| === cpg is now known as cpg|away | ||
| === shantorn_ is now known as furian | ||
| === pug_1 is now known as DaveR | ||
| === cpg|away is now known as cpg | ||
| === n0ts is now known as n0ts_off | ||
| === cpg is now known as cpg|away | ||
| === n0ts_off is now known as n0ts | ||
| === n0ts is now known as n0ts_off | ||
| === fenris is now known as Guest70212 | ||
| === Guest70212 is now known as ejat | ||
| === fenris is now known as Guest50405 | ||
| === n0ts_off is now known as n0ts | ||
| === cpg|away is now known as cpg | ||
| === n0ts is now known as n0ts_off | ||
| === n0ts_off is now known as n0ts | ||
| k1ng | hi | 05:50 |
|---|---|---|
| k1ng | ii want to copy everything except upload dir and config.php with rsync. how do i do that? | 05:50 |
| blkperl | k1ng: `man rsync` search exclude | 05:53 |
| === kInOzAwA is now known as Guest37435 | ||
| lvmer | k1ng: ooooo lmk if you find a good tutorial for rsync or make progress. I need to find a consistent way to backup my server too. xD | 06:10 |
| k1ng | i actually did:) | 06:11 |
| k1ng | rsync -avn --exclude=upload --exclude=config.php --exclude=fixer.sh --exclude=apfixer.php --exclude=HN9200FU_6.10.0.12.zip --exclude=backup --exclude=news.txt --exclude=images/logo.png /home/backup/sims /root/test | 06:11 |
| lvmer | k1ng: xD | 06:15 |
| blkperl | lvmer: why not use backup software? like bacula or amanda | 06:24 |
| === kInOzAwA is now known as Guest94576 | ||
| === didel0815_ is now known as didel0815 | ||
| === smb` is now known as smb | ||
| === cpg is now known as cpg|away | ||
| === _ruben_ is now known as _ruben | ||
| === n0ts is now known as n0ts_off | ||
| jibel | smoser, the user 'ubuntu' is not added to sudoers in latest raring cloud images, is it known? | 11:50 |
| Daviey | jibel: is it added to admin? | 11:59 |
| === n0ts_off is now known as n0ts | ||
| jibel | Daviey, it is a member of adm but not admin | 13:05 |
| Daviey | jibel: nice | 13:09 |
| === gary_poster|away is now known as gary_poster | ||
| jibel | bug 1080685 , feel free to reassign to the right package | 13:13 |
| uvirtbot | Launchpad bug 1080685 in cloud-init "raring cloud images: User 'ubuntu' cannot sudo" [Undecided,New] https://launchpad.net/bugs/1080685 | 13:13 |
| Kartagis | hi | 13:30 |
| Kartagis | should I be worried about this? warning: process /usr/lib/postfix/trivial-rewrite pid 18501 exit status 1 | 13:30 |
| smoser | jibel, :-( no. i'll take a look. thanks for raising. | 13:33 |
| Akendo | Heloo Guys | 13:48 |
| Akendo | Hello Guys | 13:48 |
| Akendo | I'm using OpenStack Folsom from the Ubuntu Cloud Archive, i found and fixed a bug and want to comit this. How can i do this? | 13:49 |
| jamespage | Akendo, hi! | 13:50 |
| Akendo | Hello jamespage | 13:50 |
| jamespage | Akendo, is there a handy bug report for this problem that we can use? | 13:50 |
| Akendo | That is what I'm looking for | 13:51 |
| jamespage | Akendo, OK - so whats the problem you have found? | 13:51 |
| Akendo | There is a problem in glance, creating or uploading via RBD (RADOS Block Device) to Ceph. | 13:52 |
| Kartagis | hi. should I be worried about this? warning: process /usr/lib/postfix/trivial-rewrite pid 18501 exit status 1 | 13:52 |
| Akendo | When i create a Image it's raising a excepction | 13:52 |
| jamespage | Akendo, stack trace would be helpful | 13:52 |
| Akendo | 2012-11-19 14:32:45 ERROR glance.api.v1.images [3dfbce68-310a-4fab-8bac-ff1b133d7ea9 None None] Traceback (most recent call last): | 13:53 |
| Akendo | File "/usr/lib/python2.7/dist-packages/glance/api/v1/images.py", line 437, in _upload | 13:53 |
| Akendo | image_meta['size']) | 13:53 |
| Akendo | File "/usr/lib/python2.7/dist-packages/glance/store/rbd.py", line 234, in add | 13:53 |
| Akendo | image_size, order) | 13:53 |
| Akendo | File "/usr/lib/python2.7/dist-packages/glance/store/rbd.py", line 205, in _create_image | 13:53 |
| Akendo | librbd.create(ioctx, name, size, order, old_format=False) | 13:53 |
| Akendo | TypeError: create() got an unexpected keyword argument 'old_format' | 13:53 |
| jamespage | Akendo, pastebin is better | 13:53 |
| Akendo | Oh ;-) | 13:53 |
| jamespage | Akendo, which version of ceph are you using? | 13:54 |
| Akendo | The Question I have first: Does i can be, that this 'old_format' is requiert by a newer version of Ceph? | 13:54 |
| Akendo | Stable Version, 'old_format' | 13:54 |
| Akendo | Stable Version, ceph 0.41-1ubuntu2.1 | 13:54 |
| * jamespage looks | 13:54 | |
| jamespage | Akendo, yes - it does; 0.48 onwards will do the trick; | 13:56 |
| Akendo | hm.. | 13:56 |
| jamespage | thats currently in-flight to land in the folsom cloud archive - lemme see if I can nudge it along | 13:56 |
| Akendo | We | 13:57 |
