=== waspinator_ is now known as waspinator === CiPi is now known as cipi === andol_ is now known as andol [05:28] How bad of an idea is it to install php 5.3.10 on ubuntu server 15.10? With mcrypt, curl, and mysql support. [05:42] cynixx3, precise 12.04 lts comes with 5.3.10 [05:43] if it is a bad idea, don't know, you sure have a reason [05:46] could I use the 12.04 repository of those packages to install on 15.10? === Piper-Off is now known as Monthrect [07:50] hi, i sucessfully installed landscape and openstack. Now i want to extend openstack with other units like ceilometer, etc... is there any documentation howto do it? [13:05] morning === Lcawte|Away is now known as Lcawte === balloons is now known as Guest56507 === Guest56507 is now known as balloons_ === balloons_ is now known as balloons === Lcawte is now known as Lcawte|Away [14:44] jamespage: is that pylxd with py3 or py27? [14:46] zul, py27 - I was trying to backport the pylxd package for the UCA [14:46] jamespage: ack [14:46] so its a package build failure rather than just in git [14:47] zul, I think that the constants in the ssl module have changed - might need to tweak based on version in use [14:47] the pull request for websockets had something in it to that effect I think [14:48] jamespage: yea [14:48] jamespage: looking at it now === smb` is now known as smb === Monthrect is now known as Piper-Off [15:22] jamespage: this works http://paste.ubuntu.com/13247939/ [15:27] zul, looks ok - you might what to push the hasattr in the runtime code out of the method call, so its called once rather than one every socket creation? === quantic_ is now known as quantic [15:30] jamespage: yeah [15:45] utlemming: or rcj: can one of you guys submit fixes to this answer? http://askubuntu.com/a/125252/235 [15:46] there are questions people are asking about how the mirrors are set up that I can't answer [15:46] also I totally forgot we moved away from s3 for those things === cipi is now known as CiPi === Lcawte|Away is now known as Lcawte === CiPi is now known as cipi [16:54] hello [16:55] i have problems with squid3 [16:56] http://pastebin.com/kE3kbVGT [16:56] this is my config [16:57] http://pastebin.com/p1zAkhDQ [16:57] and this is the php script [16:57] but squid seems to be not filter website's [16:58] i have use iceweasel with proxy enabled [16:58] port is ok [16:59] how can i make it work === cipi is now known as CiPi [17:54] paule32: hmm, be careful with that php script, you've written an sql injection bug into it on line 15; perhaps squid mangles inputs enough that users can't actually exploit it, but I wouldn't want to rely on that. [17:54] paule32: what happens if the query returns nothing? will the $row[2] reference blow up? [18:03] sarnold: i run the in the console, it waits for input and iff i type in "foo" press return/enter key it print outs "ERR" [18:04] paule32: ah, good; that just leaves the sql injection :) [18:05] but squid don't realize it [18:09] any ideas? [18:11] paule32: I don't see %DST described here, http://www.squid-cache.org/Doc/config/logformat/ -- are you sure that part is correct? do you get any errors or warnings in any of the log files at squid startup or when querying the squid proxy? [18:13] http://pastebin.com/t88d0pf3 [18:14] i need to be able to connect to sftp to save my backups (only used rsync before), how do i generate a cert and connect? (yes i tried connecting normally with just user@host) [18:15] Deliant: sftp is just a subsystem of ssh. Look up key-based authentication for ssh. === Lcawte is now known as Lcawte|Away [18:16] so i should just be able to ssh-copyid it as if it was rsync? [18:17] naa.. uha [18:18] normally ftp up the .pub file as .ssh/authorized keys then? [18:22] Deliant: I like to -append- the public portion to ~/.ssh/authorized_keys -- you may have one or two there already for other systems or other programs.. [18:23] sarnold: ? [18:24] paule32: how about other logs? [18:25] in the log: [18:25] helperHandleRead: unexpected read from blockscript #Hlpr0, 4 bytes 'ERR [18:25] sarnold: sorry, i think thats what i meant. rename the .pub to authorized_keys and normally ftp it to the server [18:26] paule32: interesting, that may be an error from connecting to the mysql daemon. add more debugging around the connect failure [18:45] hm ok, so i ftp'd over the .pub key from the cert i generated to my sftp backup server and put it in ".ssh/authorized_keys", but it still won't allow me to connect through sftp [18:49] Deliant: check ls -ld ~/.ssh ~/.ssh/authorized_keys -- the ssh daemon is very picky about those having correct permissions [18:50] its 600 [18:51] Deliant: and ~/.ssh/ ? [18:51] when i try adding a key to my desktop filezilla client it says the key format is not supported for sftp and asks if i want to convert it to .ppk? [18:51] ah right, not 600 [18:53] Deliant: are you sure with filezilla that you're not confusing sftp and