[00:15] <halvors> I'm running a Postfix SMTP server, i use the mail-stack-delivery package. But i'm not able to login using SMTP port 25, only using Submission port 587
[00:16] <halvors> I get, relay access denied :(
[00:17] <ScottK> Use port 587.  That's what it's for.
[00:21] <halvors> ScottK: But why am i unable to connect using port 25?
[00:21] <halvors> And why not use port 25?
[00:22] <ScottK> IIRC, mail-stack-delivery may not set up smtp auth for port 25.
[00:22] <ScottK> 587 is for mail submission.
[00:22] <halvors> I know.
[00:22] <halvors> But how to enable it?
[00:22] <ScottK> Look in /etc/postfix/master.cf
[00:23] <ScottK> If you see what's different for the smtp service and the submission service, you'll probably be able to figure it out.
[00:24] <ScottK> 587 is better for people who travel a lot because it's not generally blocked or redirected by ISPs/hotels/etc.
[00:25] <halvors> I cannot see any difference...
[00:25] <halvors> http://pastebin.com/0dJBhnZj
[00:28] <ScottK> Interesting.
[00:28] <ScottK> It may be something in main.cf too.
[00:31] <halvors> http://pastebin.com/Ca2LqtyB
[00:31] <halvors> :)
[00:32] <ScottK> OK.  No idea why it works on one port and not the other then.
[00:32] <halvors> Have to be something with the authentication...
[00:34] <halvors> ScottK: Should i uncomment the "-o" lines under the submission line?
[00:35] <qhartman> halvors, where are you trying to login _from_? Many ISPs block outbound SMTP connections on 25 as a spam prevention policy. Trying doing it from the server itself if you have not.
[00:42] <halvors1> qhartman: I've done that, if you try telneting the server on port 25 (s1.halvors.org) you'll see that it responds...
[00:43] <sarnold> halvors1: 250 2.0.0 Ok: queued as C5DB4D60B06
[00:44] <halvors1> Yes :)
[00:44] <halvors1> Dec  6 01:43:20 halvors-server postfix/smtpd[29286]: C5DB4D60B06: client=c-71-237-200-29.hsd1.or.comcast.net[71.237.200.29]
[00:44] <qhartman> halvors, cool
[00:44] <sarnold> halvors1: yay :D
[00:44] <halvors1> But it doesn't work when trying to authenticate...
[00:45] <sarnold> oh. I never learned how to do that by hand.
[00:45] <sarnold> but if you're authenticating, why not use 587?
[00:45] <sarnold> you presumably have control over the clients that intend to authenticate..
[00:45] <halvors1> sarnold: Many people try using port 25...
[00:45] <halvors1>  5.7.1 <halvors@skymiastudios.com>: Relay access denied
[00:46] <halvors1> When trying to login...
[00:46] <halvors1> SMTPS SSL/TLS just times out.
[00:47] <halvors1> sarnold: What port did you use?
[00:47] <halvors1> :)
[00:47] <sarnold> halvors1: 25
[00:48] <halvors1> Without authenticating?
[00:48] <sarnold> halvors1: right
[00:48] <halvors1> Isn't that bad?
[00:48] <sarnold> it's how email works, for better or worse. :)
[00:49] <halvors1> Oh you act like a mail server?
[00:49] <sarnold> halvors1: yeah
[00:49] <halvors1> And connect on port 25 to deliver mail?
[00:49] <sarnold> halvors1: http://paste.ubuntu.com/1413692/
[00:49] <sarnold> I typed all the lines without status codes :)
[00:50] <halvors1> If you try sending to halvors@skymiastudios.com does it reject then?
[00:51] <ScottK> halvors1: Yes.  That's how you enable it.
[00:52] <sarnold> halvors1: denied
[00:53] <sarnold> funny, even mail from: sarnold@halvors.org, rcpt to: halvors@skymiastudios.com fails
[00:54] <EntropyWorks> if your planning on netboot install 12.10 be ready for some headaches
[00:55] <halvors1> sarnold: hmm.
[00:55] <halvors1> Cause that works when using port 587 and starttls
[00:55] <sarnold> halvors1: you probably set the relay rules by certificate?
[00:56] <halvors1> Um. I really don't know, where is that options set?
[00:57] <halvors1> http://paste.ubuntu.com/1413705/
[01:03] <halvors1> sarnold: Any idea?
[01:03] <sarnold> halvors1: no, sorry :)
[01:03] <sarnold> I was expecting to see a relaymap or something...
[01:04] <sarnold> .. and with the destination domains not including skymiastudios.com I was a bit surprised that mail gets there on port 587, but .. it's been a long time since I've _used_ postfix.
[01:09] <halvors1> http://paste.ubuntu.com/1413721/
[01:10] <halvors1> sarnold: Do you think the following lines should be commented out?
[01:10] <sarnold> halvors1: no idea there :) sorry :(
[01:17] <halvors1> sarnold: Ok thank you for your time :)
[01:17] <sarnold> halvors1 :)
[01:17] <sarnold> I just wish I had the answers..
[01:18] <halvors1> sarnold: :)
[02:22] <jdstrand> re libvirt merge> sure thing
[02:22] <jdstrand> hallyn: ^
[05:24] <imachine> hello
[05:24] <imachine> anyone use ubuntu 12.04 or 04.1 as -vserver?
[05:26] <imachine> I have a debian squeeze install with 3 vservers. Can I painlessly (i.e. set up/config vservers on ubuntu server box, copy the vserver bodies from debian, start services, profit) migrate that to Ubuntu Server 12.04.* ?
[05:27] <imachine> The big question is, can I run debian squeeze inside of ubuntu-12.04 server?