| jamespage | Daviey, ^^ can we get ceph out of folsom-proposed and into folsom-updates please | 13:57 |
| david_ | I am using public/private keys to log into my server. I want to see in auth.log which key is being used for each login. Anybody know what parameter is needed in sshd_config ? | 13:57 |
| Akendo | I want to use the Stable release, but If I'm right there will relese a new version soon, right? | 13:57 |
| jamespage | Akendo, 0.48 is the current stable release | 13:57 |
| Akendo | Ah | 13:57 |
| Akendo | But | 13:58 |
| jamespage | bobtail is due based on 0.55 of ceph - a few weeks off yet | 13:58 |
| jamespage | that won't be put into the cloud-archive for Folsom; but will make it for Grizzly | 13:58 |
| Daviey | jamespage: yes | 13:58 |
| Akendo | So, I'm using an outdated Version? | 13:58 |
| jamespage | Akendo, basically yes | 13:58 |
| Akendo | hm... | 13:58 |
| jamespage | Akendo, this is one of the reasons we made the call to put ceph into the cloud archive as well | 13:59 |
| Akendo | We have this package out of the stable ceph repo | 13:59 |
| Akendo | So | 13:59 |
| Akendo | I guess that is the problem | 13:59 |
| jamespage | Akendo, which 'stable ceph repo' are you referring to? | 14:00 |
| Akendo | http://ceph.com/docs/master/install/debian/#add-stable-release-packages | 14:00 |
| Akendo | Like there do it in this documenation | 14:00 |
| jamespage | Akendo, the version you quoted is the version from the main Ubuntu archive in precise | 14:01 |
| Akendo | hm.. | 14:01 |
| Akendo | Strange... | 14:01 |
| Akendo | ok, just updating this | 14:02 |
| jamespage | Akendo, this will all shake out once the new version of ceph lands in the folsom cloud archive - that will push it up to 0.48.2 which is compatible with folsom | 14:02 |
| jamespage | Daviey, thanks | 14:03 |
| Akendo | So there is no bug report to commit ;-) | 14:03 |
| Akendo | Thank you | 14:03 |
| Daviey | jamespage: the tooling is on my other machine, so can it wait an hour or so? | 14:03 |
| jamespage | Akendo, no problem | 14:04 |
| Akendo | That helped a lot | 14:04 |
| jamespage | Akendo, FYI we are about to integrate ceph into the automated testing we do for OpenStack on Ubuntu | 14:04 |
| jamespage | which should help spot this sort of issue going forwards... | 14:05 |
| Akendo | That would be great ;-) | 14:05 |
| Akendo | Sounds exciting | 14:06 |
| Akendo | ^^ | 14:06 |
| uvirtbot | Akendo: Error: "^" is not a valid command. | 14:06 |
| Kartagis | hi. should I be worried about this? warning: process /usr/lib/postfix/trivial-rewrite pid 18501 exit status 1 | 14:06 |
| lamont | Kartagis: that depends entirely on why it's exiting | 14:09 |
| lamont | what else does syslog have to say about it? | 14:09 |
| === n0ts is now known as n0ts_off | ||
| Kartagis | lamont: http://ccd12e76108915a3.paste.se/ | 14:13 |
| Akendo | So, I updated now to a newer version of ceph. I restarted the server, but i can't see anymore the image | 14:26 |
| jamespage | Akendo, hmm | 14:32 |
| * jamespage rubs his chin | 14:32 | |
| Akendo | I can found the entry in the MySQL Db, but glance image-list is empty | 14:32 |
| Akendo | also does the rbd list -p images displaying the uploaded images | 14:33 |
| jamespage | Akendo, did you re-upload the image? I'm wondering if something in the DB is not quite right due to that error before | 14:35 |
| Akendo | Yes | 14:35 |
| Akendo | Also the re-uploaded image is not displayed | 14:35 |
| Akendo | Both images a list in the MySQL DB | 14:35 |
| Akendo | I'll going to try Cinder with Ceph now. Maybe the glance image-list have a issue with accessing rbd devices? | 14:40 |
| peta_ | Hello guys | 14:50 |
| smoser | smb, around ? | 14:55 |
| smb | yees... (wondering whether that is good) | 14:55 |
| peta_ | I want to roll my custom php5.4 build for my 10.04 production server and I am wondering what is the "best" method to create portable deb packages. I hear about checkinstall and it sounds too easy to be true. What would you suggest? | 14:56 |
| peta_ | Side note: My production server is a 10.04 amd64 and I have an exact copy of that system as virtual machine on my local computer -- that's where I do all the build stuff before something is moved to the production server. | 14:57 |
| lamont | Kartagis: command usage for trivial-rewrite in master.cf is bad in some way | 14:59 |
| smoser | smb, i'm hitting https://bugs.launchpad.net/ubuntu/+bug/1078926 | 15:00 |
| uvirtbot | Launchpad bug 1078926 in ubuntu "raring instance failed to find EC2 datasource" [High,Confirmed] | 15:00 |
| smoser | and its exhibiting really strange behavior. | 15:00 |
| smoser | i'm bothering you primarily because of the fact that networking was not reliable after reboot. | 15:00 |
