[00:00] not any hardware raid? [00:00] meh. don't ask. you'd know why I'm worried :) [00:00] zfs raid, j4200, and a... t1000 as a controler [00:01] iirc that's pretty standard hardware - you should be able to grab the drives and mount them wherever you want [00:01] yup I hope so ;) [00:01] but on the same rpool, make sure to remove /etc/zfs/cache or whatever it's called [00:01] zfs layout is being cached on the rpool [00:01] t1000 not doing too well. Am a bit more cautious with that type of assumptions though since that reiserfs issue [00:02] I learnt "filesystem regressions" the hard way ;) [00:02] I've been running a couple of 50TB opensolaris boxes for one and a half years [00:02] they just work [00:03] I know, so do I... its the t1000 I'm worrie dabout, its not much redunded, single ide disk [00:03] hardware failure quite likely in the next 2 years I'd say [00:03] can't you just attach a new drive and create a mirror? [00:03] I should really reinstall a new, better controller [00:03] I could, but t1000 still suck as a zfs controler :) [00:03] zpool attach rpool origdev newdev [00:03] better get a new pci-express card for another server [00:04] yann2: install a new drive in a usb dock or whatever [00:04] yann2: attach it to the rpool [00:04] yann2: install grub, and you have a mirror [00:04] the controller is setup on ufs I think [00:05] the controller? [00:05] the t1000 [00:05] oh [00:05] how is the storage attached? [00:05] FC? [00:05] I got a raidz2 on 6 disks on the j4200, attached via sas to the t1000 [00:05] ok [00:05] just order a new pizzabox from somewhere, move the SAS controller, done [00:06] that SAS card is several hundred quids, is what has been blocking me for the moment :) [00:06] yeah... you heard about the difference between pci-e and pci-x? :D [00:06] oh, pci-x card? :) [00:06] I read about it when I tried what you suggested /o\ [00:06] lsi has some pretty decent cards that doesn't cost too much [00:06] "for some reason it wont fit.... press harder... GGNNNN" [00:06] :P [00:06] 9211 is one of them [00:07] the worst part is I think we got it to fit :'( [00:07] the new box will be pci-e, right? [00:07] 9211? is good? [00:07] the 9211 will fit there [00:07] 9211 is very fast, but you'll get WWN-based device names [00:07] writing that down... cant remember which one is -e which one is -x though [00:07] which somewhat sucs [00:07] -e is the new one [00:08] want to replace the t1000 by a x4100 I have spare, should be much faster [00:08] I m supposed to get a new budget in a couple of months, I'll see if it fits in the list :) [00:08] with only four drives, I think that will be the bottleneck [00:08] cant be much worse than right now [00:09] you're saying you trust 100% the implementation of zfs on linux though? [00:10] installing that solaris box was like... urrrrrg :'( [00:10] wouldnt really want to do it again :) [00:12] thanks for your recommendation on the card btw, I'll write that one down ;) [00:14] yann2: we changed to some 3Gbps SAS boards instead [00:14] just to get rid of the WWN naming [00:15] supermicro and that controller didin't speak well [00:15] so we didn't know which drive was where [00:15] and with 160 drives, you don't want to lookup the WWN [00:18] * He4D ist away (Forever Alone) [00:18] yann2: we ended up with some sas3081 controllers and their internal counterparts [00:19] 3801 [00:19] works well [00:24] yann2: http://pastebin.com/Atkpzux5 [00:35] hi guys. I'm in the process of rebuilding a (file)server from scratch. planing on installing 10.4.1 LTS, and has a hardware RAID instaled. could anyone please recommend a filesystem that would be best-suited for optimum thoughput & stability? [00:36] kieppie: most will work, how much data do plan to put there? [00:37] * RoyK wonders if kieppie even watched the discussion after his initial quiestion [00:38] hi RoyK: thanks for the response. currently there's about 600 GB I'm backing up (`rsync -av`), but tit could very well grow well above the TB's mark in the coming year or so. [00:39] kieppie: ext4 is safe, well tried and works, nothing fancy but it works [00:39] i did try & follow the discussion after my initial question, but it either didn't seem pertinent, or I missed out on a chunk... [00:39] RoyK, hey, be happy to have cache disks :) [00:40] kieppie: and as I tried to tell you earlier tonight, try to read up on zfs if you want to do serious data storage [00:40] yann2: wot_ [00:40] ? [00:40] cool, thanks. that's the de-facto default I would've gone with, but just wanted a second opinion whether something like zfs or brtfs wasn't better-suited [00:41] http://pastebin.com/Atkpzux5 < you got ssd caches right? :) [00:41] kieppie: zfs kicks ass, if you need it [00:41] RoyK: thanks for ZFS advise. I'll lokk into it [00:41] kieppie: http://hub.opensolaris.org/bin/download/Community+Group+zfs/docs/zfslast.pdf [00:42] cheers [00:42] RoyK, I think opensolaris is dead though, not sure about new deployments... [00:42] kieppie: don't misunderstand, zfs is great for large storge, but is a bit on the heavy side for small stuff [00:42] yann2: openindiana :) [00:42] not sure about the exact status though but it didnt seem very bright [00:43] well, you might not be very bright, and that's ok [00:43] RoyK: sweet. by heavy, do you mean in terms of system resource overhead, or something else? I'm setting up a new server from scratch, so if ZFS will deliver