[01:39] http://isup.me/old-releases.ubuntu.com says "It's not just you! http://old-releases.ubuntu.com looks down from here." [01:40] Where should I look for info about this? [01:40] (I want to install python3.1 on a couple of not-quite-dead lucid servers.) [01:40] Looks like it's up at this moment? [01:40] andrewII: heh, I see that too now [01:41] Networks are fickle [01:41] It was down five minutes ago [01:41] I don't doubt it in the least :) [02:13] For MAAS, where are the installation logs for deployments? [02:17] I'm really hoping it's not /var/log/maas/rsyslog/././messages as that's just a giant pile of ureadahead lines [02:19] andrewII: could you just do a find / -xdev -mmin -10 or something, to find the newest/oldest files [02:21] twb: thanks - that's the next thing I'll try (I just rediscovered `sed /asdf/d | less` which is helping: apparently the node can't see the internet?) [02:22] yeah that or grep -v [02:22] The chief advantage of sed //d (over grep -v) is that it won't return an error exit status if no lines are removed [02:23] Son of a... one of these days I need to just read grep's man page all the way [02:23] Yeah, that would have been perfect === pavlushka is now known as Guest78383 === Guest78383 is now known as pavlushka [04:40] Hey all, I'm having some trouble getting outgoing mail set up on my Ubuntu server. I haven't actually looked into postfix settings yet, because i can't successfully `telnet gmail-smtp-in.l.google.com 25`. Would love some help on this, I'm thinking my firewall settings are off. Here's my iptables-save: http://codepad.org/gMJAp8tH [04:41] * sarnold hugs jdeeburke [04:41] finally someone testing the basics first [04:41] jdeeburke: do you want to be a "leaf" MSA that just sends everything to gmail? [04:41] jdeeburke: or do you want to be a first-class MTA that does all the MTA things [04:41] jdeeburke: chances are good your ISP blocks outgoing traffic to port 25 [04:41] @sarnold not likely, this is running on DO [04:42] In the first case, I strongly recommend using msmtp, and for gmail using 587/tcp ("submission"), with TLS and auth. [04:42] jdeeburke: "maybe" https://www.digitalocean.com/community/questions/port-25-smtp-external-access [04:42] @twp I'm not really sure of the differences, it just needs to be really basic. Sending mail via postfix from a webapp [04:43] jdeeburke: msmtp would be fine for that, except if you're sending it via gmail, the From will be set (by gmail) to your gmail account [04:43] It's for a temporary basis so it really doesn't need to be perfect, it will be gone in a month or so [04:44] jdeeburke: that may fit msmtp -- it's a blocking program, not a queueing program, so it may not be a perfect fit, but it's worth a quick look [04:44] Being an MTA is a lot of work to get right; being an MSA is much much simpler [04:44] There are other "nullmailers", msmtp is just my personal preference [04:44] sarnold: I'll open a ticket to check on that, I had no idea they sometimes blocked access. Had no other problems on my account though [04:45] twb: correct me if i'm wrong, but even switching from postfix (which I don't plan to do), wouldn't solve my problem since I can't even telnet on port 25 [04:45] jdeeburke: MSA talks on 587, not 25 [04:45] 25 is only for MTA-to-MTA traffic [04:46] Oh ok. This is an area I have no knowledge in, in case that wasn't abundantly clear already :P [04:47] try telnet smtp.gmail.com 587 and see :) [04:47] No luck on 587 either [04:47] In what way does it fail? [04:48] times out [04:48] SMTP conversations are initiated by the client side [04:48] So if you don't say anything, the server will hang up [04:48] (This is the same as HTTP and different from SSH.) [04:48] gmail prints a banner immediately [04:48] 220 smtp.gmail.com ESMTP i7sm1770483paf.9 - gsmtp [04:49] Oh- oops [04:49] yeah you should receive a 220 if you connect, I tested that on my local machine [04:49] You're right, I was confused [04:49] I was thinking of the 250 in response to the EHLO [04:50] Also, I finally remembered how to drive supybot: [04:50] /msg knoba whatis #postfix nullclient [04:51] Do my iptables look off to y'all? [04:51] http://codepad.org/gMJAp8tH [04:51] sorry, I never figured out iptables :( [04:52] -s 127.0.0.1 should be -i lo [04:52] The --sport cases would be silly in a stateful firewall, but you're not using conntrack [04:53] I would also disable mysql TCP support entirely -- use AF_UNIX sockets for local connections [04:54] Otherwise it looks OK to me [04:54] Damn [04:54] #netfilter is the best place to ask about linux firewalls, here is one I prepared earlier: http://www.cyber.com.au/~twb/doc/iptab [04:54] Yeah, I