[00:09] is there any way to see what packages depend on a given package? [00:11] apt-cache rdepends package [00:13] ah, cool, thanks [01:25] someone reads what I white? [01:27] Can I ask something about UEC? [01:38] hi, I have installed and configuring the UEC following this guide https://help.ubuntu.com/community/UEC/CDInstall. [01:38] I can run an image and login it via ssh, but only from the pc that I've the cloud controller. [01:38] The problem is that I can't ping or login ssh the image from a pc of the LAN. [02:04] hi, I have installed and configuring the UEC following this guide https://help.ubuntu.com/community/UEC/CDInstall. I can run an image and login it via ssh, but only from the pc that I've the cloud controller.The problem is that I can't ping or login ssh the image from a pc of the LAN. [02:08] !weekend | FFF666 [02:08] FFF666: It's a weekend. Often on weekends the paid developers and a lot of the community may not be around to answer your question. Please be patient, wait longer than you normally would or try again during the working week. [02:14] uu, ok I'll ask tomorrow === erichammond1 is now known as erichammond [04:44] hey guys, have you ever experienced a system freeze on high file transfer speeds [04:45] on either gigabit ethernet or local hard to hard drive [06:35] amstan: I have, but only with unreliable hardware. ;) [08:27] Hi, I extended a volume group spanning two soft raid devices (/dev/md0 and /dev/md1). I want to reduce the volume group by removing /dev/md1 and tried using 'vgreduce raid /dev/md1' which returns 'Physical volume "/dev/md1" is still in use.' I'm stuck and can't seem to figure it out. I'm using Hardy Server i386. http://pastebin.com/hkQ4D68A [08:38] netritious: if you want to reduce the size, you have to unmount the partition [08:38] reducing the size of filesystem isn't as simple as extending [08:40] ivoks: Thank you. I knew I was missing something. Will give that a try. [08:50] ivoks: He's wanting to remove a pv from a vg, not reduce the size of a filesystem, I think. [08:50] netritious: You probably want to look at pvmove. It lets you move the used PE's from one PV to another. [08:51] netritious: So, in your case, you'd run "pvmove /dev/md1". [08:51] netritious: When that's done, you should be able to run your vgreduce command. [08:56] soren, ivoks: My system still is not cooperating with me. http://pastebin.com/iHjqYPhB Where to go from here? Maybe detach the drives physically from the system and then try again? [08:58] back :) [08:58] wb ivoks :) [08:58] soren: well, he should reduce fs before reducing lv/vg, right? :) [08:59] netritious: you can't unmount partitions you are using :) [08:59] pvmove is good suggestion [09:00] netritious: let's start from begining [09:00] netritious: you have volume group that uses /dev/md0 and /dev/md1, right? [09:00] ivoks: yes, and ty for your patience :) [09:01] phone [09:01] np..I'm here all night [09:02] ivoks: Well, maybe. Depends on whether they'd all fit on the single md device. [09:03] Would more detail help? [09:04] netritious: Yikes, not pvremove. pvmove. [09:05] soren: ok, I think I have some explaining to do. :) [09:07] I have five drives; 1x160GB IDE that is LG 'system'. Then there are 4x1TB drives. sda and sdb make up /dev/md0 which I added a few days ago. sdc and sdd make up /dev/md1 which I added about an hour ago. [09:08] ok [09:08] I initially created LG raid on /dev/md0 and extended it to /dev/md1 (just an hour or so ago) [09:09] I performed 'umount -a' which has unmounted LG 'raid' which is on /dev/md0 and /dev/md1 [09:09] and now you want to take /dev/md1 out of LF? [09:09] LG [09:10] out of LG 'raid' yes [09:10] then what soren said [09:10] pvmove [09:10] pvmove -v /dev/md1 [09:11] k..will try now, and again thanks for your patience...I am really new to mdadm and lvm. [09:12] vgreduce raid /dev/md1 [09:13] pvremove /dev/md1 [09:13] that should be it [09:13] What's an LG? [09:13] logical group, probably :) [09:14] And what's that really? [09:14] An LV or a VG? [09:14] or south korean manufacturer :) [09:14] VG [09:14] soren: you're right..I should have used VG [09:14] i'm not sure you'll be able to move everything [09:14] # [09:14] PV Name /dev/md1 [09:15] # [09:15] Allocatable yes (but full) [09:15] I'm very confused by all of this: [09:16] netritious: You have 1 160 GB drive and 4 1 TB drives, and then you speak of sda, sdb, sdc, and sdd. I'm not sure how to map the drive sizes to the device names. [09:16] ...especially since there are 5 drives, but only 4 device names mentioned. [09:16] soren: he has two volume groups [09:16] soren: 160GB is system, 1TB are /dev/md0 and /dev/md1 in raid VG [09:16] Also, if an "LG" really is a VG, it doesn't really make sane that a "umount -a" unmounts an LG that is called raid. [09:17] soren: http://pastebin.com/hkQ4D68A [09:18] * soren doesn't get it at all :) [09:18] 'pvdisplay' is relevant part :) [09:19] Ah, there's and sde as well. [09:19] ...which is the 160 GB drive. [09:19] soren: in system VG [09:19] soren: correct [09:19] so, ignore it :) [09:19] exactly :) [09:19] Oh, and "raid" is the name of the volume group? [09:19] Apologies for the confusion [09:19] soren: yes [09:19] You don't unmount volume groups. [09:20] So this makes no sense: 08:09 < netritious> I performed 'umount -a' which has unmounted LG 'raid' which is on /dev/md0 and /dev/md1 [09:20] of course, it's not mounted at all [09:20] ivoks: Precisely. [09:20] soren: it probably unmounted LVs on raid VG [09:21] ivoks: You don't mount LV's either :) [09:21] :p [09:21] You mount filesystems on LV's.... and the VG doesn't care either way. [09:21] Anyways... [09:21] Ah, more confusion from me again..I unmounted /mnt/storage which is mounted on /dev/mapper/raid-storage [09:21] netritious: I stand by my instructions, nevertheless. "pvmove /dev/md1" (add a -v for more info, if you please).. then do "vgreduce" like you did. [09:21] there we go :) [09:22] ...and then pvremove if you're feeling bored. [09:26] soren: http://pastebin.com/jGgFy9ce [09:29] soren: 'system' is a volume group and 'raid' is a volume group, and nothing is mounted on a logical volume in volume group raid. So I'm still missing something. [09:30] ivoks: You had just come and gone... http://pastebin.com/jGgFy9ce [09:31] Anyone here with clue where to report broken jigdo images on the releases mirrors? [09:31] Jeeves_: #ubuntu-release, probably. [09:32] netritious: Could you try "pvdisplay -m /dev/md1"? [09:33] i guess he has a filesystem of 2TB on /dev/mapper/raid-storage [09:33] netritious: when you added /dev/md1, did you extend logical volume and filesystem to 2TB? [09:33] soren: Thanks [09:33] soren: http://pastebin.com/1eHhQnNv [09:34] backups? [09:35] ivoks: Do you still have the first pastebin thing open? It's expired so I can't go and see. [09:35] soren: nope :/ [09:35] oh, wait... [09:36] soren: http://pastebin.com/eYpAUkSz [09:36] ivoks: Great, thanks. [09:36] Oh, that just had pvdisplay :( [09:37] netritious: Can't help you. You've spent all the space on md1 on an lv called "backups". [09:37] I am trying to look in my bash_history to find all the commands I've ran and in what order [09:38] if you don't need backup, delete that LV :) [09:38] That's not really important, is it? [09:38] no, backups has nothing in/on it...was what I intended to remove from the volume group 'raid' [09:38] netritious: you can't remove disk which is filled with data, without losing that data [09:39] There is nothing in 'backups' [09:39] netritious: then delete that LV [09:39] netritious: Oh, hang on. [09:39] netritious: You just want to remove that lv? Nothing else? [09:39] netritious: You don't actually want to remove md1 from the vg? [09:39] yes, just that LV, not 'storage' [09:40] wait, let me be more