: let me ask you this then, should __USE_LARGEFILE64 be undefined? What sets it for ubuntu kernels?
[18:52] <jiboumans> (sarnold got the cause for the error right with that)
[18:52] <sarnold> jiboumans: Wouldn't it be something you've got to set in your makefiles, similar to __GNU_SOURCE or related feature macros?
[18:53] <sarnold> (i'll admit a strong weakness here...)
[18:53] <jiboumans> sarnold: by all intents and purposes, i shouldn't be allowed to code C, so i'm relying on y'alls input here :)
[18:53] <sarnold> jiboumans: hehe :D
[18:57] <jjohansen> jiboumans: define  _GNU_SOURCE in your source
[18:59] <jiboumans> jjohansen: kernel source yeah? i'm obviously out of my depth here.
[18:59] <jiboumans> and if ubuntu kernels don't build with this, i'm assuming a lot of 'm don't.. meaning aio_write wouldn't work for them either? sad face.
[18:59] <jjohansen> jiboumans: no, at the top of your C prog, put
[18:59] <jjohansen>  #define _GNU_SOURCE 1
[18:59] <jjohansen> and then do your includes
[19:01] <jiboumans> ah interesting. in the redis src there's a "#define _GNU_SOURCE" - no value. wonder why they did that
[19:01]  * jiboumans chagnes and tries
[19:03] <jiboumans> still undefined reference to aio_write64. i'll dig further - thanks for pointing me in the right direction jjohansen, sarnold.
[19:04] <sarnold> jiboumans: oh man :/ good luck :)
[20:11] <zul> adam_g: https://code.launchpad.net/~zulcss/glance/glance-fix-ups/+merge/152048
[20:54] <keithzg> Arghh, why am I having such a hard time configuring an IMAP server? Fail, me.
[20:59] <keithzg> ...wait, IMAP authentication works fine if I'm using a local user? Why wouldn't users auth'd via LDAP work, since PAM is being used for auth and thus ostensibly querying LDAP?
[20:59] <sarnold> keithzg: does your /etc/pam.d/<imap service name> config file allow ldap?
[21:02] <keithzg> sarnold: I don't see anything in /etc/pam.d that'd be related, actually (I'm trying Courier at the moment, btw)
[21:03] <sarnold> courier-imap: /etc/pam.d/imap
[21:03] <sarnold> keithzg: does that file look sane?
[21:04] <keithzg> sarnod: oh, durr, don't know how I didn't see that. But yeah, it looks sane to me, @include lines for common-auth, common-account, common-password and common-session.
[21:05] <sarnold> keithzg: and do the common-files work for e.g. ssh or ftp or lightdm or login or su or sudo or .. some other easily-testable service? :)
[21:07] <keithzg> I can certainly ssh into them; the server is headless so can't try lightdm. But it also works if I ssh in as a local account then 'su username' with that user's password.
[21:09] <keithzg> sarnold: Although auth *does* take a troublingly long time (lag of several seconds at least)
[21:10] <sarnold> keithzg: hrm, failing dns queries are often a source of noticable lag.. are you logging reverse-lookup names instead of ips in your connection logs?
[21:15] <keithzg> sarnold: I must admit I'm not sure what you're referring to; do you mean mail.log?
[21:15] <keithzg> sarnold: if that's the case, I'm seeing bits like "connect from siddhartha.gmcl.internal[10.1.187.23]" and "LOGIN, user=blahblah, ip=[::ffff:10.1.187.23], port=[60758], protocol=IMAP"
[21:16] <sarnold> keithzg: that'd be the 'siddhartha.gmcl.internal' portion
[21:17] <sarnold> keithzg: though in that case it succeeded :) hehe -- how long does it take to run 'host 10.1.187.23' on that machine? or nearby IPs?
[21:22] <keithzg> sarnold: a long lag as I installed bind9-host, heh, but about 1 second to actually resolve IP to name or vice-versa
[21:22] <keithzg> sarnold: I have a suspicion the auth delay is thanks to the virtual server that's running being periodically snapshotted to discreet external-file snapshots; it's really helpful for backup purposes, but unfortunately the version of kvm-qemu that ships with 12.04 isn't new enough to blockpull the snapshots back to collapse them, thus eventually leading to an extremely long backing chain, heh.
[21:23] <sarnold> keithzg: ooooof.
[21:23] <sarnold> keithzg: that sounds painful :)
[21:23] <keithzg> sarnold: yeahhhh, I'm probably going to eventually give in an upgrade to a non-LTS!
[21:24] <keithzg> Other possible option is I could transfer the VMs to a 12.10 or later box, blockpull, and then copy them back. Kindof an ugly way of doing it, though (and far from automatic or foolproof)
[21:26] <sarnold> keithzg: It might be worth considering an SRU for that new feature, it sounds useful...
[21:31] <keithzg> sarnold: not a bad idea a'tall
[21:35] <keithzg> sarnold: hmm, if I try a manual login via telnet I get "* BYE Temporary problem, please try again later"
[21:35] <sarnold> keithzg: o_O
[21:35] <sarnold> that's .. odd
[21:39]  * keithzg is the king of strange and confusing errors ;)
[21:51] <keithzg> Yeesh this is an extremely hard issue to Google for, since most people using LDAP and IMAP are doing rather different and far more complex things than I.
[22:02] <keithzg> Is it possible that Courier cannot authenticate via ldap through the 'authpam' module, and would have to actually use its own discrete 'authldap' module? That'd be rather annoying for my use-case, sigh.
[22:19] <Seveas> keithzg, it's been a while since I used courier, but I'm pretty sure I used pam (and pam used ldap)
[22:19] <keithzg> Seveas: Okay, that's good to know. So this *should* be working, then.
[22:23] <Seveas> keithzg, what does authtest say? And does tcpdump see any ldap bind attempts?
[22:30] <Seveas> keithzg, I've always found http://www.courier-mta.org/authlib/README.authdebug.html helpful in debugging courier auth issues (and I've had a few...)
[22:33] <keithzg> Seveas: oh, thanks! I'll check that page out.
[22:35] <keithzg> ...alas, everything relevant in there I have already tried.
[22:35] <keithzg> Seveas: authtest sadly albeit predictably gives the same error as mail.log show, ie. "Authentication FAILED: Input/output error". Working on tcpdump, I feel like I'm missing something since I don't see anything at all when I try to connect, even though on the client end I'm getting responses to my telnet commands.