excellent performance, then it seems a good idea... [00:44] zfs has a toll of memory [00:44] you want 4 or 8 gigs of memory for a start [00:45] if you start playing with dedup, which you SHOULD NOT DO, well, if you still do, add a lot of L2ARC (that is SSD for caching) [00:45] but then, dedup on zfs is not stable [00:45] I've been testing it quite extensively [00:45] and chosen not to use it [00:45] cool. well, memory scales well, so I can add as much is needed. [00:46] 4-8gigs should be sufficient for a decent storage server [00:46] that is, 20TB or so [00:46] just don't use dedup [00:46] and if you're in doubt, don't use dedup [00:47] for your 4TB server, 2TB will be fine [00:47] 2GB even [00:49] well, this seems to be a old-ish server (2 or 3 years), & has only 1 G. I think I'll pop out & go get more RAM; 8 G (4 x 2G) [00:50] dell "poweredge" xeon 3.2 G [00:50] how much storage do you plan for that one? [00:52] RoyK, for 4TB ext3 should be fine :) [00:53] so, other than "just selecting" ZFS as my FS of choice when doing the install, & avoiding dedupe, are there any other consideration I need to look out for? [00:53] At the moment it has 3 x 1 TB SATA's on a Hardware RAID (which I'll leave as-is) , which will have to do for a while. I'll clean it up a bit once the box is on-line again, but in terms of adding more storage, I'll rather wait for $$$ to become available & then build an entire new box from scratch, possibly with fibre-channel SSD's, etc [00:55] no need for fibre channel [00:55] sata is just as good [00:55] cool [00:55] just read up on zfs and you'll see why you should or should not use it [00:56] * RoyK er farinna að sofa [00:58] reading the wikipedia article now & hav a few other relevant tabs open. seems a good fit; just meed to throw more RAM at it [00:58] kieppie: just one thing - hardware raid is bullshit compared to zfs in terms of safety [00:59] RoyK: to be safe, I'll use both [01:00] kieppie: no, you misunderstand [01:00] kieppie: if zfs has access to the drives directly, it can prevent data loss, far better than any hardware raid [01:00] oooooh! [01:01] if you use zfs on top of hardware raid, zfs can merely detect data loss [01:01] I've seen that a few times [01:01] ok, then I think I'll hold off for now (may be a bit overkill just yet), & do some experementing for a future build... [01:02] hardware raid systems with zfs filesystems on top and oops, corruption [01:02] oh, I see: "double correction"? [01:02] thing is, those hardware raid systems don't see data as a whole, only blocks, and merely that [01:03] if, no, when, you get silent errors from a drive, you want the filesystem to fix that [01:03] most filesystems relies on the drive reporting wheather the data is ok or not [01:03] if it reports ok, the filesystem just sends the data up to user or kernel or whataver [01:04] but with terabytes of data, you _will_ get silent errors [01:04] but in terms of physicality, is it not always faster for have multiple disks to read from? [01:05] the difference isn't big between reads and writes [01:06] if you leave it to a dumb raid controller to sort out what's good or not, silent errors will make corrupt data [01:06] that's a fact, not fiction [01:07] so, drop the h/w raid controller & add system resources to handle FS overhead. [01:07] hmmmmm.... FreeNAS.....? I think it uses ZFS internally, & already optimized for thoughput & FS functions.... [01:08] just setup an opensolaris system or something with stupid controllers [01:08] & it's BSD ..... :D [01:08] direct access to the drives [01:08] freebsd if you like [01:09] know FreeBSD better than solaris.. [01:09] freebsd has a very old zpool version, so I wouldn't recommend it [01:09] think there's much life left in Solaris? [01:09] well, solaris isn't that far apart [01:09] there's a lot of solaris users [01:09] and openindiana [01:10] I just setup two 100TB boxes on OI, and I don't regret it [01:10] wow! 100TB? pretty sweet :) [01:11] two of them [01:11] 160TB raw storage in each [01:11] but leave some redundancy, and we're at 2x100TB [01:11] planning on starting a hosting co, or just a killer media-center? [01:12] heh - bacula backup storage :) [01:12] SaaS? hosted backups? [01:13] just some supermicro boxes with SAS controllers and a truckload of WD Black 2TB drives [01:14] cost us about $20k a piece [01:16] for long term storage, are the WD green's not better? [01:16] eveything will probably work [01:16] but the scrub times for those greens will be terrible [01:16] I have a 30TB setup with those [01:17] ah [01:20] thanks for your help & advise, RoyK: I'll head out & go get some more RAM while I wait for this backup to finish. [01:25] * He4D ist away (Forever Alone!!!1) === He4D is now known as He4D|OFF === loadedwithwyrms is now known as loadedwith|sex === loadedwith|sex is now known as loadedwith|shhh [01:58] is https://help.ubuntu.com/community/NetworkConnectionBridge still the best way to bridge? [01:59] uses pre-up brctl bla bla [01:59] too many lines have they made somthing simpler to go the networking file? [01:59] go in the [02:00] is there an ubuntu debian way I mean [02:01] rather than just using brctl directly [02:01] no? [02:05] sabgenton: that page is overly complicated [02:07] sabgenton: i privated you a simple configuration === Tankenator_afk is now known as Tankenator === Tankenator is now known as Tankenator_afk [02:43] How do you