| smoser | i'm guessing a flakey network driver could have cuased the issue on first boot also. | 15:01 |
| Kartagis | lamont: rewrite unix - - n - - trivial-rewrite | 15:01 |
| Kartagis | I'll be back tomorrow | 15:02 |
| smb | smoser, Hm, ok. So in micro its netfront. Though I need a bit of time to look trhough the messages in detail | 15:02 |
| lamont | Kartagis: I don't have time to dig into it - it could just be parameters in main.cf that are tripping it up as well | 15:02 |
| smoser | smb, you want to poke at the instance at all ? | 15:04 |
| smb | Oh, it is still up, then maybe yes | 15:04 |
| smoser | smb we've seen this "networking doesn't come up" *very* rarely previously, i'd guess on order of 1/1000 or more. | 15:04 |
| smoser | but it seems to come up more regularly on raring | 15:05 |
| smb | smoser, Ah so even with older releases? | 15:05 |
| smoser | well, really really rarely. | 15:05 |
| smoser | such that i believed it was a platform (ec2) failure. | 15:05 |
| smoser | (and i'm still not convinced it is not for the less common case). | 15:05 |
| smb | smoser, Yeah it might be nice to see the logs from the domain creation. Though I know that won't happen | 15:07 |
| smoser | smb, ok. stefan-bader-canonical can go into backdoor@ec2-50-16-73-126.compute-1.amazonaws.com | 15:08 |
| smb | smoser, Heh nice naming scheme... | 15:09 |
| smoser | fwiw, you might find use for lp:~smoser/backdoor-image at some other point in time. | 15:09 |
| smoser | smb, so, to show failure on that instanc,e right now, i'm trying to get /var/log/syslog off of it | 15:10 |
| smoser | and this fails (hangs) | 15:10 |
| smoser | ssh -C -v backdoor@ec2-50-16-73-126.compute-1.amazonaws.com 'sudo cat /var/log/syslog' | 15:10 |
| smb | smoser, Hm, even a dmesg hangs after a bit. Feels like network works a bit only... Something like this I only had with HVM domUs that had interrupt issues... | 15:14 |
| smoser | smb, one interesting thing though was that 'apt-get install pastebinit' worked. | 15:21 |
| smoser | which goes to s3 | 15:21 |
| smoser | which might point to a packet length issue | 15:22 |
| smoser | wow. there is just all sorts of stuff seemingly busted there. | 15:25 |
| smb | smoser, It seems a dmesg into less also works better... hm not yet at netfront but "blkfront device/vbd/2049 num-ring-pages 4 nr_ents 128" now if I could remember things better I might know whether that was more than 1 before... | 15:25 |
| smoser | smb, hm.. | 15:26 |
| donspaulding | Hey there, I've got a running installation of Karmic server out in a datacenter. It doesn't have anything valuable on it, so I'm wondering if I can repurpose it as an openstack controller. I want to install the Quantal on it, but I don't have console access to the machine. Is there a way I can install ubuntu server from the running install of Karmic? | 15:34 |
| roaksoax | jamespage: oh btw... I don't know if you noticed but there miught be an issue with rabbitmq | 15:36 |
| roaksoax | jamespage: sometimes the instances fail due to rabbitmq failing to start | 15:36 |
| jamespage | roaksoax, might their? | 15:36 |
| jamespage | is that on instances? | 15:36 |
| roaksoax | jamespage: yeah | 15:36 |
| jamespage | we tear it up and down pretty regular in the lab. | 15:36 |
| roaksoax | jamespage: yeah, this is only in the instances, sometimes happens, sometimes doesn't, pretty weird | 15:37 |
| smb | smoser, Did you just stop the instance? | 15:39 |
| smoser | probably | 15:40 |
| smoser | :-( | 15:40 |
| smoser | smb, i hvae another just a minute | 15:41 |
| smoser | sorry | 15:41 |
| smb | smoser, Oh, no worries. Not sure I can really get much more right now. Just was a bit surprised. | 15:41 |
| smoser | backdoor@ec2-174-129-111-177.compute-1.amazonaws.com | 15:41 |
| smoser | smb, if you want, you can use that. | 15:42 |
| smoser | it seems to me that multiple things at play | 15:42 |
| smoser | * race condition for ifup | 15:42 |
| smoser | * plymouth dies (/var/log/boot.log doesn't get to /dev/console) | 15:42 |
| smoser | * flakey network driver/settings/something | 15:42 |
| smb | smoser, Ok, and I probably should try whether things are the same if I run on my version of Xen 3.4.3 | 15:43 |
| smoser | smb, do you think there is useful info in the fact that local(ish) networking seems to work | 15:45 |
| smoser | (ie, to s3) | 15:45 |
| smb | smoser, It is a bit confusing as anything goes through the same virtual nic | 15:46 |
| smoser | well, but the packet size would differ | 15:46 |
| smoser | no? | 15:47 |
| smb | smoser, Oh, hm... could it be more of a console device issue... | 15:47 |
| smoser | ? | 15:47 |
| smoser | no. | 15:47 |