[05:51] <ScottK> I'm pretty sure you can.
[05:51] <ScottK> But I haven't done it myself.
[11:31] <jamespage> zul, I'm looking at bug 1085038; looks like the generated entry points are all easy-install so not relevant for the packaging
[11:37] <Nafallo> guys. are there any plans on compiling libvirt with vpx support?
[11:49] <Nafallo> that might come down from debian...
[11:49] <Nafallo> the exclusion of esx that is
[12:25] <jamespage> yolanda, zul, adam_g: review up for quantum - https://code.launchpad.net/~james-page/quantum/grizzly-updates-1/+merge/138438
[12:26] <jamespage> also worth being aware of bug 1030195
[12:26] <jamespage> I'm pretty sure that alot of postrm scripts are being generated to get rid of that error when its actually a false positive
[13:02] <Nafallo> ooooh
[13:03] <Nafallo> vpx/esx is disabled because it would
[13:03] <Nafallo> put libvirt in multiverse, isn't it?
[13:35] <hallyn> zul: hey, when you get some time, i think i need some packaging help with https://launchpad.net/~serge-hallyn/+archive/crossc/+sourcepub/2825655/+listing-archive-extra
[13:36] <zul> hallyn: sure just getting in
[13:36] <zul> which means still waking up
[13:36] <hallyn> zul: me too, tbh
[13:43] <saban> hi, i m using debian.. and i m getting this error when apt-get update W: GPG error: http://ppa.launchpad.net lucid Release: The following signatures couldn't be verified because the public key is not available: NO_PUBKEY B18637BB5175BC68... ? how is this posiblle?
[13:45] <rbasak> saban: did you use add-apt-repository to add your PPA? If not, the easiest way to explain how to fix it is just to remove it from sources.list and use add-apt-repository to add it again
[13:46] <hallyn> zul: so really i have two questions, the non-urgent one is what i should call the version # on the new qemu package, given the debian version is < both the ubuntu qemu-kvm and qemu-linaro versions
[13:47] <zul> erm..
[13:47] <hallyn> zul: the more urgent one is what to do with qemu-utils and qemu-kvm breaks/replaces
[13:47] <zul> jamespage: ^^^
[13:48] <saban> rbasak: as i remember i didnt add anything. and my source.list is this http://pastebin.com/h9wBdkUY
[13:48] <zul> hallyn: so uh what? :)
[13:48] <hallyn> zul: hold on, i'll ask you in 29 mins
[13:48] <hallyn> 20 that is :)
[13:48] <zul> k
[13:48] <hallyn> thx
[13:51] <rbasak> saban: check /etc/apt/sources.list.d/. And it looks like you're running Debian, not Ubuntu!
[13:52] <jamespage> zul, was ^^^ in reference to quantum or hallyn question?
[13:52] <zul> hallyn: question
[13:53] <zul> er..
[13:53] <zul> jamespage: hallyn's question
[13:55] <saban> rbasak: yes i know :P i was just asking here becouse i hadnt any clue why i have ppa problem in my debian server.. i got it now one of program was from ubuntu package or something.. now i know where is the problem... tnx
[14:04]  * jamespage thinks
[14:07] <jamespage> hallyn, can you explain the version # question more? whats the difference between ubuntu+debian?
[14:07] <hallyn> jamespage: the more urgent problem I have is:  qemu-kvm depends on qemu-system depends on qemu-utils, qemu-utils does break/replace qemu-kvm, but it wnts to be upgraded before qemu-kvm (bc of the relations)
[14:07] <hallyn> so short version is 'apt-get dist-upgrade breaks'
[14:08] <jamespage> ugh
[14:08] <hallyn> right.  and because of the way files are being moved between packages, everything needs to be replaced at once
[14:08] <soren> Are none of those relationships versioned?
[14:09] <hallyn> all are
[14:09] <hallyn> isn't helping
[14:09] <hallyn> when qemu-utils is being upgraded, it's seeing the old qemu-kvm...
[14:10] <soren> Is this already uploaded or do I need to look in bzr to see the control file?
[14:10] <soren> If the latter, which branch do I need?
[14:11] <hallyn> soren: uploaded to ppa:serge-hallyn/crossc for raring
[14:13] <jamespage> soren, dget -u https://launchpad.net/~serge-hallyn/+archive/crossc/+files/qemu_1.2.0-z-dfsg-8.dsc
[14:14] <soren> qemu-system doesn't seem to depend on qemu-utils, fwiw.
[14:14] <soren> It recommends it, though, so the result is probably mostly the same.
[14:15] <jamespage> hallyn, I'm not sure you need Breaks/Replaces on qemu-kvm itself as its a transitional package
[14:15] <hallyn> jamespage: i need the files in the old qemu-kvm to be removed before qemu-system is installed...
[14:15] <hallyn> but yeah, qemu-system break/replacing *should* be enough
[14:15] <hallyn> i'm just getting desparate :)
[14:16] <soren> At least your kvm package will need an epoch added to it.
[14:16] <hallyn> yeah i'll get rid of the break/replaces on qemu-kvm altogether.
[14:17] <hallyn> soren: what do you mean?
[14:17] <soren> hallyn: Sorry, I phrased that poorly.
[14:17] <soren> hallyn: Your kvm package will need an epoch.
[14:17] <hallyn> soren: that also sounds like part of an answer to my first question - how to version these
[14:18] <soren> hallyn: The others won't need it, just the kvm one. Otherwise it won't supersede the existing kvm package.
[14:19] <jamespage> hallyn, I'm wondering whether not having a (= ${source:Version) on Recommends: qemu-utils might be something todo with
[14:19] <jamespage> it
[14:19]  * jamespage grasps at straws
[14:19] <hallyn> soren: is the kvm package still needed?  it was transitional...