configure which services start at boot time... any tool for that? [02:50] Pupeno[work], depends on what services it is :) [02:54] patdk-lap: mysql, postgresql, apache. [02:55] pmatulis: so bridge_ports eth0 eth1 would do all that I need without needing to type brctrl ? [02:55] use update-rc.d [02:56] if i just want to bridge eth0 and eth1 with bridge utils [02:58] patdk-lap: actually, it is for crashplan... and it didn't work. [02:59] pmatulis: ok sweet [02:59] found http://manpages.ubuntu.com/manpages/lucid/en/man5/bridge-utils-interfaces.5.html [03:00] sorry for ever dobuting you [03:00] https://help.ubuntu.com/community/NetworkConnectionBridge is way out of date then [03:00] works but not showing the ubuntu way at all [03:04] hi, I have a esxi box P4, that has two 500 gigs disk, and one usb flashi key (2 gigs), I want esxi on the flash key, and I want raid 1 on the two 500 gig disk then I want to have one freenas applaince, and one ubuntu server applaince that all is that possible if so how is that possible [03:07] mrroth: would you like some fries with that? [03:07] hmm [03:07] mp [03:07] no just help pmatulis [03:27] hello all, i was wondering if someone could help me please with wordpress [03:28] ##wordpress [03:29] thanks [03:31] * sabgenton put a one line update @ https://help.ubuntu.com/community/NetworkConnectionBridge [03:33] just told people to go to the bridge-utils-interfaces man page for the bridge_ports option [03:41] mrroth: thanks for passing me on to #wordpress but no one will talk to me on there, despite the fact there are people talking on there [03:42] :( [03:42] sorry, Im sorry to winge, I've just used irc a few times now for help and ended up getting no support at all [03:46] ledzeplin1989: what's the problem you need help with [03:47] I tryed to install wordpress earlier, I used a guide from the ubuntu site: https://help.ubuntu.com/community/WordPress [03:48] I followed all the instructions to install it on to my server but for some reason it will not load the wordpress page up to install it [03:49] thank you delimiter :) [03:50] got someone now talking to me on #wordpress but thank you very much [03:50] its greatly appritiated!! :) === loadedwithwyrms is now known as loadedwith|wyrk === Tankenator_afk is now known as Tankenator === SquishyNotHere is now known as squishy === Tankenator is now known as Tankenator_bed === loadedwith|wyrk is now known as Paradox === loadedwithwyrms is now known as trollface [07:56] the version of zabbix is pretty old, any ppa out there? [07:58] 1.8.2 and newist is 1.8.4 [07:58] not that old [07:59] 1.8.4 is release on january 4th === He4D|OFF is now known as He4D [11:07] Hi! I was thinking that is it possible or would it be usefull to run enterprise datawarehouse inside ubuntu server private cloud? What benefits could there be? [11:08] And is it possible for example run PosgtgreSQL inside cloud? [11:14] skoude: you can run anything in a cloud, but for specific things like databases, I'd recommend dedicated hardware [11:18] Yes currently we are planning physical servers:) This was just an idea... It would be nice to see which kind of performance loss there would be when running the DW in cloud :) [11:24] * He4D ist away (Forever Alone!!!1) === He4D is now known as He4D|OFF [11:26] skoude: for a database server, I'd separate that from the cloud [11:26] and then just have the cloud clients access that db server [11:36] hmm. is there any other solutions thn greenplum, gridSQL or pgpool that uses postgres? [11:36] or is working with postgres:) [11:37] sorry wrong channel :) [11:52] hi guys what should i do installing ubuntu server and ask me for driver- my SATA HDD si not detected..any idea guys..? [11:54] ruben23: pastebin lshw output [11:55] where should i run this command im on the instalaltion menu. [11:56] ruben23: shift+left should take you to a console [11:56] ok ill do it now [11:56] you may have to boot on a live cd to get the output out of there :P [11:57] ok ill do it [11:58] RoyK:ill still goto the console right.. [11:58] with a live cd, you can just open a terminal to get that info [12:00] iwhere can i get livecd ubuntu 8.04 LTS [12:00] the ubuntu desktop cd is a livecd [12:01] btw, have you tried 10.04? [12:01] might be newer drivers there for your controller [12:02] RoyK: yes i tried problme is i got application that dont run at all on ubuntu-10.04 LTS [12:02] eaccelerator for php [12:08] ruben23: http://blog.up-link.ro/how-to-install-and-integrate-eaccelerator-into-php5/ [12:08] first hit on google [12:10] is anyone here knowledgeable regarding likewise-open? in particular, verison 5.4 which comes with ubuntu 10.4 LTS server [12:11] DrNick_: I didn't know that one - gotta try that when I get back to work :) [12:13] the basic authentication seems to work fine, however the problem i'm having (and some others) is assuming the default domain. i can change the setting in the likewise "registry", and it applies and refreshes OK, however it doesn't actually work, i.e. trying to authenticate without the domain fails [12:13] i'm tempted just to un-install likewise and do it manually with samba + winbind instead :) [12:14] well, that works too [12:15] would be really nice if they backported the latest likewise-open (which apparently doesn't have the problem) to the LTS version of ubuntu server, as obviously I don't want to upgrade to a non LTS version of the distro - as