| smoser | scp hangs. | 15:47 |
| smoser | whic hprobably doesn't allocate a console device | 15:47 |
| smoser | i woudln't think (if you're speaking of a pty) | 15:47 |
| smoser | but i could be wrong on that | 15:47 |
| smb | smoser, Was rather thinking of the console one attaches to at login. But if a remote scp fails as well... Package size may differ but I would rather think internal networking would use the biggest usable size | 15:49 |
| smb | smoser, But yeah, we use pts here, so not what I was thinking | 15:50 |
| smb | smoser, The only other difference would be that apt-get install would have more receiving data. Which direction was you scp? | 15:52 |
| smoser | smb, i was trying to scp /var/log/messages from the instance | 15:53 |
| smoser | also, wget http://ubuntu-data.s3.amazonaws.com/ebs/ubuntu-images-milestone/ubuntu-lucid-10.04-beta1-amd64-server-20100317.img.tar.gz -O /dev/null | 15:53 |
| smoser | that works fine | 15:53 |
| smoser | reliable ~200M transfer at 10M/s | 15:53 |
| smb | smoser, So that scp was also sending, as are the dmesg or cat that seem to lock up | 15:54 |
| smoser | hm.. | 15:54 |
| smoser | smb, i can put up anoter instance in that zone if you'd like | 15:55 |
| smoser | then you can try moving data back and forth between it if you 'd like | 15:55 |
| smb | smoser, I am just pulling something bigger from people which seems to work too | 15:55 |
| smoser | os it woudl seem maybe we're busted on traffic going into the instance. | 15:56 |
| === matsubara is now known as matsubara-lunch | ||
| smb | smoser, To me it seems higher traffic out of the instance causes issues. | 15:57 |
| smoser | "higher" == > 10k | 15:57 |
| smb | smoser, Let me try to recreate that locally | 15:57 |
| ninjix | does anyone have thoughts or tips for managing the maintenance portion of long running Ubuntu cloud instances? I've been searching around but haven't really found any good writeups on the subject. | 15:57 |
| smoser | i had to use 'split --bytes=4095' to use 'ssh user@host cat file' reliably. | 15:58 |
| smoser | (4095 not reliably determined. i jumped from 10000 to that just on a whim) | 15:58 |
| roaksoax | jamespage: oh btw... i think there's a problem with l3-agent | 15:59 |
| jamespage | roaksoax, oh yes? | 15:59 |
| roaksoax | jamespage: so I have setup the HA cluster. I "kill" the quantum/0, it fails over to quantum/1, but l3 fails to start due to being unable to connect to whatever it needs to connect | 16:00 |
| roaksoax | jamespage: i haven't yet look at what might it be... just wanted to let you know :) | 16:00 |
| smoser | smb, on this particular instance, and right now, i'm seeing this works: | 16:01 |
| smoser | ssh backdoor@ec2-174-129-111-177.compute-1.amazonaws.com dd if=/dev/urandom bs=1024 count=10 >/dev/null | 16:01 |
| smoser | this hangs: | 16:01 |
| smoser | ssh backdoor@ec2-174-129-111-177.compute-1.amazonaws.com dd if=/dev/urandom bs=1024 count=11 >/dev/null | 16:01 |
| ninjix | most of what I find written is geared toward ephemera instances. I want to move more of our core systems to Ubuntu Openstack arch | 16:02 |
| smoser | ninjix, what does "maintenance portion" mean ? | 16:02 |
| smoser | applying updates? | 16:02 |
| smb | smoser, wtf... | 16:02 |
| roaksoax | jamespage: or the connection was refused | 16:03 |
| ninjix | smoser: life cycle of an instance | 16:04 |
| roaksoax | jamespage: ./deployer.py -c openstack.cfg openstack-quantal-quantum | 16:04 |
| roaksoax | err | 16:04 |
| roaksoax | jamespage: http://paste.ubuntu.com/1370377/ | 16:04 |
| ninjix | I'm wondering about how people are taking care of their long running instances | 16:05 |
| ninjix | the question came to mind this morning while I was having to perform a number of full reboots to pickup kernel updates | 16:06 |
| smoser | smb, random information: | 16:09 |
| smoser | ssh backdoor@ec2-174-129-111-177.compute-1.amazonaws.com dd if=/dev/urandom bs=$((1024*5+(256+111))) count=1 >/dev/null | 16:09 |
| smoser | that is the first block size to hang fo rme. | 16:09 |
| smb | smoser, But you do nothing that causes any different traffic on any block or net device... Execept maybe a slightly different timing... | 16:10 |
| smb | smoser, Oh wait urandom... lack of entropy maybe | 16:21 |
| smb | smoser, At least one I did is waiting on the read to finish | 16:22 |
| smoser | urandom doesn't use entrop | 16:23 |
| smb | smoser, I think it does but should stop if low on entropy... But I jused messed up the repeat. Doing 1024 5MB records is a bit insane. Seems while being logged in I can do 5*5MB without locking | 16:31 |
| smoser | smb, to where? | 16:32 |
| smb | backdoor@ip-10-212-103-115 | 16:32 |
| smb | but also from urandom to null | 16:33 |