[14:20] <hallyn> jamespage: that sounds right!
[14:20] <soren> hallyn: Nope, you should be able to just get rid of it. Good point.
[14:21] <hallyn> soren: though will ppl get confused if they can't apt-get install kvm' ?
[14:21] <jamespage> hallyn, Recommends can do weird stuff with ordering and configuration
[14:22] <soren> hallyn: Other things that you can get rid of also include:
[14:22] <soren>  Conflicts: kvm (= 85+dfsg-4.1)
[14:22]  * jamespage thinks of bacula as a case in point
[14:22] <soren> It's been a while since that was useful (if ever).
[14:22] <soren> hallyn: Yeah, that's true.
[14:22] <soren> hallyn: Things might still depend on kvm rather than qemu-kvm or whatever.
[14:23] <hallyn> soren: current raring package does not have an epoch on kvm...
[14:23] <soren> hallyn: Yes, it does?
[14:24] <soren> Hm.
[14:24] <soren> From what I can tell, apt is doing exactly what it's being told.
[14:24] <soren> You're specifying that qemu depends on qemu-system (>= same version).
[14:25] <soren> So you're telling it that qemu-system must be installed before qemu.
[14:25] <soren> Same for qemu-system -> qemu-utils.
[14:25] <hallyn> i'm not asking for qemu
[14:25] <soren> Oh, sorry.
[14:26] <soren> Well, same problem, really.
[14:26] <hallyn> the lack of versioning on recommends as jamespage suggested sounds a likely culprit,
[14:26] <soren> qemu-kvm depends on qemu-system (= same version).
[14:26] <soren> So it must be installed first.
[14:26] <hallyn> right
[14:26] <soren> And then qemu-system says it needs qemu-utils first.
[14:26] <soren> But qemu-utils says it'll break qemu-kvm.
[14:26] <hallyn> << its own version
[14:27] <soren> Right.
[14:27] <hallyn> so i want qemu-kvm removed.
[14:27] <hallyn> that should do the right thing right?
[14:27] <hallyn> it takes hours for a build to happen so i want to look for other possible problems before submitting :)
[14:27] <soren> Are you attempting to get rid of qemu-kvm?
[14:27] <soren> Is that the goal?
[14:27] <hallyn> yes
[14:27] <soren> Ah. I see.
[14:28] <hallyn> qemu-system provides everything
[14:28] <soren> I missed that entirely.
[14:28] <hallyn> (it's the debian-experimental git branch, merging qemu+qemu-kvm)
[14:28] <hallyn> sorry - that would be confusing then!
[14:29] <soren> Which package replaces qemu-kvm?
[14:29] <hallyn> qemu-system
[14:30] <soren> Ok. So qemu-system should say that it replaces, conflicts and provides qemu-kvm.
[14:30] <hallyn> oh, provides
[14:30] <soren> Yeah.
[14:30] <soren> The problem is that you have versioned dependencies on it.
[14:31] <soren> I don't think we can do versioned provides yet.
[14:31] <soren> I could be wrong, though. It's been a while.
[14:32] <soren> Does anything have a versioned dependency on qemu-kvm?
[14:32] <hallyn> so right now i have http://paste.ubuntu.com/1414805/
[14:32] <hallyn> as the new debian/control
[14:32] <hallyn> they just have breaks/replaces on versioned qemu-kvm
[14:34] <hallyn> soren: jamespage: all right i'll try the pastebin'd version;  my other question then was,
[14:35] <hallyn> no,
[14:35] <hallyn> oh i didn't type that
[14:36] <hallyn> debian version is 1.2.0-dfsg, ubuntu version is 1.2.0+noroms and 1.2.0-2012
[14:36] <hallyn> apart from updating the source to qemu 1.3.0, or doing what i'm doing in ppa (using 1.2.0-z-dfsg),
[14:36] <hallyn> is there something else i can do with the versioning when switching to the debian packages?
[14:37] <hallyn> i don't wnat to do an epoch really since the problem goes away with 1.3.0
[14:47] <hallyn> soren: jamespage: thanks, let's see how that goes  (build queued)
[15:01] <luminous> hi, I have had a hard time getting this sorted out in #ubuntu, my issue may require some more expertise,  so I'm here :)   I'm trying to install redmine on a ubuntu server, so it is relevant.
[15:01] <luminous> http://dpaste.com/842127/ <<< libmagick-dev won't install properly for me, having cancelled aptitude mid-run
[15:01] <luminous> aptitude install & cancel looked like:  http://dpaste.de/qi8dq/
[15:54] <jamespage> adam_g, I think we need to wrestle the swift charms back off my finnie
[15:55] <jamespage> there ain't much review going on around this commits.
[15:57] <hallyn> jdstrand: hey,
[15:57] <hallyn> jdstrand: so i'm working on a patch to allow 9p mounts in libvirt,
[15:57] <hallyn> but the example used in the bug report was a 9pfs under /home
[15:57] <hallyn> virt-aa-helper expliclty disallows paths under /home
[15:57] <hallyn> do we want to enforce that for this too?
[15:58] <jdstrand> hallyn: I'm not familiar with 9p mounts. what do they do?
[15:59] <hallyn> i'm not either in particular, but they offer a dir on the host through the 9pfs network fs to the vm,
[15:59] <hallyn> and actually  my patch is treating them and fstype=bind the same way
[16:00] <hallyn> (this is re bug 943680)
[16:01] <hallyn> wait, i don't see home explicitly listed
[16:01] <jdstrand> hallyn: yeah, me either
[16:02] <hallyn> but i do get virt-aa-helper: error: /home/ubuntu/9p
[16:02] <jdstrand> hallyn: are you talking about the apparmor profile for virt-aa-helper itself, or the profile it generates for the vm?