this server will be in produciton [12:15] * production [12:16] test & dev servers I don't mind running whatever really, but production stuff I like to keep on LTS only versions [12:22] guys im connecting ubuntu 8.04 LTS and ubuntu server 10.04- im installing php but is it ok it will communicate with the other php and mysql..? even they dont have same version..? [12:24] DrNick_: apt-get source likewise-open [12:24] then extract the new source in that dir, overwriting the old [12:25] make a new package, intall it [12:25] dpkg-build-package iirc [12:25] it's pretty trivial [12:25] most of the magick is in the debian/ directory of the source package [12:25] configure string and so on [12:26] New bug: #700288 in amavisd-new (main) "amavisd-release not working with quarantine_subdir_levels" [Undecided,New] https://launchpad.net/bugs/700288 === oubiwann_ is now known as oubiwann [12:54] ruben23, there shouldn't be any problem [13:15] lenios:..? [13:15] running different versions on 10.04 and 8.04 of php and mysql [13:16] is it fine,actually im connecting this two server [13:17] unless you're using features only available on latest versions, yes [13:33] guys any chance how to chnage resolution on ubuntu server 10.10 to more higher [13:35] ruben23: try with dpkg-reconfigure console-setup, in the end you are asked for the font size you want [13:39] its asking for keyboard model [13:39] guntbert: askign for keyboard model [13:40] ruben23: pc105 (intl) should be fine [13:41] what you mean..? [13:41] im resolution not for keyborad -sorry [13:41] ruben23: that reconfigures the "console", consisting of keyboard and screen [13:42] what should i select..? tell me [13:42] ruben23: for keyboard? I already suggested ^^ [13:44] ruben23: ŝlovakish? [13:45] RoyK: im suing US keyboard- im a Filipino- philippines [13:52] i created a dir in /var/run, but everytime i reboot, it disappear and a software starts complaining about missing directory [13:53] Error404NotFound: that's because /var/run is designed that way [13:53] Error404NotFound: you could create it on every boot via rc.local [13:54] gobbe: hmmm, the init scripts run before or after rc.local? [13:54] rc.local is last one [13:54] hmm, then its no use :p [13:55] well, you can configure your init-scripts also [13:56] anyway quite weird that software needs folder and it's not able to create it [13:56] yup, thinking to add a mkdir there. [13:56] gobbe: its a source compiled install. [13:56] yeah [13:57] then manipulating init-scripts might be the best way [14:00] Error404NotFound: just create a newe init script - see the skeleton file [14:00] gobbe: both are S20, what if i create a new script and do it as S10? [14:00] RoyK: thats what i thought as well [14:01] where can i find skeleton file? [14:01] found it [14:03] yep [14:03] RoyK: actually why do i need skeleton? i just need to define PATH, do one mkdir and chown [14:05] it might look nicer if you do it the ubuntu way [14:05] but it's by no means necessary [14:05] originally, unix had one file /etc/rc === Tankenator_bed is now known as Tankenator [14:09] i will need to manually make symlinks right? or can i specify a number with update-rc.d command [14:10] http://twitter.com/timmartin2/status/23365017839599616 [14:10] rofl [14:12] found it [15:02] New bug: #700337 in migrationtools (universe) "No documentation for process "migration"" [Undecided,New] https://launchpad.net/bugs/700337 [15:26] binBASH: ROTFL === Tankenator is now known as Tankenator_away [15:50] how do I repair a pooched grub loader on 8.04? it decided to die after I did a dist upgrade [16:00] jeeves_Moss: grub-install ? [16:00] is this grub1 or 2? [16:00] normally grub1 on 8.04 [16:00] grub 1 [16:01] I'm trying to get booted into the LiveCD. Slow external ROM [16:03] RoyK, still here? I'm booted via the liveCD now [16:03] I'm here [16:03] RoyK, ok, ideas on how to fix this? LOL. dumb upgrade [16:04] hopping around on fucking crutches [16:04] jeeves_Moss: I'd say, sudo in as root, mount the root volume, chroot into it, mount -a, and run grub-install [16:04] RoyK: you going to be around all day? [16:05] no idea, but I guess I'll be around for a while [16:05] ok so I'll ignore my client flashing [16:05] * RoyK is pretty immobilised by a broken leg [16:05] client flashing? [16:06] RoyK, was the broken leg caused by a party induced accident? [16:06] yeah every time you swear my irc client goes nuts [16:06] jeeves_Moss: nah, just tripped in a staircase in Iceland [16:06] on my way to a party [16:06] I just had one wish, to stay a few more days, and that was granted :P [16:06] that sucks!! I could see it being better on the way HOME from the party [16:06] they do like their parties in iceland [16:07] iceland rocks! [16:07] nothing else to do there [16:07] RoyK, ok, I have the partition in question mounted into /mnt/temp, now just run grub-install? [16:07] I flew past it once - nothing but white [16:07] jeeves_Moss: chroot /mnt/temp [16:07] mount -a [16:07] ok, [16:07] then grub-install [16:07] that'll leave you with your old system [16:07] yes [16:08] but I'm unsure about the arguments as of now [16:08] kk [16:08] iirc it just takes the device name, as in /dev/sda or something [16:08] that's the device name of the drive you're on [16:08] ahh [16:09] obviously [16:09] the old /etc/fstab should show the correct partitions [16:09] I should just install grub2 and get it over [16:09] nah [16:09] keep it on grub1 [16:09] grub2 is a PITA if you don't know it [16:10] true, I was thinking more that it's installer should fix the issues @ hand! [16:11] there is a choice in the boot menu for fixing things iirc [16:11] fix a broken system or so [16:11] hummmm [16:11] one sec, going to reboot to see if I can find it [16:11] I don't know if that does grub install, but I would be somewhat surprised if it didn't [16:11] and to think, all of this came from an upgrade [16:12] happens a lot [16:13] weird, I booted into "recovery mode", and now, I get udev_monitor_new_from_netlink: error getting socket: invalid argument [16:17] hi are there any default firewall on ubuntu server 10.10...? [16:17] ruben23, configured and working by default? No. [16:18] RoyK, how can I mount this root partition on this server so I can do a dpkg -reconfigure on it? [16:22] RoyK, how can I get this dumb thing out of read only mode? === squishy is now known as SquishyNotHere [16:22] jeeves_Moss: just mount the root partition, chroot into it, mount -a (to mount proc etc) [16:23] that should be all [16:24] thanks [16:25] jeeves_Moss: I would think running an fsck -f on that filesystem first might be a good idea [16:25] I SOOOOOO don't want to do a fresh install right now [16:28] I'm wondering if I just copy all of my root drive to an external and do a fresh install. My biggest concern is that I have a RAID array mounted (software) at MD0 [16:28] the raid should be safe [16:28] linux will read the config from the drives [16:29] RoyK, that's a 100% sure that it'll be smart enough? I can't loose that data (and yes, there's no backups) [16:29] jeeves_Moss: while in the livecd, can you see md0? [16:29] cat /proc/partitions [16:29] cat /proc/mdstats [16:31] dosn't like the last one! [16:31] jeeves_Moss: do you have a spare drive you can use for a new root? [16:32] RoyK, no :-( [16:32] or an usb pen? [16:32] s/an/a/ [16:32] yes [16:32] well, unplug the root drive, install ubuntu server on the usb pen [16:32] you'll see quite quickly if you can address the raid [16:32] what I was thinking was backing everything up off of /dev/sda to an external drive, then reinstalling [16:33] jeeves_Moss: up to you :) [16:33] could I boot from the liveCD and see if it sees it? [16:33] I'm not sure all the drivers will be loaded [16:33] the livecd sees my 3ware raid [16:33] I was more concerned about the software raid [16:37] well, I guess this is a lesson learned then. [16:37] if it 'aint broke, don't f**k with it [16:38] what version did you upgrade to? [16:38] I just forced a dist upgrade from the CLI [16:39] jeeves_Moss: do-release-upgrade? [16:39] or just apt-get dist-upgrade? [16:39] the latter should be trivial [16:39] do-release-upgrade [16:39] hm... [16:40] :) [16:40] playing with matches [16:40] and a wee bit of gazoline [16:40] jeeves_Moss: if I were you, I'd disconnect the current root disk and try a fresh install on another drive, USB pen or spinning crap, doesn't matter [16:44] I think we may have something here!!! [16:44] http://www.linode.com/forums/viewtopic.php?t=5276%3E [16:49] well, we now have a 10.04 splash screen, so, it's time to see if this works [16:49] ... and we have a login [16:50] :) [16:50] and the raid is there? [16:50] not yet. I'm fixing broken packages right now. [16:51] dpkg --configure -a [16:51] :) [16:51] it's "limping" along right now. [16:51] lol [16:51] apropos limping http://karlsbakk.net/xray.png [16:52] bolly crap man! [16:52] *holly [16:53] it'll heal [16:53] do they take out the metal someday? [16:54] dunno, they said it could stay there if it didn't bother me [16:54] untill you hit a MRI machine! LOL [16:54] just loks like they screwed one bone to the other, so I would think it would lesson movement [16:54] looks [16:55] brb, going to switch laptops [16:55] one of the screws goes through fibula and into tibia, but that'll break, the doctor said, and should't pose a problem [16:55] wow [16:56] another operation, and I'll be grounded for some more weeks [16:57] good thing it's titanium, less hazzle at airports [16:58] dont you have some volcano going off up there? [16:58] Eyjafjallajökull has fallen asleep [16:58] some 8 months ago [16:58] ahh, good [16:58] and, no, I'm not Icelandic :) [16:59] but I've studied the language for some time [16:59] btw http://www.youtube.com/watch?v=9jq-sMZtSww [17:01] I could never pronounce it [17:02] * RoyK can [17:03] but, beleive me, it took me some time to learn the intonations in that language, and I'm still not there [17:14] hi folks. I've just set up a Dovecot IMAP server on Ubuntu 10.04 (migrating from a host running old-ish Debian), I can connect to port 993 locally (e.g. with telnet) but not from other hosts. is there some obvious thing I should be checking, like a default firewall? (#ubuntu said I should try here!) [17:14] oh, to pre-empt the obvious first answer, I have "ssl_listen = *:993" in my dovecot.conf, which is copied from the old server. [17:26] jkg: does lsof -p $pidofdovecotserver show that it actually listens? [17:27] dovecot 14458 root 7u IPv4 6911728 0t0 TCP *:imaps (LISTEN) [17:28] suggests yes to me, but I'm not entirely sure what I'm looking at :-) [17:39] might be firewall [17:40] yeah, that seemed the logical answer - but I don't _think_ I'm running one - iptables -L doesn't show any rules, and I've not installed anything else [17:40] looks like its