| smoser | smb, i was just doing network transfer | 16:33 |
| smoser | that was the thing | 16:33 |
| smoser | used /dev/urandom (badly) to avoid compression. | 16:33 |
| smoser | smb, but interestingly, this fails same way: | 16:34 |
| smoser | ssh -o Compression=no $uhost dd if=/dev/zero bs=$((1024*5+(256+110))) count=1 >/dev/null | 16:34 |
| smoser | (well, the 111 fails, 110 passes) | 16:35 |
| smb | smoser, Oh, doh! | 16:35 |
| smb | smoser, Ok, I see, you discard the data locally. | 16:35 |
| smoser | right. | 16:35 |
| tonyyarusso | Anyone know how ufw/iptables would impact ARP? | 16:43 |
| tonyyarusso | (An entry was not showing up on the switch with ufw enabled; disabled ufw and the entry I was looking for popped right up.) | 16:44 |
| SpamapS | tonyyarusso: AFAIK, ufw doesn't mess with any of the ARP settings | 16:45 |
| vezq | don't have very good experiences with ufw | 16:49 |
| smb | smoser, Ok, for now I am off the instance and you can rip it down if you want. I want to try that locally | 16:50 |
| nopz | Hi there, anyone using lsyncd ? | 16:52 |
| smoser | smb, k. | 16:52 |
| jdstrand | ufw does not do anything with arp | 16:57 |
| tonyyarusso | I wouldn't think it would, but somehow it seems to be affecting something. Very confused. | 16:59 |
| === matsubara-lunch is now known as matsubara | ||
| tonyyarusso | How would I specify that the IP addresses associated with each interface should only be accessible through that interface, not the other? For instance, if eth0 is .2 and eth1 is .3, no traffic to .3 should could through eth0. | 17:03 |
| rbasak | tonyyarusso: is http://www.tldp.org/HOWTO/Adv-Routing-HOWTO/lartc.kernel.rpf.html what you need? | 17:05 |
| rbasak | tonyyarusso: also see: http://lkml.indiana.edu/hypermail/linux/kernel/0002.2/0880.html | 17:06 |
| === mcclurmc is now known as mcclurmc_away | ||
| tonyyarusso | rbasak: I don't think so? | 17:11 |
| tonyyarusso | rbasak: To expand a bit, I have two interfaces that are on the subnet, connected to the same switch, on the same vlan. The switch is .1, eth0 is .2, and eth1 is .3. eth0 is connected to switch port 2:20 and eth1 is connected to port 2:15. My goal is to totally separate traffic between the physical interfaces by IP address, so that if eth1 gets saturated by traffic to .3 (incoming sflow packets only) I can still do management ... | 17:13 |
| tonyyarusso | ... stuff to eth0 on .2 (all two-way traffic). | 17:13 |
| tonyyarusso | Man, this ufw thing is reproducible. Disable ufw, and the ARP entry for eth1 pops up on the switch, but it's not there when ufw is enabled. Weird. | 17:14 |
| rbasak | Oh, I see. You can do that with multiple routing tables and advanced routing (see the howto), but I'd recommend against it. Everyone else with that need would use a separate management lan or vlan, and that's probably the path of least resistance | 17:15 |
| tonyyarusso | perhaps | 17:16 |
| jdstrand | tonyyarusso: are you using bridging? | 17:17 |
| tonyyarusso | jdstrand: no | 17:17 |
| BrixSat | Hello, is there any daemon with webinterface for network monitor? | 17:19 |
| jdstrand | tonyyarusso: I suggest doing something like 'sudo ufw logging high' and then look in /var/log/ufw.log to see what is going on (in fact, with 'logging low' (the default), you might already have stuff there. it seems pretty clear that you will need to add iptables rules to /etc/ufw/before*.rules to handle your setup (see man ufw-framework for details) | 17:19 |
| Kartagis | how do I know whether postfix is in chroot? | 17:19 |
| lamont | Kartagis: if you didn't chagne the config, it's in a chroot by default on debian and ubuntu | 17:20 |
| Cuacrzz | hello, i need some help with a l2tp vpn config please! | 17:26 |
| jamespage | roaksoax, that would indicate the l3-agent is not configured at-all | 17:28 |
| jamespage | roaksoax, ignore me | 17:28 |
| * jamespage scrolls to bottom of log | 17:29 | |
| jamespage | roaksoax, is 10.55.60.163 the IP of the node you took out? | 17:29 |
| jamespage | l3-agent will be getting URL's for quantum from keystone; and if one has dropped we need the haproxy frontend to keystone as well. | 17:29 |
| raub | Does anyone know if the Intel x520-DA2 (chipset 82599ES) works well under ubuntu 12.04LTS? https://wiki.ubuntu.com/HardwareSupportComponentsWiredNetworkCardsIntel might be a bit dated | 17:35 |
| jamespage | roaksoax, note that the scalability of the l3-agent is a documented limitation - it has to query quantum to figure out what it should be doing | 17:35 |
| roaksoax | jamespage: right, this is not scaling though, it is a failover | 17:37 |
| roaksoax | jamespage: so l3-agent is simply started in quantum/1 node, if quantum/0 fails | 17:37 |