[16:02] <hallyn> virt-aa-helper: error: skipped restricted file
[16:02] <hallyn> the profile it generates for the vm
[16:02] <hallyn> gah, sorry, biam
[16:03] <jdstrand> hallyn: that is probably this check:
[16:03] <jdstrand> case S_IFDIR:
[16:03] <jdstrand>   return 1;
[16:03] <jdstrand>   break;
[16:08] <hallyn> jdstrand: oh.  that's right.  thanks!  i guess i need to make that '${x}/**'
[16:08] <jdstrand> hallyn: what type of rule do you want to add in the end?
[16:08] <jdstrand> right
[16:08] <jdstrand> so...
[16:08] <jdstrand> that will require more work
[16:09] <hallyn> snprintf :)
[16:09] <jdstrand> hallyn: maybe create a vah_add_dir()
[16:09] <jdstrand> fyi, upstream doesn't like snprintf
[16:09] <hallyn> sigh
[16:09] <hallyn> double sigh
[16:09] <hallyn> they wont reply to me anyway :)
[16:10] <hallyn> ok i think i'm going to put that off then.  i thought it might be a quick addition to the other patches i sent to you for review,
[16:10] <jdstrand> hallyn: possibly because you didn't run 'make check' 'make syntax-check'
[16:10] <jdstrand> make syntax-check will tell you about things like snprint
[16:10] <jdstrand> f
[16:11] <hallyn> and tell me to write in a custom language?
[16:11] <jdstrand> they will want you to use something like virBufferAsprintf
[16:12] <jdstrand> so would I btw, that is what virt-aa-helper does
[16:12] <jdstrand> I got dinged for that when I submitted my patches initially
[16:12] <jdstrand> hallyn: so, it doesn't have to be overly complicated
[16:13] <hallyn> no but there are other things to do, and i'm chafing at custom language
[16:13] <jdstrand> vah_add_file() grows a 'bool dir' arg
[16:13] <jdstrand> then you can have a different call to virBufferAsprintf when its true
[16:14] <jdstrand> valid_path could also grow a 'bool dir'
[16:14] <hallyn> do we want the same restrictions on dirs as on files?
[16:14] <hallyn> (besides "can't be a dir" :)
[16:15] <jdstrand> then in get_files() if it is 9p, set dir=True, otherwise default to False
[16:16] <hallyn> jdstrand: not quite like that,
[16:16] <jdstrand> hallyn: re same restrictions> for now, I think yes
[16:17] <hallyn> i'm specifically running through the ctl->def->fss and looking for 'mount' and 'bind' fstypes,
[16:17] <hallyn> but right
[16:17] <jdstrand> well, ok
[16:17] <jdstrand> basically, default to False unless you need it to be True :)
[16:17] <hallyn> (just want to make sure you're not suggesting I do this elsewhere through a call to reload_profile)
[16:18] <jdstrand> no, just in get_files()
[16:19] <hallyn> ok.  thanks.  i guess first (once my disgust settles) i'll do the virBufferAsprintf thing for the patches i've already sent.
[16:19] <jdstrand> hallyn: so, I was going to suggest making 'make syntax-check' happy. should I just wait for you to do that before I review the merge?
[16:19] <hallyn> soren: jamespage: ok, still having the same problem...
[16:20] <jdstrand> hallyn: sorry about that. actually, once you get used to it, their functions work pretty well... but I empathize having gone through it myself
[16:20] <hallyn> jdstrand: well assuming it's a simple replacement, i'd say no, there's a lot more to those patches which could have real bugs
[16:21] <hallyn> so as long as there's no problems with the fundamental patches, then i can avoid bugging you next time i send :)
[16:21] <hallyn> soren: jamespage: so qemu-kvm upgrade causes qemu-system install recommends qemu-utils install which breaks on the old version of qemu-kvm, which isn't installed bc of this loop.
[16:22] <hallyn> hm, lemme check the pkg contents.  maybe that last conflicts isn't strictly needed
[16:24] <hallyn> drat.  it is.  qemu-io is in both
[16:25] <hallyn> wait, if i have qemu-system providing qemu-kvm, does that mean i don't need a qemu-kvm metapackage?
[16:26] <jamespage> hallyn, I might be tempted to mock this up in an empty package
[16:26] <jamespage> so I could tweak the depends and see effects quickly
[16:27] <hallyn> do i maybe want conflicts instead of breaks?
[16:31] <hallyn> jamespage: yes, except i'd probably spend several hours setting that up without breaking the rest of the packaging so nothing installs :)
[16:31] <hallyn> lemme ask this q in -devel
[16:32] <jamespage> hallyn, I think that might be a good idea
[16:41] <hallyn> uh, wait.  i see another stupid thing i did.  not necessarily related, but perhaps.  biab
[16:42] <jamespage> yolanda, looking at your revised MP for nova/upstart now
[16:43] <yolanda> jamespage, i was trying to finish the testings anyway
[16:43] <yolanda> so maybe it isn't correct
[16:47] <jamespage> yolanda, the changelog is a little fuddled and not correct formatted - but aside from that looks OK
[16:48] <yolanda> jamespage, i have it pending to be corrected, yes, i need to push the changes
[17:05] <jamespage> adam_g, zul: any objection if I create some new views in the lab?  I like the OS series centric stuff we have in the public instance
[17:05] <zul> no objections here if it makes life easier
[17:20] <jamespage> yolanda, merged your changes; deploying into the lab right now....