ftps? ftps is every bit as terrible as ftp and should be avoided. [18:54] im very familiar with filezille (i usually use rsync), but when i try to add keyfile under "SFTP" tab it wont accept the rsa key i just generated [18:55] its just saying its not a supported format by filezille though, so maybe not applicable [18:56] and its on my server i'm trying to do this anyways [18:56] damnit, why couldnt they just have used rsync :| === Piper-Off is now known as Monthrect [21:21] for mdadm, do i need to create a partition table for my drives before creating an array via mdadm? [21:24] i dont can't filter url's/domains with helper mysql [21:24] anyone experinces with it? [21:35] Is there anyone here with experience with the Apache 2 MPM ITK module on Ubuntu 14.04? I am trying to use it but get the "No MPM loaded" error when I start Apache. [21:40] OEP: do you have a LoadModule directive in /etc/apache2/apache2.conf (as described on this page: http://mpm-itk.sesse.net/)? [21:40] Pwnna> you need to create a partition table [21:41] yeats: We do have "LoadModule mpm_itk_module ..." in a modules.conf, we are using a custom /etc tree though [21:43] OEP: anything in /var/log/apache2/error.log (or similar)? [21:45] capricorn_1: i just tested in virtualbox and it seem like i just did mdadm --create /dev/md0 --num-devices... /dev/sda /dev/sdb ... i didn't actually explicitly create a partiton table? [21:46] yeats: It doesn't seem to be producing any log output. [21:46] I have this in my notes: [21:46] mdadm --create --verbose /dev/md0 --level=0 [21:46] --raid-devices=2 /dev/sdb1 /dev/sdc1 [21:46] mdadm -Cv /dev/md0 -l5 -n5 -c128 /dev/sd{a,b,c,d,e}1 [21:46] mdadm -S /dev/md0 stop array === kasubai_ is now known as Razzdoll [21:47] mdadm --detail --scan [21:47] you may output the above command into mdadm.conf file which might look like [21:47] this: [21:47] DEVICE /dev/sda1 /dev/sdb1 /dev/sdc1 /dev/sdd1 [21:47] ARRAY /dev/md0 devices=/dev/sda1,/dev/sdb1 [21:47] ARRAY /dev/md1 devices=/dev/sdc1,/dev/sdd1 [21:47] mdadm -As /dev/md0 start an array [21:47] right [21:48] but if you jsut want to use the entire disk do you even need partition tables? [21:50] I would think so. How do you know what kind of formatting is going to be there? I've always partitioned drives for linux ext[3,4] types [21:50] that makes it easier to recover in case of problems. Since you are playing in virtual environment you may try both and see what happens. [21:50] Pwnna: partitioning is optional for non-boot devices [21:51] well don't you just create a lvm ontop of the md[0-9]+ devices and then do mkfs.[fav-fs] /dev/mapper/lv-vg1 or whatever [21:51] Pwnna: for example, LVM LVs rarely use partitioning, unless they're for VM guest images [21:51] TJ-: is this documented somewhere? I can't seem to find it on search [21:51] hmm [21:51] Pwnna: bootable requires some form of partitioning unless it's a floppy (emulated) boot, for example [21:52] OEP: you'll want to make sure that logging is correctly configured - otherwise, you're flying blind [21:52] hmm i guess that makes sense [21:52] because only on boot does the partition table make a difference, or if you want multiple partitions. but it doesn't matter because we're creating logical structures on top of multiple disks with mdadm and lvm? [21:52] Pwnna: ISO 9660 El Torito is another exception, in there's no MBR/GPT required, but that's a different thing altogether [21:52] but the question is is this documented somewhere? [21:52] you are dealing with software raid here so it's necessary to take care of "lower levels" first. No? [21:53] Pwnna: is 'what' documented, precisely? whether or not to use partitions? It's an administrator decision usually [21:53] yeats: Well, if I start it with prefork, it will produce log output, but not with itk. [21:53] TJ-: i suppose i'm looking for why something needs/don't need to have a partition table. everyone on the internet just say use gdisk to create a partition table without justifying this choice. [21:54] OEP: from the glance I took of the page I shared above, it looks like itk is supposed to run alongside mpm-prefork [21:54] (but I may be wrong) [21:54] never having used it ;-) [21:54] Pwnna: if the array is on bootable disks, either those disks need to be partitioned, and the array built on partitions, or if whole-disk arrays, the MD device needs partitioning *and* it needs to use metadata version 1.0 (or 0.9) so the meta-data isn't at the start of the underlying devices [21:55] yeats: Yeah, I can load the two modules together, but when it comes the time to setgid I get a failure in the logs. It makes me think apache doesn't see itk as a compatible mpm. :P [21:55] hey all- was wondering.. im not the greatest, and absolute beginner, is this 300MB storage limit? http://pastebin.com/TAk6urGf [21:55] They have some pretty funny version requirements there on that page. [21:55] TJ-: so if that means I can