asked my question in there too, but no response so far [04:56] Hmm [04:56] Just tried telnet'ing port 22 on another server of mine, and getting the same timeout [04:57] That sounds like your firewall, ya [04:57] Either debug it or just make it stateful, so that return packets are always allowed [04:58] oh [04:58] you're not allowing any ICMP at all [04:58] That means PMTUD won't work [04:59] And reject messages also won't arrive [04:59] twp: i'm googling so many things talking to you [04:59] https://en.wikipedia.org/wiki/PMTUD [04:59] The short version is that if you don't allow *some* ICMP through, your system won't work properly [04:59] So do i need something along the lines of: -A PRELUDE -p icmp -j ACCEPT [04:59] That's what I do, but it's broader than really necessary [05:00] It's common to rate-limit incoming ICMP echo requests (i.e. pings). [05:03] I just added icmp -j accept to test, but still can't telnet [05:04] OK so either tcpdump (on both ends), or add -j LOG rules (or -j-less rules & watch packet count), or disable firewall entirely for a bit [05:09] Alright. I'll go back into my cave until I get totally stuck again. Thanks so much for the help twb, it's much appreciated [05:09] You are very welcome [05:09] jdeeburke: one option, is to use a front end like ufw [05:09] (and sarnold too) [05:09] it is a refreshing change for someone to start by posting iptables-save instead of yukky iptables -L ;-) [05:09] you're welcome, hope you sort this out :) [05:09] twb: -and- testing the underlying connections first :) hehe [05:10] twb: ;) I may not know much, but I know enough to not be a complete idiot [05:12] sarnold: I actually was just about to look into using ufw instead. 90% of my day-to-day server management is obfuscated and I'm not sure the time investment right now is worth it to chase this rabbit [05:13] jdeeburke: yeah, I understand completely; sometimes it's worth knoiwing something -all- the way through, and sometimes a nice front-end is perfect :) [05:13] it's sometimes hard to know which is which, but ufw is nice enough and easy to replace later if you wish [05:14] ufw is beneficial if you need the firewall to be modular enough that end users, or packages, can extend it [05:17] Soooo [05:17] deleted all rules disabled firewall completely, still no telnet. [05:17] Gonna open a ticket w/ DO [05:18] At that point you need to check "ip r", "ip a", and obscure things like window scaling [05:18] And DNS resolution, obviously, if you're referring to hosts by name [05:19] Also have assumed 100% IPv4 so far, but if there's IPv6 involved you need to care about all that [05:23] So IPv6 does seem to be involved, at least in my testing [05:23] gmail-smtp-in.l.google.com resolves to the ipv6 address in testing [05:23] with no firewall rules, it times out. If I resolve the ipv4 address and telnet that, it works like a gem [05:23] if I restore my rules and telnet the ipv4, it times out [05:24] Why are you using that gmail-... instead of smtp.gmail.com ? [05:24] i'm not sure. just something i found online that someone else used to test. [05:25] Same result though [05:25] http://askubuntu.com/questions/32298/prefer-a-ipv4-dns-lookups-before-aaaaipv6-lookups [05:25] (Not sure if that's a good idea, but it'll address the immediate symptom) [05:39] twb: Alright I think i've got it sorted, let me know what you think about this: [05:39] disabled firewall, I'm able to telnet to ipv4. postfix sends successfully [05:39] re-enabled my rules and it's broken again [05:39] added a rule for 587, no working still [05:39] removed all my 25 and 587 rules and replaced with this: [05:39] iptables -I INPUT -m state --state RELATED,ESTABLISHED -j ACCEPT [05:40] Don't use -m state [05:40] -m conntrack --ctstate ESTABLISHED [05:40] (and RELATED if you use things like FTP or SIP) [05:41] *shudder* ftp [05:41] And yeah - don't use FTP :-) [05:41] Oh not to worry, no FTP on this or any other servers I run... [05:41] 14:54 [...] here is one I prepared earlier: http://www.cyber.com.au/~twb/doc/iptab [05:44] http://sprunge.us/LfQG?ini is an even simplierified one [05:45] * jdeeburke hugs twb === chmurifree is now known as chmuri [05:45] Alright I think I'm all set [05:45] ha :) nice [05:45] just successfully sent mail with all firewall rules enabled [05:48] thanks to everyone involved; 10/10 would do again ;) [05:49] 10,000 schrute bucks to all [05:49] \o/ [05:49] NOW to do the other thing I started doing 2 hours ago === Guest97201 is now known as lordievader [07:14] Good morning === Guest4584 is now known as aki === lordievader is now known as Guest20559 === aki is now known as Guest29539 === Guest29539 is now known