[17:20] <jamespage> (nova is the trigger for a re-deployment test - and adam_g appears to have the grizzly deployment on precise working now)
[17:27] <yolanda> great!
[17:27] <yolanda> i trust more in the changes, now i can test it properly with the new environment
[17:53] <adam_g> jamespage: +1 to both swift charms and modifying the jenkins views. i was hoping todo some swift work this week.
[17:54] <jamespage> adam_g, the swift charms contain some bad juju atm the moment
[17:54] <adam_g> jamespage: is cloud-init the main blocker for quantum's use of config drive on 12.04?
[17:54] <samba35> rebooted
[17:54] <jamespage> adam_g, its the guest that needs the later version of cloud-init
[17:54] <adam_g> jamespage: ya, last i checked the keystone stuff was hard-coded config options. actually it seems whoever pushed the changes to the official charms didn't push all of them
[17:55] <adam_g> jamespage: oh, ok, so its a guest-only issue?
[17:55] <jamespage> adam_g, yes
[17:55] <jamespage> adam_g, quantal cloud images should just work
[17:55] <jamespage> adam_g, the other thing you need todo is create a tenant network
[17:55] <adam_g> jamespage: okay. i tried boht last night with no luck. the instance never got its IP, which i assume it gets from the node running the dhcp agent?
[17:55] <jamespage> adam_g, the ext_net is shared across tenants and has not DHCP - its for floating IP's
[17:56] <jamespage> adam_g, nova-cloud-controller has a 'quantum-tenant-net' script which will create one for you
[17:56] <jamespage> I created one for the admin tenant and then ran nova instances as that admin user - that worked OK
[17:56] <adam_g> jamespage: but for fixed-ips.. i had the network and ports created so that new instances got new fixed IPs from the tenants network associated with it, but the IP never made it to the instance via DHCP
[17:58] <adam_g> jamespage: does all of the internal traffic flow over eth0? or is it possible/required to set that stuff to use a specific network, similar to flat_interface /w FlatDHCP?
[17:59] <hallyn> jdstrand: lp:~serge-hallyn/ubuntu/raring/libvirt/libvirt-hugepages updated to use virBuffer
[17:59] <hallyn> recon i need to resend to libvirt list
[18:01] <jamespage> adam_g, it does; its encapsulated in gre tunnels between the ovs instances in each compute node and the quantum gateway
[18:01] <jamespage> adam_g, its possible todo other network configurations but thats the one the charm supports right now
[18:01] <jamespage> I do want to add other ones....
[18:02] <jamespage> adam_g, I've not tried what you did with the tenant network;;;
[18:03] <jamespage> you should end up with a dnsmasq instance running on the gateway which dishes DHCP
[18:05] <adam_g> jamespage: yup. all *looked* okay. there was dnsmasq running with a host file that was updated when instances came and went, but DHCP appeared not to be working from the instances.
[18:05] <adam_g> ill redeploy soon and have another look
[18:06] <jamespage> adam_g, OK
[18:06] <jamespage> worth checking the tunnels have setup ok with 'ovs-vsctl show'  - you should see one for each endpoint on quantum-agent and nova-compute nodes
[18:07] <jamespage> adam_g, I stuck in a fix for the python-dnspython problem for all three charms - see my comment in the bug report as to why this is required
[18:07] <adam_g> jamespage: ack
[18:13] <nidazole> can anyone help me about  how to combine 3 dd images from 3 SCSI physical disk to one image? the images got through RAID controller. i am trying to launch those to VMWare!!
[18:14] <patdk-wk> heh?
[18:15] <patdk-wk> your in the wrong place then
[18:15] <patdk-wk> and if you combine them into one image, you will mess it all up
[18:15] <sarnold> nidazole: (a) prepare for, and mentally accept, that you cannot do this. (b) _maybe_ there are md or dm tools that might be able to do this, but it'll probably be annoying, difficult, and you may be the first...
[18:15] <patdk-wk> you need to convert them to vmdk files
[18:15] <patdk-wk> oh wait
[18:15] <patdk-wk> your attempting to join 3 disks from a raid?
[18:16] <patdk-wk> ya, I did that last month
[18:16] <patdk-wk> you have to write a program to do it :) or use a very expensive windows program
[18:16] <sarnold> patdk-wk: hardware raid controller? :)
[18:16] <patdk-wk> I missread images got through raid, as images not through raid
[18:17] <nidazole> md didnt work
[18:17] <patdk-wk> in my case, hardware raid died
[18:17] <patdk-wk> why would md work?
[18:18] <nidazole> how to convert split dd images to vmdk?
[18:18] <nidazole> LIVEVIEW?
[18:18] <patdk-wk> nidazole, you write a program to do it :)
[18:18] <patdk-wk> and define split
[18:19] <patdk-wk> I seriously doubt your *raid* was setup for concat
[18:19] <shauno> qemu has a helper tool qemu-img which is fantastic for converting between the various formats
[18:19] <patdk-wk> yes, that will come after you joined the 3 drives back into 1
[18:20] <nidazole> patdk-wk: ok :)  how?
[18:20] <patdk-wk> make a program :)
[18:21] <sarnold> patdk-wk: did you need documentation from the raid card vendor or were you able to figure out the data format by inspection alone?