specific.. [09:40] Yes, please :) [09:40] but that's not what you said before :) [09:40] Hello everyone [09:41] ivoks: Apologies for the confusion..still a n00b with lvm... [09:41] Guys we are going to deploy around 100 Ubuntu Server in virtualization - which ubuntu server would you go for? 9.10 or 8.04.4 LTS? [09:41] it's easy partition -> PV -> VG -> LV -> FS [09:41] i need to understand how stable is 9.10 vs 8.04.4 [09:42] netritious: It's much easier if you explain what you want to achieve rather than what you think you need to do to achieve it. [09:42] Wilco_: Why? [09:42] Wilco_: real question is what do you need server for [09:42] servers [09:42] coffe [09:42] Wilco_: There's a new LTS coming out on Thursday. You may want to wait for that. [09:42] we are building some kind of service which mostly run tomcats [09:42] Please accept my condolences. [09:43] so im going to have like 100 VMs running tomcats [09:43] the question which ubuntu to use [09:43] 10.04 :) [09:43] i need something super stable [09:43] oh, 8.04 then [09:43] soren: k, I want to remove /dev/md1 from logical group 'raid' and create another volume group just for /dev/md1 [09:43] :) [09:43] but it has old tomcat [09:43] netritious: Why? [09:43] we are going to use the tomcat from tomcat web site [09:43] 10.04 should be stable [09:43] and the java vm from sun [09:44] then use 8.04 [09:44] ok [09:44] ivoks: Hm? why? [09:45] soren: I don't want to span the logical volume across /dev/md0 and /dev/md1 [09:45] soren: cause it's 'super stable' :) [09:45] i noticed that 9.10 when installed installs EXT 4 [09:45] is ext 4 stable enough ? [09:45] you can change that to any FS you want [09:45] these will be production servers so i really need something that wont dye on me [09:45] dye=die [09:45] yeah [09:46] i was just suprised when i saw it default to ext4 [09:46] netritious: Why? [09:46] the issues i have with 9.10 on servers is upstart [09:46] soren: Why not? [09:46] explain ? [09:46] i use VMware to run the ubuntus and they load up pretty fast [09:46] soren: is that a bad idea? [09:46] even vmware-tools works fine [09:46] netritious: Maybe. Hence the "why"? [09:47] soren: would it be wise to use the latest LTS [09:47] coming on thu ? [09:47] Wilco_: Usually, yes. [09:47] hmm [09:47] dilemmas dilemmas [09:47] Wilco_: You can install it now and start testing. [09:47] Wilco_: No need to wait. [09:47] soren: It just makes sense to me to only have to deal with one volume group per raid device. [09:48] soren - our machines are very simple. a tomcat, java jdk, apache, some php and thats it [09:48] netritious: Ok. Well, if that's wha tyou want, that's fine. I just don't see the point. [09:48] Wilco_: on 9.10, nfs is started before network, for example [09:48] ivoks: thats lame - dont you get a timeout here? [09:48] it's not the problem for services [09:48] 9.10 had a lot of problems like that. They should all have been addressed in 10.04. [09:49] problem is when your /home is NFS :) [09:49] hehe [09:49] and in 8.04 its ok ? [09:49] then boot fails [09:49] 8.04 and 10.04 are ok [09:49] great. [09:49] ok - another issue that im puzzled with is MYSQL Enterprise 5.1 - (not the one that comes off the ubuntu) - do yo uknow if its runs in 64bit on ubuntu 8.04? [09:49] i would suggest 10.04 then [09:50] just make sure your app works with php 5.3 [09:50] they have a weird matrix [09:50] of supported OS [09:50] but they are vague on ubuntu [09:50] don't know [09:50] i never understood people that but mysql... :/ [09:50] buy [09:50] soren: To be completely honest I probably don't need LVM at all...I'm really not doing anything with it unless I span the two RADI1 devices, correct? [09:50] netritious: The awesomeness of lvm is not just that you can dynamically resize LV's, but also that you can have a VG that spans multiple LV's. Unless you're expecting to move one of the raid sets to