as _aki === _ruben_ is now known as _ruben === Guest20559 is now known as lordievader === _aki is now known as aki [09:28] Hello guys, can you please tell me if there will be packages for Openstack Newton for Ubuntu Trusty release available in cloud archive? === iberezovskiy|off is now known as iberezovskiy [09:46] hello, folks [09:46] I've been trying to get trust's cloud image to run with kvm [09:47] but none of the tutorials I found have worked for me [09:47] does anyone have pointers for this? [09:50] Ubuntu's wiki has a tutorial which uses a local cloud-init file. Tried that one, can't log in [09:52] I wanted to set some environment variables permanently for them to be consumed by a process. So I added the variables to /etc/profile, and added exit 0 at the end. Now I cannot SSH into the server. The connection closes as soon as it is established [09:56] try sftp-ing into your server. does that log out immediately too? [09:56] alternatively, assuming you're on a linux machine, you can use scp [09:57] probably not [09:58] I can use scp but I need to make changes to the server, which requires SSH [09:59] Can I replace /etc/profile.d/ using scp? [09:59] winterfell, ^ [09:59] no, I thought you might be able to copy over /etc/profile, but you would not be able to write it back, so that won't work [10:00] depends on the permissions [10:00] did you create the script as su or as a regular user? [10:01] jgrimm, it's still in a silo, and still in progress. it comes with regressions and needs a fixup (revert) [10:01] It is root owned. So I guess that is not going to work [10:01] Could it be something else altogether? I am guessing it is the exit 0 statement at the end [10:02] it is most likely that. As a new shell is being created, it is exited as soon as the scripts in /etc/profile.d are being applied [10:02] gargsms, ^ [10:05] Yes, most likely. I can sftp. Let me try changing the file with SFTP [10:08] Not allowed to write there. Permission error. :( [10:11] hi, i'm setting up an openstack environmant on a local server - conjure-up openstack ... how can I skip the pre deployment tasks to repeat the final steps? [10:11] Hi [10:12] is there anybody around? I could use some help with conjure-up [10:13] Hi Teme :-) I need also help - already questioned but until now no answer - maybe we have to wait :-) [10:13] maybe I can help you? [10:14] hi, i'm setting up an openstack environmant on a local server - conjure-up openstack ... how can I skip the pre deployment tasks to repeat the final steps? [10:15] deltatree: If you could it would be great! I've been trying to conjure-up a nova-lxd environment, but finalizing the installation goes to error everytiime [10:16] in which state are your containers? [10:16] specifically step 3 keypair. It's giving me an error, "Could not read output from step " [10:16] juju status [10:17] all of the containers are up and running, I can access the openstack-dashboard with browser and juju-gui is responding as well [10:17] could you send me the output of "juju status" [10:17] sure [10:19] gargsms, do you have root credentials for this server? === degorenko is now known as degorenko_away [10:29] someone here who can help me with conjure-up [10:29] deltatree: sorry, my collegue just walked in, heres my juju status http://paste.ubuntu.com/23177476/ === G_ is now known as G [10:39] deltatree: actually now that i'm thinking this, I think I have pretty much the same question as you do. How to selectively execute the final steps [10:43] is everybody at lunch still? Or is it normal that its this quiet? [11:09] @teme: they are sleeping ... :-) [11:09] :) === olegb_ is now known as olegb [11:25] @teme: https://gist.github.com/anonymous/9eabdf74bf72fa26287aaed18e281140 [11:26] with that you can reset juju and restart conjure-up with a clean env [11:35] deltatree: ok, thnx. I'll try again. [11:35] deltatree: so you think that my problem is more of a missed configuration issue? Or is this just fyi? [11:38] deltatree: I've been doing the same thing with just "juju models" - "juju controllers" - "juju remove-model XX" - "juju kill-controller xx". Havent actually never just removed that cache or config [11:53] How to know offet value, starting value end value of P.E, And which particular P.E is being used in LVM anyone? [13:10] is there a way to manually launch openstack nova-lxd neutron networking creation scripts? [13:52] jamespage: I can't make the time on todays call in about 2 hours from now - I don't hae anything on dpdk to add, is there anything you want to know from me inadvance or is all fine? === Daviey_ is now known as Daviey [14:15] so, the latext Xenial images for vagrant