[18:21] <patdk-wk> sarnold, I didn't need anything
[18:22] <patdk-wk> I knew how I setup the raid, so I knew the params, and just wrote a program to unstripe the data
[18:22] <patdk-wk> also added sparsefile support into it, cause I have a habbit of zero'ng my disks, so it saved a crapload of time
[18:22] <sarnold> patdk-wk: there was no metadata on the drives that you had to skip? no funny geometry problems? :)
[18:22] <patdk-wk> in my case, raid10 setup with 128k blocks
[18:22] <sarnold> patdk-wk: hahahaha
[18:22] <sarnold> patdk-wk++
[18:23] <patdk-wk> no metadata, if there was, it was at the end of the disks
[18:23] <patdk-wk> dumped finished image to new disk, and reboot :)
[18:24] <sarnold> patdk-wk: hooray for documenting / remembering your setup. Nice. :)
[18:25] <patdk-wk> I also recovered a raid50 setup, last jan
[18:25] <patdk-wk> that was much more fun :)
[18:26] <patdk-wk> raid50 with 7 disks, 3 bad disks, raid card totally screwed
[18:27] <patdk-wk> finding working disks, getting them in order, recover each of them using ddrescue
[18:27] <patdk-wk> took almost a week
[18:28] <patdk-wk> those had metadata, but it was easy to locate a *correct* mbr, and then adjust for it
[18:28] <sarnold> ooo
[18:29] <patdk-wk> so the question still remains, how did he have his raid setup? raid1 with spare? raid5? maybe raid6 but I doubt it
[18:30] <nidazole> thanks, patdk-wk.
[18:30] <nidazole> about expensive programs?
[18:32] <nidazole> what programs?
[18:32] <nidazole> and what kind of params i need to write a program?
[18:36] <patdk-wk>  is what I used at one time: www.ufsexplorer.com/download_pro.php
[18:38] <nidazole> patdk-wk: can i combine the images with this?
[18:39] <patdk-wk> I think so, been a long time since I used it
[18:40] <nidazole> patdk-wk: and one more question please. what kind of params i need to write a program?
[18:40] <patdk-wk> heh
[18:41] <nidazole> just heh?
[18:41] <nidazole> :)
[18:41] <patdk-wk> ask an understandable question
[18:42] <nidazole> i cant, but thank you very much.
[18:43] <nidazole> very usefull to me.
[19:13] <hallyn> stgraber: for lxcapi get/set_cgroup_item, i was going to pass in a container obj and use the container's cgroup.  Not name a cgroup.  Based on your email i'm not sure that's what you mean.  di dyou mean to pass a cgroup name in?
[19:14] <stgraber> hallyn: oh yeah, my e-mail was a bit confusing, I guess the C implementation would be something like get_cgroup_item(c, subsystem, key)
[19:16] <hallyn> kewl
[19:16] <hallyn> (actually it'll be 'int (*get_cgroup_item)(container, char *subsys, char *retv, int inlen)
[19:17] <stgraber> ok, same as get_config_item, works for me :)
[19:18] <hallyn> stgraber: exactly :)  thanks
[19:18] <stgraber> after the binding, I expect it to be possible to do things like lxc.Container("p1").get_cgroup_item("blkio", "blkio.throttle.io_service_bytes") in python which would return the content of /sys/fs/cgroup/blkio/lxc/p1/blkio.throttle.io_service_bytes
[19:19] <stgraber> it's unfortunate that we need to repeat the blkio part in the key, but well, that's how cgroups are and it's not worth working around :)
[19:29] <cryptonite> hello
[19:30] <mysteriousdaren> hi cryptonite
[19:31] <cryptonite> i was interested in setting up a home mail server, what do you think will the rasberry pi be enough or would it be better to run it on a vps?
[19:33] <shauno> I'd recommend against the Pi simple because residential connections and smtp don't mix well
[19:33] <sarnold> cryptonite: storage is probably the weak point, I'm not sure I'd want a mail spool on sd or usb..
[19:34] <sarnold> good point shauno :)
[19:35] <cryptonite> shauno: what do you mean?
[19:36] <cryptonite> i have also an atom setup which could be used too
[19:36] <shauno> cryptonite: many isps block smtp on residential connections, because the primary use-case for it is trojan'd windows boxen.
[19:36] <shauno> and I have to say I agree with them.  the valid few are far, far out-numbered
[19:37] <sarnold> … and many other ISPs also put their dynamic IPs on RBLs to help _others_ block their users as well :)
[19:38] <cryptonite> i think i could disable the smtp port blocking for my connection...
[19:39] <cryptonite> but you say that other people might not get the emails i send them?
[19:39] <Ng> so the cloudimg images, they boot with console=tty1 and console=ttyS0, and it seems like it boots with the kernel displaying tty1, but in the plymouth world that means you just see kernel messages and not anything from/above init. Is there an easy way round that? I'd quite like to see all the output
[19:40] <sarnold> cryptonite: yeah, that's happened to me before...
[19:40] <shauno> cryptonite: it's quite possible that the more stricter recipients might not look so kindly on such connections, yeah.  which is less relevant if you're only taking inbound
[19:40] <shauno> but I/O is still a consideration on the Pi specifically
[19:44] <cryptonite> i would like to be able to do both, send and receive
[19:56] <zastern> Anybody using ActiveMQ or mcollective on ubuntu? I'm getting a weird error that seems to be specific to the ubuntu packages - when I try to start with /etc/init.d/activemq console activemq, I get ""File /etc/activemq/instances-enabled//activemq.xml not found." Note the double slash.
[19:58] <sarnold> zastern: how about "service activemq start" or similar?
[19:59] <zastern> sarnold: it does start, and I see the java process with ps aux, but it's not listening on the specified port
[19:59] <zastern> thats why im trying to do service activemq console to get debug info
[19:59] <zastern> i already disabled ufw so its not that
[20:00] <sarnold> zastern: is it bound to localhost:port or a specific IP:port, just not the one you need?
[20:06] <zastern> sarnold: its not bound to anything
[20:06] <zastern> netstat -a shows nothing for it
[20:07] <sarnold> zastern: darn. :(
[20:08] <sarnold> zastern: is there anything in the logs?