another machine, I just don't see the point in splitting them into two VG's. [09:51] have one VG or don't have VG at all [09:51] (in this case) [09:51] soren: I'm trying to account for that exactly...in the event this old hardware gives out on me. [09:51] ivoks: we need to for support purposes [09:51] netritious: Then you would take one of the raid sets and put in one box and the other raid set in another? [09:51] the drives are new, but the motherboard/ram/cpus are 8+ years old [09:52] Wilco_: still... :) [09:52] soren: can't until I get more hardware [09:52] netritious: I'm asking.. [09:53] netritious: If your server exploded, would you move one of the raid sets to one machine and the other raid set to another? [09:53] netritious: Or would you move them both to the same box? [09:53] soren: very possibly [09:53] the first [09:53] netritious: Ok. Then just split them. You're just losing a lot of flexibility that way, but ok. [09:53] soren: I just want the option [09:55] soren: splitting them makes it less flexible in the event I have to move to two different machines? Just trying to make sure I understand. [09:56] soren: or are you saying I have less flexibility with LVM in general by not spanning across the raid devices? [09:56] netritious: No, splitting them means that you now have to worry about where your data is kept. [09:56] netritious: One of the great things about lvm is that you just have a great big blob of storage and don't need to worry about where the data is. [09:57] You just say "hey, I need an LV that's 400 [09:57] GB for my databases" and it gives it to you. [09:57] It may span multiple PV's or not. You don't need to know. [09:57] You just know that you get a blob of storage you can use. [09:57] If you want to extend it later on, you just do. [09:58] ...and that may expand it into another pv. You don't need to know. [09:58] If you split them, you always need to consider whether you're good on space on this VG or if you'd rather put it on the other VG or whatnot. [09:59] soren: that makes sense [10:00] soren: that was kind of my point about using LVM in the first place...It's probably not needed with what I think I want out of it, but on the other hand if I need to move each volume group to two different servers and then span across more physical or RAID devices it would be optimal to split it now..I'm guessing at that of course. [10:01] but 'sotrage' and 'backup' have distinct purposes, but atm 'storage' is acting as both, which is why I wanted to add more drives in the first place [10:06] guys any of you use landscape to manage patch management on multiple ubuntu servers [10:06] we are considering getting it so we have a baseline of patches across all servers [10:07] Wilco_: we have it on test, considering it [10:07] it seems to have its limitations, but looks good nevertheless [10:07] i only need it to patch the servers :) [10:07] have you tried it ? [10:08] We've tried it [10:08] Wilco_: for just patching servers, landscape might be overkill :) [10:08] Couldn't really find an advantage over our existing Nagios, Munin, and 'apt-get upgrade' strategy. [10:09] Jeeves_: apt-get upgrade is an issue [10:09] Wilco_: How? [10:09] there's stuff like cron-apt [10:09] because in different times you may not have the same patch level across servers [10:09] i must have all my servers using the same updates [10:09] just update them all at the same time. [10:09] unattend-upgrade configured to update everything + a nagios check to make sure there isn't any package "blocked" by dpkg is usually working perfectly fine [10:09] Wilco_: Than you should always apt-get upgrade at the same time [10:10] Wilco_: landscape is convenient for that, yes, especially with a large number of servers [10:10] but on production critical servers you sometimes cant [10:10] i gonna have like 200 servers in a year [10:10] i need ORDER :) [10:10] * Jeeves_ gives Wilco_ cfengine and autoapt.pl [10:10] Jeeves_: what are these? [10:11] scriptz? [10:11] Wilco_: landscape costs $150 per server (though less for non-commercial and less with many servers). with 200 servers, I'd say it might be worth it [10:11] yeah they giviing me superb price [10:11] * stgraber is managing something between 600 and 1k Ubuntu servers (mostly Ubuntu containers with vserver/openvz + some vmware/kvm VMs + some physical hosts) [10:11] Wilco_: cfengine is a scriptinglanguage which allows you to configure servers out of svn [10:11] price is not the issue [10:11] I'd just use puppet, or cluster-ssh. [10:11] autoapt.pl install's software when you want it, and configures it unattended [10:11] Wilco_: then go for it - less hassle than using cfengine and the likes [10:11] ok [10:11] ic [10:11] thats great support here guys :) [10:12] stgraber: how do you patch your 1000 servers? [10:12] does anyone here have some script that hardens an ubuntu server? [10:12] or at least checks that its hardened? [10:12] Nope [10:12] 'harden'? [10:12] security [10:12] scans the os [10:13] turns of what ever requires turns off [10:13] Wilco_: bcfg2 (similar to puppet) which make sure package lists + files match what we have in our repo every hour + unattended-upgrade at fixed time every night for the upgrades + munin/nagios to make sure everything is consistent (including package version) [10:13] it should be pretty decent out of the box [10:13] Wilco_: It doesn't start anything that you didn't install :) [10:13] Wilco_: so long as you don't install unneeded stuff, it should be quite hardened by default [10:13] Wilco_: and you buy support, pih... :) [10:13] Wilco_: there's tons of ways to do that. [10:13] i just install openSSH server [10:13] ufw enable might help too [10:14] Wilco_: and fail2ban [10:14] ic [10:14] Wilco_: propper iptables setup, rkhunter and fail2ban and patching is a good start. [10:14] ok [10:14] and I see some already said that :P [10:14] Did rkhunter stop whining about ssh allready? [10:14] i wont be using iptables - gonna use vShield Zones [10:14] Jeeves_: no :) [10:15] i'll check rkhunter and fail2ban [10:15] ewook: Than it still sucks :) [10:15] Jeeves_: or, depends what the wining is about :P [10:15] stgraber: very intresting [10:15] Jeeves_: you meen the blacklisted sums? [10:15] mean [10:15] ewook: No. The root-login stuff. [10:15] stgraber: what happens when a new kernel releases - do you update it too ? dont you restart the server in this scenario ? [10:16] Jeeves_: I've never seen that one on a ubuntu.. or so I think. [10:16] stgraber: you said "our repo" - you have your own repository ? and if so why ? [10:16] Wilco_: yes you do. but that is planned maintenence to take down a server :) [10:16] Jeeves_: What root login stuff? [10:16] lil_cain: 'You are allowing a root login! That is very dangerous!' [10:16] Wilco_: If you're managing >3/4 machines that run anything non-standard, it's worth having your own repo. [10:17] Jeeves_: there's no root enabled in my configs [10:17] ok [10:17] i dont. [10:17] Jeeves_: so you might wanna check your setup :P [10:17] i just use ubuntu main [10:17] Wilco_: as I said, most of these are containers, containers use the host kernel and don't have their own. For physical and VMs, we upgrade the kernel and we have hooks in /etc/kernel.d/ to send e-mail in cases where there's a custom module that need manual building + maintenance window. For others, it really depends on the support contract we have with the customer regarding security, in some cases we reboot, in some cases we wait for anot [10:17] very simple [10:17] Jeeves_: Ah. I always thought that was a stupid default on the part of Ubuntu. [10:17] stgraber: nice [10:18] ewook: We allow root-logins by default. Because it makes stuff much easier for us. [10:18] Wilco_: having your own repo is the same thinging as having a