have broken resolvconf too [14:16] seriously, is there anything I can do to help here [14:16] seems perpetually in one state of broken or abother and I can't see why it's being released if so [14:16] a few BATS tests would show the level of brokenness before shipping [14:17] i.e can't get dns servers from vagrant as /etc/resolv.conf is a file, not a symlink to resolvconf [14:17] so resolvconf -u does nothing [14:17] and therefore empty resolver list and no resoultion == broken provisioning [14:18] having to add loads of extra cruft to just work around broken images and it just feels wrong [14:18] could make my own in packer and be done with it, but conisdering there's been 22 millions downloads of trusty.. you'd think having them working for xenial would be something worthwhile [14:19] (and not using chef's bento or puppetlabs bo etc etc etc) [14:24] cpaelzer, nah - thats fine I'll just use it as a catchip on 2.6 release [14:25] Odd_Bloke, ^^ [14:25] re joelio's broken xenial vagrant image [14:25] it's not mine, it's ubuntu's :) [14:25] but yea, sure :) [14:26] joelio: also, do bugs exist for the issues you're hitting? [14:26] anyway I can help, I'm happy too - makes more sense to fix upstream than run another set of packer json that only I see [14:26] rbasak: I can raise some, it got updated a few days ago and resolv.conf broke there [14:26] before that it was a bug in naming, so you couldn't run more than one instance etc [14:27] joelio: please do. I'm not directly involved but I'm sure the team that looks after it would welcome your help. [14:27] before that it was vboxsf so /vagrant mount failed [14:27] rbasak: will do, ta [14:27] could mop these up before shipping with a better test harness afaict [14:28] joelio: There is a bug filed, and I believe the fix is in xenial-proposed. [14:28] joelio: thanks! https://bugs.launchpad.net/cloud-images may be the right starting place unless someone knows of somewhere better. [14:28] joelio: perhaps you could contribute tests? [14:29] more than happy! [14:29] will go over and check it out, I've got fixes for all this, but just irks that it's not in upstream :) [14:29] appreciate this is best efforts stuff but I use it quite a lot and always catches me out after a new box update [14:32] Odd_Bloke: do you have an id for the resolvconf bug ooi? [14:33] joelio: Not off-hand; will find it once I'm out of my current meeting. :) [14:33] sure thing, understand [14:33] just don;t want to raise a new if one exists [14:59] jamespage, hey man, I'm talking with Christian Ehrhardt by e-mail, about new DPDK and OVS, I think that we'll need to backport those two, from 16.10, to 16.04, via Newton Cloud Archive, can you do it? [15:00] nice, just talking about ovs+dpkg at $work today [15:00] ovs/dpdk rather [15:01] that going into Mitaka? [15:11] joelio: https://bugs.launchpad.net/cloud-images/+bug/1621393 [15:11] Launchpad bug 1621393 in cloud-images "xenial64 image (20160907.1.0) has a broken (empty) /etc/resolv.conf" [Undecided,New] [15:11] cheers Odd_Bloke [15:11] joelio: So semiosis has been doing a lot of the work to get the xenial box back in to good shape. [15:12] good to hear :) [15:13] I would like to install openstack on ubuntu using conjure-up https://insights.ubuntu.com/2016/06/09/conjure-up-an-introduction/ - what is the recommended HW spec for this? how much RAM? disk ? CPU ? - thank you [15:15] JonathanGe: I think that depends on how many resouces you want to manage from your infra.. as a guide line a basic devstack instance with bare bones is about 6GB RAM, just to stand it up [15:15] not sure what conjure-up is recommending though, sure guys and gals here could chip in [15:16] conjureup is openstack on lxd [15:18] http://conjure-up.io/docs/en/users/#getting-started seems to answer my Q [15:19] joelio, yes, OVS with DPDK is available on Xenial (not sure about Mitaka integration) but, it is super unstable. [15:20] ThiagoCMC: oh, appreicaiate that. Literally been watching videos from the dpdk conference today and the "up to 8x better perf than vanilla ovs" seemed enticing. Look forward to testing it out :) [15:21] JonathanGe: ah neat, uses ZFS too [15:24] Hello I have tried to resize my logical volume. And it seems like i am failed to do so http://paste.ubuntu.net/23178304/ [15:28] bipul, looks like you did [15:28] but cause you said Yes, destroy my filesystem by making it smaller, you did just that, and destroyed your filesystem [15:32] sorry Shrinking the logical volume. patdk-wk [15:32] To reduce the size of a logical volume, first unmount the file system. You can then use the lvreduce command to shrink the volume. After shrinking the volume, remount