[20:08] <zastern> sarnold: no, thats why im trying to use console!
[20:08] <zastern> haha
[20:08] <zastern> console is debug output
[20:08] <zastern> tehres no logs at all from what i can tell
[20:09] <sarnold> zastern: hehe, I figured you'd look there first, but it never hurts to double-check :/
[20:22] <adam_g> jamespage: still around?
[20:53] <sh_t> has anyone here had issues with ubuntu hanging on shutdown/reboot? "Unmounting local filesystems..." is what i'm seeing. fresh installation with 16 JBOD disks setup in fstab.. all working/test mounted prior to reboot
[20:53] <sarnold> sh_t: sync can take a while..
[20:54] <sarnold> sh_t: run 'sync ; shutdown -h now', and see if it is still slow at the same point?
[20:54] <sh_t> server was idle
[20:54] <sh_t> but i'll try that
[20:54] <sarnold> hrm.
[20:55] <sh_t> i read some thread about it on the ubuntu forums.. some people having to put "killall"'s into /etc/init.d/umountroot
[20:58] <jamespage> adam_g, yep
[20:58] <guntbert> sh_t: I take a guess: DELL ?
[20:59] <sh_t> guntbert: no, supermicro
[20:59] <sh_t> guntbert: it doesn't happen in 10.. only 12 (which ive decided to start using recently)
[21:00] <guntbert> sh_t: then its probably different from what I experienced, I remember (answered) questions around that topic on askubuntu
[21:05] <jamespage> adam_g, juju set quantum-gateway ext-port=eth1
[21:06] <jamespage> and magic will happen
[21:06] <adam_g> jamespage: okay, i think thats what i probably needed. i had everything up on quantal. instances were getting their IPs, but iw as unable to reach them myself.
[21:06] <adam_g> jamespage: don't think i had ext-port set, ill try that next time.
[21:07] <jamespage> adam_g, you can set it now
[21:07] <adam_g> jamespage: do floating ips (created and associated via quantumclient) work as expected?
[21:07] <jamespage> adam_g, yes
[21:08] <jamespage> adam_g, the folsom/stable branch also includes the integration between nova and quantum so you don't have to use quantum at all
[21:08] <jamespage> adam_g, however the charm does not support that yet
[21:08] <jamespage> thats for next iteration
[21:08] <adam_g> jamespage: cool
[21:09] <adam_g> jamespage: i just put up some MPs for the swift charms to get things rolling there. just migrated to using the openstack-common stuff for cloud archive installation.
[21:09] <jamespage> adam_g, you can actually see the instance from quantum-gateway
[21:09] <jamespage> "sudo ip netns exec qrouter-5ba6e7b8-5bdd-4cb7-8a3a-361561d8e771 ping 192.168.21.131"
[21:10] <jamespage> quantum uses ip namespaces heavily on the gateway
[21:10] <jamespage> adam_g, did you take the deb out of the charm?
[21:10] <adam_g> jamespage: but from elsewhere its accessible via 192.168.21.131?
[21:10] <adam_g> jamespage: yup. that gets installed now from the archive, if folsom (or later) is being installed
[21:10] <jamespage> adam_g, once you set the ext-port for the gateway yes
[21:11] <jamespage> all that does is add the port to the bridge and set the link to up
[21:11] <adam_g> im half tempted to rewrite the swift stuff in python. theres to much config file sync'ing that could be done better using templates
[21:12] <jamespage> adam_g, I need to sync up the quantum charm to work with common as well
[21:15] <adam_g> jamespage: i think we can get your stuff merged first, and then sync into there. i mentioned in one of those reviews, id like to have just one configure_network_manager() we can use from both nova-compute and nova-c-c.
[21:15] <jamespage> adam_g, yeah +1 to that
[21:15] <jamespage> adam_g, can we make the commons project an official one as well please....
[21:22] <adam_g> jamespage: sure.
[21:22] <jamespage> adam_g, looking at you comment re testing for IP in compute/controller
[21:28] <adam_g> jamespage: https://launchpad.net/openstack-charm-helpers
[21:29] <jamespage> adam_g, nice
[21:34] <jamespage> adam_g, I just pushed a fix for the [[ -n private_address ]] bug to cloud/compute branches
[21:35] <zastern> Anybody running into issues getting activemq running on ubuntu for mcollective? The package seems possibly broken.
[21:39] <adam_g> jamespage: cool. i wanna do another deploy and then ill merge those
[21:48] <jamespage> adam_g, ack - thats me done for today - ttfn
[21:56] <adam_g> jamespage: cya
[21:59] <adam_g> Ursinha: ping
[22:14] <Ursinha> adam_g, pong
[22:24] <adam_g> Ursinha: er, just sent an email. :)
[22:25] <Ursinha> adam_g, will reply :) do you want/need the answer now or tomorrow is fine?
[22:25] <adam_g> Ursinha: no rush. tomorrow's cool
[22:25] <Ursinha> adam_g, cool :)
[22:54] <hallyn> stgraber: bleh!  what a pain!  you can't get size of a cgroup file through stat or seek.
[22:55] <hallyn> i sort of knew that, but was hoping someone had fixed it by now :)
[22:57] <jeeves_moss> if I have the %maildir% dump from my pooched dovecot install, is there a way to get the e-mails back out of the database so I can import them into a mail program, and then offload them to Exchange?
[23:01] <Patrickdk> heh?
[23:01] <Patrickdk> dovecot doesn't do databases
[23:02] <Patrickdk> you shouldn't even import them
[23:02] <Patrickdk> you should just setup an imap program and transfer them
[23:02] <stgraber> hallyn: oh, that's fun, so how do you read those files? just read until you get nothing?