| jamespage | roaksoax, I know | 17:37 |
| roaksoax | jamespage: let me deploy everything again | 17:38 |
| jamespage | roaksoax, ack | 17:39 |
| roaksoax | jamespage: btw.. you didn't make any more changes to the quantum charm right? | 17:39 |
| * jamespage thinks | 17:39 | |
| jamespage | I'd have to check the branch but I don't think so no | 17:39 |
| jamespage | last commit on friday | 17:40 |
| tonyyarusso | jdstrand: well, I'm not seeing anything obvious yet at least. | 17:41 |
| Akendo | Hello jamespage , sorry for disturb again. My cinder have troune using this uuid password for ceph. It looks like it that it's no working | 17:44 |
| Akendo | Any idea? | 17:44 |
| Akendo | troube* | 17:44 |
| jamespage | Akendo, yeah - that is fiddly | 17:44 |
| Akendo | Make me a little bit crazy..... why a uuid? | 17:44 |
| jamespage | Akendo, take me through how you are setting up and managing the uuid | 17:44 |
| jamespage | specifically on cinder and on the nova-compute nodes.... | 17:44 |
| Akendo | http://ceph.com/docs/master/rbd/rbd-openstack/#setup-ceph-client-authentication | 17:45 |
| jamespage | (I wrote all of this stuff into the Juju charms for OpenStack so I know how awkward it is) | 17:45 |
| Akendo | Like the ceph guys it do | 17:45 |
| jamespage | right - I banged my head against that for a while as well. | 17:45 |
| roaksoax | jamespage: another thing. The ha cluster will need at least 2 interfaces ... so in the quantum charm I'll have to specify: 1. network address of interfaces, 2. multicast address to use. 3. and multicast port to use | 17:46 |
| roaksoax | jamespage: how can we make the config to dinamically add interfaces | 17:46 |
| jamespage | Akendo, you can do it one of two ways | 17:46 |
| roaksoax | (in case we want to add a third, fourth) | 17:46 |
| Akendo | ok | 17:46 |
| jamespage | 1) Create the uuid on the cinder host first; and then create it using exactly the same uuid on all of the nova-compute nodes | 17:46 |
| jamespage | or | 17:46 |
| jamespage | 2) Create it individually on each node and specify the rbd_secret_uuid for each host individually (they can all be different) | 17:47 |
| jamespage | roaksoax, wellll.... | 17:48 |
| jamespage | that is awkward | 17:48 |
| jamespage | Akendo, one second - I'll dig out a bit of code for you | 17:49 |
| roaksoax | jamespage: ackward, but required, unless we *always* require a second nic | 17:49 |
| Akendo | hm.. | 17:49 |
| Akendo | Thank you, I'll try it again | 17:50 |
| jamespage | Akendo, this is the config in nova-compute for the second option - http://bazaar.launchpad.net/~charmers/charms/precise/nova-compute/trunk/view/head:/hooks/nova-compute-relations#L131 | 17:50 |
| jamespage | set_or_update just sets options in nova.conf | 17:50 |
| jamespage | roaksoax, its tricky because to an extent you have to make assumptions about the physical machine you are using | 17:50 |
| jamespage | having a config option that specifies the physical nic to use would make sense | 17:51 |
| jamespage | but it has to be consistent across all service units | 17:51 |
| roaksoax | jamespage: exactly, but that's why this cluster uses the network address, and with it , it selects the interface | 17:51 |
| roaksoax | jamespage: so to me, it makes more sense to send the network address of the iface to use | 17:51 |
| jamespage | OK - so lets ignore that issue for the time being and assume that is the case | 17:51 |
| roaksoax | rather than a nic | 17:52 |
| Akendo | Ah | 17:52 |
| Akendo | I now understanding how this work, awesome link! | 17:52 |
| jamespage | roaksoax, but you are assuming the nic is already configured then? | 17:52 |
| Akendo | Now I * | 17:52 |
| roaksoax | jamespage: i';m assuming that I know the network address to which the cluster is going to be connected to | 17:52 |
| jamespage | Akendo, https://javacruft.wordpress.com/2012/10/17/wrestling-the-cephalopod/ has a bit more on deploying ceph with openstack using juju charms | 17:52 |
| Akendo | hm.. | 17:53 |
| Akendo | Ok | 17:53 |
| jamespage | Akendo, even if you don't want to use juju its worth reading the code to plunder the knowledge :-) | 17:53 |
| jamespage | roaksoax, is this a multicast address? | 17:54 |
| roaksoax | jamespage: so the thing is the communication between cluster nodes is based on "rings" | 17:54 |
| roaksoax | jamespage: each rings needs, multicast addresss, multicast port, and network address | 17:55 |
| jamespage | roaksoax, network address per service unit right? | 17:55 |
| roaksoax | jamespage: right, the network address to which the service unit is connected to | 17:55 |
| roaksoax | jamespage: every node in the cluster nees to be configured with the same network address, mcast address/port | 17:56 |