listening. [17:41] ubuntu doesnt ask to install a firewall, if I remember [17:41] what are you typing into the remote host? [17:42] name or ip address? [17:42] telnet nephos.uk-cvs.com 993 [17:42] try ip address [17:42] but I get the same result by IP - "Trying 84.22.181.182..." [17:42] jkg: pastebin iptables-save output [17:44] http://paste.ubuntu.com/551855/ [17:44] try tcpdump to see that traffic actually comes [17:44] in [17:45] why it is so hard to type correct with mobile phone :-) [17:45] I'm a bit rusty on tcpdump - do I want: "sudo tcpdump 'tcp port 993'"? [17:47] interesting: from the machine itself, I can connect to 84.22.181.182:993 (although my tcpdump command is obviously wrong, since it didn't pick that up ;) ) [17:50] tcptraceroute output is interesting, http://paste.ubuntu.com/551861/ [17:56] it could be your external firewall [17:56] it seems that you are doing somekind of natting [17:57] I am behind NAT, but the server isn't -- but I get the same results from another machine not behind NAT, too. [17:57] aah, ok [17:57] can you open any other connection your server like ssh etc? [17:58] yep, I can ssh to it fine, load web pages from it fine [17:59] jkg: upload your /etc/dovecot/dovecot.conf to pastebin [18:02] hello [18:02] is php 5.3.5 available for 10.04 anywhere yet? [18:02] (I'm mostly interested in php5-fpm) [18:03] Version: 5.3.2-1ubuntu4 [18:03] so no [18:03] if you mean supported packages [18:04] ahah! after all that, it /was/ a firewall issue. I just called our hosting provider, by default they block port 993 inbound ()! [18:04] :) [18:04] jkg: asking from someone else helps usually ;) [18:04] I don't really care if the packages are unofficial since 5.3.5 fixes a pretty huge security hole [18:05] jkg: I understood it did work before (on debian) ? [18:05] sorry, the debian machine was in a different network location too, I should have mentioned that. [18:06] http://www.dotdeb.org/2011/01/07/you-really-should-upgrade-to-php-5-3-5-or-5-2-17/ [18:06] jkg: :-) [18:06] Thorn: the security fixes may be backported - that stuff happens a lot in debuntu land [18:06] Thorn: check that first [18:06] I couldn't imagine it would be the network provider's firewall to blame, so I just mentioned the stuff I thought might be relevant :-) so this means /all/ their customers have 993 blocked inbound... oddness. [18:07] I'd name and shame them, but I think I've disclosed enough information for anyone interested to work it out ;) [18:07] jkg: can you telnet to the port on the server's address? [18:07] not localhost [18:08] from the server itself? yeah. that was the final clue that I needed to ring the network provider and ask them about their firewalling :-) [18:08] jkg: and preferably from a box on the lan [18:08] jkg: ufw status [18:09] jkg: or pastebin iptables-save [18:09] nope, last update 2010-09 https://launchpad.net/ubuntu/lucid/+source/php5/5.3.2-1ubuntu4.5 [18:10] php5 (5.3.2-1ubuntu4.5) lucid-security; urgency=low ...... ECURITY UPDATE: arbitrary code execution via empty SQL query [18:10] arbitrary code execution is "low"? [18:11] RoyK: it's cool, I spoke to the vendor and they're changing the firewall config. thanks, though. [18:11] jkg: heh - ic [18:11] that's not the bug I'm looking for [18:12] Thorn: building a new package from source won't be too hard, though [18:12] apt-get source [18:13] extract new source into that dir [18:13] dpkg-build-package [18:13] iirc [18:13] there is no source package [18:13] there are source packages for all ubuntu packages [18:14] just get the source, unpack the stock php tarball into that, and make a new package of that [18:14] but as fas as I can see there is no ubuntu package which would include that fix [18:14] erm [18:14] oh, interesting [18:14] most of the magic is in the debian/ directory in the source tree [18:15] that says what to build and where to install it plus some package magick [18:15] I'll try that, thanks [18:26] * RoyK leans back and watches the usual suspects [18:33] RoyK, isn't pbuilder prefered over dpkg-build-package to build binary packages? [18:35] lenios: pbuilder uses dpkg-buildpackage. It's a higher level tool. [18:36] Generally it's better to use it. [18:37] any reason to recommend dpkg-buildpackage then? [18:37] hi guys how to make my hostname resolve to my local IP on my onw ubuntu server..? [18:38] hi all. please give me a suggestion of a good (non CLI) GUI replacement for webmin, a web interface to control some servers, mainly LAMP. but if possible, more servers on the same tool package . tks ppl. [18:39] (let me add i intend to use the interface trough lan, i.e. blocking outside world) [18:40] what else do you need outside LAMP? [18:40] ruben23, define it in /etc/hosts [18:41] lcb: what else do you need to configure outside LAMP? [18:41] lenios: got no idea, how the format would it be.. [18:42] format is : IP hostname [18:42] ruben23: read "man 5 hosts" [18:42] JanC: I'm trying to build now the server, so, let's say, FTP, for instance and as time goes i'll add some multimedia servers. [18:42] JanC: but i'll be happy if it's onçy for LAMP, at this point. [18:42] onçy/only [18:43] JanC: due to the suspended webmin for ubuntu and debian i wonder if there is any workaround to avoid the glitches on it though. [18:44] lenios: i got already : 127.0.1.1 Database.. [18:45] ruben23, paste your /etc/hosts [18:45] paste on a pastebin! [18:45] !paste [18:45] For