[23:04] <jeeves_moss> Well, the issue is that the server crashed, and I only have the backup of the maildir directory.
[23:05] <hallyn> stgraber: yup! :)
[23:05] <hallyn> stgraber: so i only do that if null pointer is passed in or 0 length
[23:06] <hallyn> if you pass in a small value your indication will be return value == what you passed in, not the full size as with lxc_get_config_item
[23:11] <Patrickdk> jeeves_moss, since you only backed up the maildir, you have to rebuild the rest of the server
[23:11] <Patrickdk> once it's build put the maildir files on it
[23:11] <Patrickdk> then you can transfer it to exchange, or whatever you want
[23:12] <Patrickdk> the space savings you got, from only backing up the maildir, is the cost of restoring things, when this happens
[23:12] <sarnold> many mail clients can grok maildir; maybe aim mutt at the maildir, and configure it to know how to imap to the exchange server, and then 'copy' the mails to a folder on the imap...
[23:13] <jeeves_moss> Patrickdk, and if the users were in a MySQL database, I'm guessing I'm going to have to totally rebuild EVERYTHING in order to get the mail out of it?
[23:13] <Patrickdk> getting access to the mail is easy, it's one file per email
[23:14] <Patrickdk> the issue is, you can't give that to exchange
[23:14] <jeeves_moss> Patrickdk, or is there a way to set it up that I can run a shell program to login to my exchange server and upload all the e-mails to the "inbox"
[23:14] <Patrickdk> you can do an imap transfer, but you need to setup an imap server for that to work
[23:14] <Patrickdk> I don't know of any maildir->imap program
[23:14] <Patrickdk> only imap->imap
[23:14] <jeeves_moss> Patrickdk, I know that each e-mail is just a file.  Can I "import" that into the local directory in Thunderbird or something?
[23:15] <Patrickdk> dunno about thunderbird
[23:15] <sarnold> mutt should do it
[23:15] <jeeves_moss> Patrickdk, I'm just trying to thin of the fastest way to move the emails around.  I'd really like to take that old server out to the parkinglot and smash it.
[23:16] <jeeves_moss> sarnold, mutt will, but I'm not sure how to move the e-mails around.  I had migrated ~50 users to Exchange to shut them up while I sorted out the other mess
[23:17] <sarnold> jeeves_moss: T lets you tag by patterns. ;C will copy all tagged messages to a folder.
[23:17] <jeeves_moss> sarnold, thanks!
[23:17] <sarnold> jeeves_moss: it may not be pretty :) but if you don't want to set up dovecot again and use an imap<->imap, it might be the thing to save your bacon...
[23:18] <jeeves_moss> sarnold, lol, lets hope.  As I said, it was a v-hosted dovecot/postfix/MySQL mess that I had thrust on me.   I was sick of explaining why M$ products didn't work.  So, I just built an exchange server and everyone shut up
[23:18] <sarnold> jeeves_moss: muttrc(5), search for "Constructing Patterns" to see the rules available...
[23:18] <jeeves_moss> sarnold, thanks.
[23:20] <jeeves_moss> sarnold, now I'm going to see if I can wake it up using WOL
[23:20] <sarnold> jeeves_moss: plus, you'll have fun when someone wants a mail "out of backups" and you say, "okay, it'll take me about four hours to clone this exchange, an hour to load the database off backup, I'll get it to you ... tomorrow." :)
[23:22] <jeeves_moss> sarnold, lol.  I've attached a few file boxes to my door.  and the note above it says "If you need something, submit a form (pick a box).  If you've submitted a form, don't bother asking me in person.  They will be accomplished in order of need of the group, not in order of personal preference.  Asking me will result in your request being moved further down the pile"
[23:23] <sarnold> jeeves_moss: oof :)
[23:24] <jeeves_moss> a few people have asked, and I have taken their request form and physically moved it to the bottom of the pile while they were standing there.  Under the pile is a garbage bucket.  They're warned that the second request will be moved lower than the pile
[23:25] <jeeves_moss> damn it.  I can't find the MAC of the server that's turned off
[23:29] <patdk-lap> sarnold? heh?
[23:29] <patdk-lap> you NEVER clone exchange to get an email out of backups
[23:29] <patdk-lap> you just mount the backup, and restore the email
[23:30] <sarnold> patdk-lap: that's how it was explained to me last time I asked an exchange admin how they got mails out of backups.
[23:30] <patdk-lap> this was echange pre 2003?
[23:30] <patdk-lap> I know in 2003 there was a backup mailspool, you restored the backup to it, then did some recovery thing
[23:30] <patdk-lap> in 2007/2010, you just restore and copy over the active mailspool
[23:31] <patdk-lap> exchange does some voodoo so it doesn't actually overwrite it
[23:31] <patdk-lap> I've done this a few times in the last couple of months
[23:31] <patdk-lap> users completely wiping out their exchange mail somehow via their iphone
[23:32] <patdk-lap> not nearly as nice as a maildir email restore
[23:33] <patdk-lap> but I have even dropped maildir/mbox on my mail servers, for mdbox
[23:33] <sarnold> patdk-lap: yeah, probably 2003, 2004, I asked about it...
[23:34] <jeeves_moss> sarnold, if I have e-mail cached in Thunderbird, is there a method to sync it back to the server?
[23:34] <sarnold> jeeves_moss: not that I know of, but I know nearly zero about thunderbird.
[23:34] <patdk-lap> the only thing I know about thunderbird is it stores email internally using mbox format
[23:35] <jeeves_moss> sarnold, lol.  Well, I think when I get home, I'm going to fire up that old box, and strip all of the crud out of it.