the file system. [15:32] And that is what i did. [15:32] *can* [15:32] how can you do that? [15:33] when you just cut off the end of the filesystem? [15:47] cpaelzer: do you remember the uvtool bug you reported that only affected Precise? I can't find it. [15:51] cpaelzer: ah, found it. Bug 1620633. [15:51] bug 1620633 in uvtool (Ubuntu) "qcow image created in precise is not usable" [Low,Won't fix] https://launchpad.net/bugs/1620633 [15:52] cpaelzer: does that mean that uvt-kvm is completely broken on Precise, or just when using the libvirt from the cloud archive or something like that? [15:57] joelio, so far, here is the big problem that I found when trying OVS+DPDK: https://bugs.launchpad.net/ubuntu/+source/openvswitch/+bug/1577256 [15:57] Launchpad bug 1577256 in dpdk (Ubuntu) "The OVS+DPDK dpdkvhostuser socket bridge, only works as expected, if the KVM Guest also have DPDK drivers loaded" [Medium,Expired] [15:58] From what I've read on DPDK / OVS mail lists, this will be always a problem, if yes, then, OVS+DPDK will be useless in a cloud enviromnent, or even on a KVM-Only env, simple because one VM interferes with the others in a very bad way. [16:03] Well, tomorrow I get to give a talk to our research dept on useing Linux in research, and will be useing the Ububtu servers I have built for them as examples :) [16:04] Gonna try to plug the maintainers of the packages we use, make a case for some support of some kind [16:14] Hello! I am looking at getting a ODROID-C2 (ARMv8), can I install the normal Ubuntu server 16.04 arm64 iso? If so, why does Odroid the company provide their own ISO? [16:15] ThiagoCMC: ok, I see.. is there anything else happening in the dn space (netmap or vpp etc) that looks more promising for VM-to-VM? [16:15] obviosuly relying on users maintaing their own dpdk is never going to happen [16:15] ThiagoCMC: thanks for the info, all useful [16:16] Hello all, I am trying to use syslog-ng on Xenial and I am finding that kernel boot messages are not being logged. after fixing the syslog.socket symlink to point to syslog-ng instead of rsyslog I am getting systemds boot messages and later kernel messages.. any idea on what I am missing to get early kernel boot messages, like the ones in dmesg? [16:17] joelio, I'm still waiting to see if OVS+DPDK can be used on OpenStack Compute Nodes... [16:17] ThiagoCMC: it'd be greay if it could [16:17] s/greay/great/ - fingers not with it today [16:17] About netmap, well, I filled another bug report about it: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1475078 [16:17] Launchpad bug 1475078 in linux (Ubuntu) "VirtIO (and probably other modules as well) is built-in, make it modular..." [Wishlist,Fix committed] [16:17] ThiagoCMC: if you need testers, let me know [16:18] Ok! :-) [16:18] we're an openstack shop, so looking to increase inter-tenant/vm throughput where possible [16:19] anything we can do to help... :) [16:55] ThiagoCMC, hey - sorry - stuck otp most of the afternoon [16:55] ThiagoCMC, actually working on that backport yesterday today [16:55] ppa:james-page/newton === iberezovskiy is now known as iberezovskiy|off [17:29] jamespage, cinder 2:7.0.2-0ubuntu1~cloud1 is ready to promote to liberty-updates [17:34] jamespage, that's good news, thank you! [17:35] apw: ping, re bug 1581076. Tested and verified that Xorg does not segfault [17:35] bug 1581076 in xorg-server (Ubuntu Xenial) "Xorg segfaults on start-up on Big Endian PPC hardware" [Undecided,Fix committed] https://launchpad.net/bugs/1581076 [17:35] with the build in proposed [17:35] MrBIOS, excellent, do put that in the bug [17:35] way ahead of you :) === InfoTest1 is now known as InfoTest [17:36] are there nightly builds of the LTS ISOs? If not, what’s the schedule for 16.04.2? [17:40] http://cdimage.ubuntu.com/ubuntu-server/xenial/daily/current/ [17:40] maybe that's what you are looking for? [17:41] nope, I know Xenial works fine [17:46] MrBIOS: xenial == 16.04.x [17:47] ah apologies, yes, that’s precisely what I’m looking for :) [17:47] 16.04.2 is just xenial + xenial-updates at a certain point. [17:47] * MrBIOS consumes more coffee [17:47] * Seveas attaches the coffee IV === andrewII is now known as andrew-ii [18:57] rbasak: bug 1620633 means that uvtool is broken when trying to access images created at any newer libvirt - not too high prio [18:57] bug 1620633 in uvtool (Ubuntu) "qcow image created in precise is not usable" [Low,Won't fix] https://launchpad.net/bugs/1620633 [19:28] If I add a node1 to my MAAS cloud, I should be able to ping node1.maas from the cli, right? === Guest56430 is now known as IdleOne [23:42] anyone here use fzf? can i get a ppa for it?