posting multi-line texts into the channel, please use http://paste.ubuntu.com | To post !screenshots use http://tinyurl.com/imagebin | !pastebinit to paste directly from command line | Make sure you give us the URL for your paste - see also the channel topic. [18:46] lenios: ------------>http://pastebin.com/VM1JG4cK [18:48] lenios: you there..? [18:48] ruben23, what does "host Database" return? [18:48] ruben23, instead of 127.0.1.1, you should use your interface's IP for your host's name, i.e. 192.168.1.123 [18:49] http://pastebin.com/P3EH3TTj [18:50] and for what it's worth, "Database" is a potentially very bad hostname if you ever have to talk to anyone about your server by its hostname [18:52] example, "I can't connect to database!", does it mean they can't connect to the server called Database, or that they can't connect to a database? [18:53] haha [18:53] qman__: ok ill chnage it [18:53] I also recommend you use lowercase hostnames... [18:54] JanC: ok, i tried testing it- replaced 127.0.1.1 Database- then i get timeout [18:55] replaced with what? [18:58] lenios: i put it- innermaxdb (new hostname) [18:59] with what IP? [18:59] which* [19:00] on the --> /etc/hosts , i ahve this ---> 192.168.2.2 innermaxdb , is this ok..? [19:01] what does host innermaxdb shows? [19:01] host-command doesn't use hosts-file [19:02] lenios: ---->http://pastebin.com/Hr7D9J7J [19:02] still not working [19:04] like i told, host-command is not using hosts-file [19:04] try ping [19:04] oh true, host doesn't use hosts file [19:04] that might sound little silly, but that's how it is ;) [19:05] i'll just have to remember that [19:06] ok, hope there are ways i can resolve my own hostane on for my local ip [19:06] ruben23: ping [19:12] if you have mysql server installed there is also resolveip [19:12] yes [19:13] and you could write a quick script in about any scripting language too [19:37] anybody understand pam and/or exim? I'm seeing log lines like http://paste.ubuntu.com/551884/ in auth.log and can't send mail (my username is uk-cvs.com-james, and I'm getting my password right) [19:38] as before, this is on exim 4.71 on 10.04, using an exim config that worked previously on exim 4.63 on Debian etch; I guess it's a pam issue, or an exim-running-as-wrong-user issue, rather than an exim config thing... [19:39] the relevant authenticator in the exim config looks like http://paste.ubuntu.com/551886/ [19:43] guys is there a reason why dnssec-keygen does not work ? [19:43] it seems to hang [19:45] MatBoy: dunno, try using strace [21:18] good evening [21:18] moin [21:18] ;) [21:19] whatever :) [21:23] ;D === He4D|OFF is now known as He4D [21:48] <_ruben> dnssec-keygen hanging .. lack of entropy comes to mind [21:52] _ruben: playing with openvpn? [21:53] or bind [21:53] I guess [21:56] <_ruben> RoyK: was a comment on an earlier q in here actually [21:57] i [21:57] erm [21:57] k [21:57] as in ok [21:58] laptop on the lap and moving around a bit can be challenging [22:05] is there any problem if enter localhost's static ip (192.168....) as the LDAP server to use? [22:19] lcb: that's not localhost [22:19] RoyK, indeed, sorry, i mean the machine ip [22:19] localhost is, by definition, 127.0.0.1 or ::1 [22:20] yes.. [22:21] i'll play around with it and see as i go if everything goes ok, with localhost, instead of any ip [22:23] is there a "repair in place" option for Ubuntu server (like a repair option on XP)? [22:23] jeeves, what do you need to do? [22:25] LinuxAdmin, I did a forced release upgrade, and now, my system is sitting @ a fsck (and has been for ~8 hours) [22:25] the orignal release was 8.04, and now the splash screen shows 10.04 === Thorn__ is now known as Thorn [22:25] LinuxAdmin, apparently, there is a LOT of complaints about that upgrade. I think it was a lack of reason that I did the upgrade [22:26] jeeves_, you probably upgrade your kernel [22:26] have you tried to run a previous version of the kernel? [22:27] kernel upgrade can sometimes break things [22:27] LinuxAdmin: eeeeeerm [22:27] LinuxAdmin, yes, and it says it can't find the kernel in question, and locks up [22:27] LinuxAdmin: most filesystems have been pretty rock solid for years, a minor upgrade won't do much [22:28] lol, so.... since this is a production box (hangs head in shame), I've killed Apache, and our e-mail [22:28] have you checked your grub configurations? [22:28] yes [22:28] I had ubuntu server box that broke my drbd shared storage [22:28] that's why I remembered that [22:28] LinuxAdmin: most filesystems are safe [22:29] drbd is another story [22:29] ok, but sometimes kernel upgrades break things [22:29] jeeves_: reboot to single and fsck it [22:29] LinuxAdmin, I was able to get ONE boot out of it (before it went into this endless FSCK loop) on 10.04. There was SOOOOO much broken stuff that I'm thinking that I'm just going to back up the 200Gb main drive, nuke it, and do a fresh install. My biggest concern is that I have a 1Tb software RAID in that box as well, and I hope I can recover it [22:29] that's why he could see if he can boot with another kernel [22:29] RoyK, hey man, it's the same issue from earlier today. [22:29] jeeves_: the raid stuff_ [22:30] ? [22:30] currently, I'm out for dinner with the wife, so I'm on my phone, I'll have to check tomorrow when I get it on the bench @ work. [22:30] jeeves_: just don't listen to that LinuxAdmin dude [22:30] yes, I had