| jamespage | roaksoax, right | 17:56 |
| roaksoax | jamespage: so I can't simply assume that eth0, for example, is the interface used for all the physical systems to connect to the same network | 17:57 |
| roaksoax | which means using network address is a better approach | 17:57 |
| jamespage | roaksoax, I still don't see how that maps to individual service units | 17:58 |
| jamespage | eth0 is universally appliable | 17:58 |
| zul | smb: still around? | 17:58 |
| jamespage | 192.168.21.99 applies to a single unit only | 17:58 |
| smb | zul, Somewhat, but a good deal of my brain is away... | 17:59 |
| jamespage | roaksoax, unless I missed something and you mean 192.168.21.0/24 (i.e. a network address)? | 17:59 |
| roaksoax | jamespage: right, but physical machine 01, is connected to network 10.10.10.0/24 on eth0, machine02, is connected to 10.10.11.0/24 | 17:59 |
| roaksoax | on eth0 | 17:59 |
| roaksoax | if we specify interface to use | 17:59 |
| jamespage | roaksoax, right - sorry - I see now - I was being dumb. | 17:59 |
| roaksoax | the nodes in the cluster will never see each other | 17:59 |
| zul | smb: did you have a script that takes the cloud-images and make it usuable for xen? | 17:59 |
| jamespage | roaksoax, I'd also never underestimate the value of common cabling in the DC classed by machine type | 18:00 |
| smb | zul, Sort of. It is usable for Xen, what it does is adding no-cloud data | 18:00 |
| jamespage | its invaluable :-) | 18:00 |
| zul | zul: care to share? | 18:00 |
| roaksoax | jamespage: oh definitely not, they are going to use a standard cabling/network per interface and stuff | 18:00 |
| smb | zul, Talking to yourself again? :) | 18:01 |
| roaksoax | i just don't want to assume | 18:01 |
| zul | smb: perhaps :) | 18:01 |
| roaksoax | jamespage: though it could be a requirement... have the same interface in the same network | 18:01 |
| jamespage | roaksoax, I have to dash now - if you want to leave your test rig running and add my LP key (james-page) and PM me the details I'll take a look first thing | 18:01 |
| zul | smb: some people's brains are not here as well | 18:01 |
| smb | zul, I can post it to you. I hope it is generic in usage, you should look at it carefully | 18:01 |
| jamespage | roaksoax, I don't think thats unreasonable | 18:01 |
| smb | zul, :) | 18:01 |
| zul | smb: please | 18:01 |
| jamespage | I make that assumption in the quantum charm already | 18:01 |
| jamespage | with the ext-port configuration | 18:01 |
| roaksoax | jamespage: ok, so i guess i can do the same then | 18:02 |
| zul | right...so libvirt 1.0 works with xen 4.1 | 18:06 |
| smb | zul, I am trying to get a xen package reviewed for raring that has 4.2 and re-adds qemu-dm since | 18:08 |
| zul | smb: linky? | 18:08 |
| smb | that was suggested by the xen folks (because upstream qemu does no migration) | 18:09 |
| zul | right | 18:09 |
| smb | zul, I asked someone else already for a review, so I don't want to create chaos by spreading it into all corners | 18:10 |
| zul | ack | 18:10 |
| smb | Hope it gets done soonish though | 18:11 |
| smb | zul, script sent | 18:11 |
| zul | thanks | 18:11 |
| smb | zul, Let me know how it works for you. I have to admit it might be a bit "evil"... | 18:13 |
| zul | smoser: hey do you have a problem with adding an upstart job for hvc0 for pv-domu guests in the cloud-images? | 18:17 |
| smoser | what would said upstart job do ? | 18:18 |
| zul | smoser: just creates a getty | 18:22 |
| zul | http://pastebin.ubuntu.com/1370663/ | 18:22 |
| smoser | id rather you think of it generically. | 18:23 |
| smoser | maybe 'virt-getty.conf' that generally tries to run a getty on any "virtual console". even potentially considering 'ttySX' to be such a console. | 18:23 |
| zul | sounds reasonable | 18:25 |
| smb | smoser, I actually have one for hvc0.conf which only starts it when /dev/hvc0 is present | 18:27 |
| zul | smb: crappers http://pastebin.ubuntu.com/1370663/ | 18:27 |
| smb | zul, smoser, http://paste.ubuntu.com/1370690/ | 18:28 |
| zul | smb: no domU drivers loaded : | 18:28 |
| zul | smb: that works for me | 18:29 |
| smb | zul, I intend to add this to the xen package in one of the next steps to get the same on dom0 | 18:30 |
| smoser | smb, you should be able to reference DEVNAME in the getty call i think | 18:30 |
| smoser | and then extend the start on to other things also | 18:30 |
| ninjix | hey guys, which openstack network mode is the server team using? | 18:31 |
| smb | smoser, Well this was started with more of dom0 in mind, so I did want to exclude the container case. Did not seem overly making sense there | 18:31 |