create a partition table ontop of /dev/md0? like using fdisk or gdisk? [21:55] s/if/does [21:56] Pwnna: yes, of course, once the block device is available you can do whatever you want with it [21:56] I would go by this: https://raid.wiki.kernel.org/index.php/Linux_Raid [21:56] but what would be the point of having a partition table on /dev/sdX when you have a raid + lvm on top? [21:56] and just put your partition table up there [21:57] Pwnna: to make /dev/sdX bootable by the firmware. The RAID might be in a LUKS-dm_crypt container [21:57] Pwnna: On GPT there'll be an EFI-SP or BIOS Boot partition [21:57] yeah so does that mean the data there cannot be raided? [21:57] unless you have a hardware raid or something? [21:57] because it has to live on one disk and one disk only [21:58] Pwnna: correct; unless the UEFI/BIOS 'knew' about the software RAID. With RAID-1 mirrors, as long as the metadata is at the end of the disks, then firmware can still boot from either drive [21:58] hardware RAID can be ignored since even the firmware only sees the 'logical' device, not the underlying devices [21:59] right [21:59] so that means my boot drive cannot be RAID0 or RAID5/6, right? [21:59] because it's not a mirror and things can be everywhere [21:59] that's interesting. [22:00] yeats: Oh, actually I think you are right. I just noticed the call to setgid() is actually coming from itk. Something else must be the matter, like the privilege drop happened too early. [22:00] Creating RAID in virtual machines is pointless. That needs to be taken care of in KVM or core OS in the first place. [22:00] capricorn_1: i'm just testing the process of creation [22:00] I know, just mentioned ... [22:00] i have to provision a handful of servers with the same RAID layout so i want to test.. [22:01] TJ-: i suppose it's possible to put /boot and /efi on one super tiny storage device (4GB flash module), and then have / be mounted from a different disk array backed with raid [22:01] or even multiple storage in RAID1, as you were saying [22:01] interesting. [22:01] Pwnna: yes [22:01] that's pretty cool. [22:02] so do you even need to create a parititon table on the lvm volumes, then? [22:02] does it even matter? [22:02] like MBR/GPT on top of /dev/mapper/lvvg0 or whatever [22:03] Pwnna: I provision servers with RAID-1 mdadm RAID, using meta-data 1.0, with the raw devices allocated to the MD device and the MD device partitioned. The partitions 'show up' on the underlying devices and therefore the system is bootable with or without RAID support in worst-case scenarios [22:03] right [22:03] right, but suppose if you want to mount.. a /data drive backed by RAID6 [22:03] Pwnna: what that means is a single disk can read the boot-loader and boot part of the way even without MD RAID support [22:03] right [22:04] Pwnna: then I'd probably not partition the raw devices, and I'd allocate the 'RAID' device to a VG [22:04] Then use LV for allocating space [22:04] and just not bother with partitions in the sense of GPT/MBR at all [22:04] is this the /correct/ way to do things? [22:05] because i don't know enough about this. this is the first time i've done a none-raid1 setup. [22:05] last time i had a single disk for boot + raid1 for /data [22:06] Pwnna: MBR/GPT is about hard partitioning... once set, its fixed [22:06] LVM takes all the 'hard' away and lets you be flexible and reassign things at will [22:06] ahh that's good to know [22:06] LVM makes it trivial to do snapshotting, cloning, creating mirrors, etc etc [22:07] do you have any experience with LVM2's builtin raid support? [22:07] assigning temporary LVs for testing and destroying them after [22:07] i've tried to do some tests and can't seem to correct things correctly.. [22:07] create things* [22:07] Pwnna: I use LVM mirror RAID for both the data and mirror log on some systems [22:07] having "basic partitions" makes it possible to recover in case you have problems. For example, I have a KVM server with simple / partition with functional OS, i.e. all utils to manage other partitions etc. After that LVM kicks in and some VMs use it as "raw partitions" [22:08] I've done a little bit of lvm raid:ing, and I'd by far trust md above that just for sheer usage base [22:09] hmm [22:09] i'll investigate that later [22:09] thanks all! [22:12] based on what I see here: https://raid.wiki.kernel.org/index.php/RAID_setup I would create partitions, single at that to make it easier. Good luck [22:13] I prefer whole disks [22:17] I'm with Neil there === Monthrect is now known as Piper-Off [22:27] would anybody know please? [22:56] TJ-, are you arround? [22:56] I have problems with my swap again, it seems everytime I update ubuntu-base it screws the swap === Piper-Off is now known as Monthrect === Monthrect is now known as Piper-Off === Piper-Off is now known as Monthrect === Monthrect is now known as Piper-Off === Piper-Off is now known as Monthrect