a RAID0 set sitting @ /dev/md0 that is 2 500Gb diss [22:31] * He4D ist away (Forever Alone!!!1) [22:31] sorry, sir, but raid0 is asking for trouble [22:31] royk, depends on what it's used for :) [22:31] for caching, spooling, something, ok [22:31] but I somehow doubt you'd need a terabyte for that [22:31] it was to be a temp soulition untill I could get $200 for 2 2Tb disks that would be mirrored. [22:31] RoyK, lol, spooling of Porn [22:32] royk, video, but anything <30tb is small for that [22:32] ;) [22:32] ... or Porn of spoling [22:32] ok RoyK, I'm not here anyway [22:32] ok, so, lessons learned here... back up, back up, back up. [22:33] real men don't use backups [22:33] I just hope I can pull a copy of the configs for e-mail, data bases, etc [22:33] real men weep [22:33] .... then we get creative and put the damn thing in the freezer! [22:34] * RoyK doesn't want his macbook pro in the freezer [22:34] heh, I had a raid50 die last sunday :( [22:34] RoyK, FRUIT??? seriously? [22:34] 7 disk (6 + 1hotspare) [22:34] 5 bad drives [22:34] recovered almost everything [22:34] patdk-lap, WTF? please tell me there was a bottle of booze found in the server room [22:34] patdk-lap: zfs would have found the errors early [22:34] zfs ftw! [22:34] the raid card should have too [22:34] nope [22:35] the raid card listens to the drives only [22:35] no it doesn't [22:35] if the drives say it's ok, the raid card doesn't give a fuck [22:35] I had consistency checking on, every other day it does a sweep [22:35] lol, that's why our "large" storage array in the data center is a 12 bay unit with 2Tb drives in it. [22:35] patdk-lap: that's just asking the drives [22:35] patdk-lap: no data consistency checks [22:36] we're thinking ZFS is the way to go for it for us. [22:36] it takes 16hours to run a consistency check [22:36] only zfs and perhaps btrfs do that [22:36] it has to check it [22:36] half of it was also the scsi cable had gone bad [22:36] patdk-lap: there is no chance your controller can run consistency checks on the drive data without pre-stored checksums, which it probably doesn't have space for [22:37] drives that where fine, where marked bad [22:37] anyone heard of the power requirements on a NAS with ZFS? I'm thinking of buying a dual CPU board for the 2 3Ghz Xenon CPUS and 8Gb of RAM I have, but according to the specs, a P4 is the best [22:37] and drives that where bad, where still good [22:37] New bug: #700492 in ntp (main) "ntp complains about ipv6 errors every 5 minutes" [Undecided,New] https://launchpad.net/bugs/700492 [22:37] jeeves_: there aren't really much power requirements for zfs, just your regular hardware [22:38] RoyK, ok, cool. saves me some $$!!! [22:38] jeeves_: and for zfs, not much cpu is needed [22:38] jeeves_: just make sure it's 64bit [22:38] hmm, isn't the network or disks the normal slowdown? [22:38] use quad qdr infiniband :) [22:39] well, I figured with 12 disks to work with, and those 12 disks spaced out over 4 4port SATA cards, the slow part should be the PCI busor the NIC [22:39] SAS3 or perhaps SAS6 should be sufficient for most use [22:39] what do the sata cards use? pci? pciex1? [22:39] and for our production boxes, we have 2Gb FC cards in them [22:39] patdk-lap, PCI (it's an old 2.4Ghz P4) [22:40] oh horrible [22:40] we're using SAS3 for this 100TB setup, works well [22:40] max the whole system will get is 100MB/sec [22:40] RoyK, how much $$ backing do you have though? My busniess partner and I are still in the "start up" phaze [22:40] patdk-lap: depends how many drives you need, how much speed you want etc [22:40] you can get a cheap ass pcie system [22:41] that would be much faster than pci slots :) [22:41] true [22:41] I should get a better board [22:41] jeeves_: about $25k for 100TB net storage, 160TB raw space [22:42] RoyK, nice. well, we're still young/poor, so, everything we have comes from making "deals" and plotting along on a shoestring budget [22:42] heh, I did mine for cheap [22:42] jeeves_: this is off the shelf from supermicro [22:43] once we get going, I'll start replacing stuff. Consolidate all the web servers into a single blade cab, etc [22:43] a $30 mb, a simple intel 945 cpu, 4gig ram, pcie intel nic, and put in some pci sata dual port cards [22:44] patdk-lap: install something like openindana on that, add a bunch of drives [22:44] setup zfs to do its business [22:44] I was thinking a cheap intel board, 5 PCI-E slots, a dual (or quad NIC), and the rest are 4 port SATA cards [22:44] that was my network storage/mythtv system [22:44] I was thinking FreeNAS [22:44] hi patdk-lap ;) [22:44] nah - use openindiana [22:44] why? [22:44] remember, sata2 drive can max out a pcie x1 if it wants :) [22:45] hello binbash [22:45] jeeves_: freenas is based on freebsd, which has a very old zfs implementation [22:46] openindiana development is quite in the game [22:46] lol, that's the point. I'm trying to reduce as many bottle necks as I can on that box [22:46] RoyK, lol, maybe I should dust off my SunBlade 1500. [22:46] the zfs implememntation in fbsd isn't good [22:46] Finally my voip phone works \o/ Installing Asterisk 1.8 fixed everything. === He4D|OFF is now known as He4D [22:47] ok boys, I'm outta here. I have to go pick up my fiance, and go. No more sitting in the car waiting!!!! FINALLY! [22:53] binBASH: I used to work with asstrix === deadsmith_ is now known as deadsmith [22:56] asstrix? :D