| ninjix | are you running Quantum for much of your testing? | 18:32 |
| smb | smoser, Oh, if you meant extend to start on different tty's... It seemed that rather each desired one should have its own conf file | 18:33 |
| === Akendo is now known as Akendo_ | ||
| smoser | smb, well i dont know. why would you want a bunch of things that differ only by 3 chars. | 18:35 |
| smoser | i was suggesting a generic one that was configured one place that said "do you want a getty on a virtual console device" | 18:35 |
| smb | smoser, You could ask why there is tty[12345].conf... ;) | 18:35 |
| smoser | then if any were there, it turned them on. | 18:36 |
| smoser | (yes, i could ask that) | 18:36 |
| smoser | especially if those jobs can be started on demand as yours shows there. | 18:36 |
| smb | smoser, I guess one thing is it makes it simpler to start/stop ttys individually | 18:37 |
| === yofel_ is now known as yofel | ||
| smb | And not sure how tight the need for one process triggered by one conf file in udev is... that is something I need to investigate too for trying to make xen start scripts upstart ones | 18:39 |
| === matsubara is now known as matsubara-brb | ||
| RoyK | what's the big deal with xen when kvm's around? | 18:45 |
| smoser | smb, its not significantly simpler to start/stop ttys individually. | 18:49 |
| smoser | you hav eto make it an instance job. and | 18:49 |
| smoser | start virt-console DEVNAME=ttyS0 | 18:50 |
| smoser | or something like that | 18:50 |
| smb | smoser, Well, yeah. I would not claim to know why it was done that way. I just usually assume there has been some reason for it (maybe just histerically because there were individual lines in inittab). Or maybe it allows for simpler additions without risking to loose those when upgrading. | 18:55 |
| === matsubara-brb is now known as matsubara | ||
| === kInOzAwA is now known as Guest53130 | ||
| tonyyarusso | How do I prevent a route from being created on boot? Specifically, I do NOT want the route that reads "206.131.129.4 * 255.255.255.252 U 0 0 0 eth1". Here's my /etc/network/interfaces currently: http://pastebin.com/pBWq7TC6 | 19:45 |
| keithzg | I swear I've done it plenty of times, but I've just upgraded a server to 12.04 and it keeps prompting me to upgrade to 12.10 upon terminal login; how do I stick it to LTS only? | 19:50 |
| ScottK | keithzg: Look in /etc/update-manager/release-upgrades | 19:52 |
| keithzg | ScottK I already did that, it has the line "Prompt=lts" | 19:53 |
| keithzg | and yet... | 19:53 |
| ScottK | Dunno. | 19:53 |
| ScottK | Did you restart after you changed it? | 19:54 |
| ScottK | I'm not sure where that gets cached. | 19:54 |
| keithzg | I didn't even ever change that | 19:55 |
| keithzg | It has, as far as I know, always said that. | 19:55 |
| keithzg | last modified date is Aug 8, looks like | 19:55 |
| RoyK | keithzg: wierd - try an "apt-get dist-upgrade" - make sure it's updated - simetimes there's a bug... | 19:55 |
| keithzg | The following packages will be upgraded: unity-scope-musicstores | 19:56 |
| keithzg | heh, methink that won't do the trick | 19:56 |
| sarnold | indeed, but at least it is a simple step :) | 19:57 |
| keithzg | sarnold truth! | 19:57 |
| tonyyarusso | keithzg: You have to delete a file in /var. Let me see if I can find it. | 19:58 |
| tonyyarusso | keithzg: /var/lib/update-notifier/release-upgrade-available | 19:59 |
| altermann | hello, is the www-data user equivalent to apache in other distros? | 19:59 |
| altermann | i was not able to find anything conclusive on google | 19:59 |
| tonyyarusso | keithzg: just remove it, and it will get recreated with the correct parameters on next run of the script | 19:59 |
| tonyyarusso | altermann: yes | 19:59 |
| altermann | thank you | 20:00 |
| keithzg | tonyyarusso: thanks! | 20:03 |
| sarnold | tonyyarusso: excellent :) thanks | 20:03 |
| === robbiew is now known as robbiew-afk | ||
| === mcclurmc_away is now known as mcclurmc | ||
| === mcclurmc_away is now known as mcclurmc | ||
| rnbrady | Hi folks | 21:32 |
| rnbrady | What is the canonical way to configure RAID post install on an Ubuntu server? | 21:32 |
| === cpg|away is now known as cpg | ||
| === jiboumans_ is now known as jiboumans | ||
| === cpg is now known as cpg|away | ||
| keithzg | arghh I've added a second ethernet card to a server for the sake of an iscsi mount, and now it's insisting on using that interface for everything. How do I get it to only use an interface for a specific IP range, and use the other interface for all other traffic? | 23:56 |
| sarnold | keithzg: I assume the iscsi endpoint is in the same netblock as all the other hosts? | 23:59 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!