WSUS for windows-machines. [10:18] Wilco_: we use the official archive + a ppa + an internal repository for non-distributable packages. "repo" in my previous sentence was refering to a configuration repository describing all "profiles" (package lists, configuration files lists, ...) that's in our case a bzr branch. [10:18] am i right that the only time you need to reboot the server is when the kernel/drivers update? [10:18] Root still doesn't have a password, only pubkeys [10:18] Jeeves_: well then, it's not rkhunter's fault :P [10:18] Wilco_: when libc6 does as well. [10:18] ok [10:18] ewook: It is, when I tell it not to whine about it and it still does :) [10:19] Jeeves_: oh. gotcha. :P [10:20] Jeeves_: I'd be happy if there's a slipstreamed and 'secured' way that it rehashes itself after a apt-upgrade [10:20] guys maybe its not relevant - a friend of mine asked me to help after he compailned that his server was hacked. i was able to confirm taht his server was hackde (root) - i could see the entire history the hacker did i.e in bash_history [10:20] however we noticed that teh server loads up with promiscious mode on [10:20] at boot time [10:20] any ideas how to turn that off :) ? [10:20] Disconnect the server, and use the console? [10:21] Wilco_: wipe & reinstall the whole thing [10:21] yeah thats what i told him [10:21] but it was interswting [10:21] I love the word promiscious, it's such a durty word :P [10:21] to understand [10:21] the moment redhat starts it says entering promiscuous mode on eth0 [10:22] i even tried forcing the network confgiguration to turn of promisucous - no worky [10:22] Wilco_: So how many passwords does the hacker have by now? [10:22] millions :) [10:22] And how many other boxes are rooted by now? [10:22] i guess all of t hem [10:22] i told him to wipe it out [10:22] he is really fucked [10:23] Wilco_: It's a good thing that he's not a customer of ours. His box would be offline by now [10:23] heehe [10:23] or is he!? [10:23] Is he in .nl?! :) [10:23] thats happens when people put a server on the internet without a firewall [10:23] Wilco_: Bullshit [10:23] his machine was open for port 22 [10:23] they scanned it [10:23] My server has been online without a firewall for years [10:24] with SSH open? [10:24] yes [10:24] they scanned [10:24] brute forced attack it [10:24] found root password [10:24] that's when inexperienced people put a server on the internet without firewall ;) [10:24] boom [10:24] server hacked [10:24] Wilco_: use fail2ban [10:24] JanC: Indeed :) [10:24] and secure passwords [10:24] Wilco_: Use sane passwords [10:24] yeah [10:24] hehehe [10:24] sure [10:24] 'koekje123' [10:24] 'toor' [10:24] you know hwo many admins use [10:24] 123456! [10:24] :) [10:24] 'thisisapassword' [10:24] Jeeves_: I'd call you brave, if you didn't patch at least the services running on it.. [10:24] os q2w3e4 [10:24] :) [10:25] 'secret' [10:25] I don't use passwords to login [10:25] ewook: Obviously, I patch my servers [10:25] ok [10:25] Wilco_: and don't allow root login over ssh [10:25] don't allow them [10:25] hehehe [10:25] RoyK: agree [10:25] And you shouldn't call me brave, but stupid :) [10:25] when it comes to security im a real bitch [10:25] Wilco_: but do use fail2ban [10:25] i'll check fail2ban [10:25] Jeeves_: I didn't wanna say out loud myself ;) [10:25] ok does anyone here knows of a wrapper for tomact? [10:26] a good one that loads it up as a service properly [10:26] ewook: :) [10:26] Wilco_: You mean 'apt-get install tomcat' ? [10:26] no [10:26] Ow, that does start tomcat as a service :) [10:26] i use tomcat off apache tomcat org [10:26] i use their latest and greatest [10:26] i need a wrapper for it to load up as as service [10:27] s/and greatest// [10:27] i.e service tomcat stop/start [10:27] I think it's included [10:27] nah [10:27] Otherwise, just use the wrapper out of the ubuntu package [10:27]