[00:09] <lorenz> TJ-: You remember monday when we tried to figure out how to get the server back online? I then reinstalled and today it happened again. Exactly the same problem. Approximately the same installation, running less than 36 hours, then hung up (network didn't work anymore, local console login timed out). I couldn't do anything besides rebooting and exactly the same problem appeared again.
[00:09] <pakcjo> how can I be sure that slapd gets started before krb5kdc?
[00:11] <TJ-> lorenz: !!!! on bare hardware or VM, or both?
[00:11] <lorenz> TJ-: Again on both.
[00:12] <lorenz> TJ-: I already checked initramfs, removed anything that isn't on my host machine and rebuilded. Seems the same. I now got detailed file logs on what has changed in the last 36 hours.
[00:14] <Gallomimia> im having trouble with the installer not giving keyboard access anymore after entering the recovery mode in its grub menu
[00:15] <Gallomimia> it's been working fine for days. then suddenly no keyboard in installer
[00:16] <TJ-> lorenz: have you done a memtest86+ on the bare metal?
[00:16] <lorenz> TJ-: No, I haven't. But I have done one  on the KVM host, so it should boot at least there.
[00:16] <RoyK> the only way to run memtest is on bare metal
[00:17] <RoyK> lorenz: a kvm host should be bare metal
[00:17] <RoyK> and it probably is
[00:17] <RoyK> lorenz: using ecc on this one?
[00:17] <lorenz> RoyK: I ran it on the host.
[00:18] <RoyK> all way through?
[00:18] <lorenz> RoyK: for a night
[00:18] <RoyK> should do
[00:18] <TJ-> lorenz: do you test it on the VM only after it fails on bare metal?
[00:18] <lorenz> RoyK: I got standard DDR3-1600, but my host system never crashed in a whole year of usage (5-8h/day)
[00:19] <RoyK> lorenz: probably not memory issues then
[00:19] <lorenz> RoyK: The server first crashed after 20d uptime and now after 36h
[00:19] <lorenz> RoyK: the 36 hour-one is a fresh install
[00:19] <RoyK> lorenz: wrote this little program to just stress memory from inside the OS, it made my system crash easily when memory was bad
[00:21] <TJ-> lorenz: I've had modules go 'funny' in severs after a long time... turned out to be vibration-related... reseating the modules after removing them fixed it
[00:21] <lorenz> TJ-: Ok, I'll see. I'll also run a quick memtest.
[00:21] <RoyK> lorenz: http://karlsbakk.net/tmp/memstress/
[00:22] <RoyK> lorenz: run that over for some time - if that can't crash the system, neither can the vm
[00:22] <RoyK> (IMHO)
[00:25] <Sachiru> Anyone using ZFS on Linux for production on limited systems? How much RAM is the minimum requirements for it? Somehow I can't find that in the documentation.
[00:26] <lorenz> RoyK: It normally runs on bare metal
[00:26] <lorenz> RoyK: KVM is just for debugging
[00:26] <RoyK> Sachiru: 2GB should do, but the more, the better
[00:27] <RoyK> Sachiru: just don't use dedup unless you know what you're doing
[00:27] <Sachiru> No dedupe
[00:27] <Sachiru> Cool
[00:27] <Sachiru> Is that 2GB just for ZFS ARC, or does that already include the OS?
[00:27] <RoyK> the zfs arc should be large enough to sustain most use
[00:28] <Sachiru> Also is it safe to use ZRAM with ZFS?
[00:28] <RoyK> no, arc is separate on linux and freebsd
[00:28] <Sachiru> Ah.
[00:28] <Sachiru> So 2GB for ARC + Whatever OS needs
[00:28] <RoyK> it's only on illumos/solaris where it's somehow integrated
[00:28] <RoyK> it's dynamic, so don't worry
[00:28] <RoyK> zram?
[00:28] <Sachiru> Yep
[00:28] <RoyK> what's that?
[00:28] <Sachiru> The compressed swap in RAM thingy
[00:29] <RoyK> probably
[00:29] <Sachiru> http://en.wikipedia.org/wiki/Zram
[00:29] <RoyK> Sachiru: how much memory do you have on the box?
[00:29] <Sachiru> 8GB RAM, but 4GB shall be consumed by VMs.
[00:29] <RoyK> add a small ssd for l2arc
[00:30] <RoyK> if the VMs reside on zfs, set aside 8GB or so for SLOG
[00:30] <RoyK> just a small partition for SLOG
[00:30] <RoyK> it'll boost performance a bit
[00:30] <Sachiru> Uh
[00:30] <Sachiru> I don't think I'd need L2ARC
[00:30] <RoyK> on the SSD, that is
[00:31] <Sachiru> Or a SLOG
[00:31] <Sachiru> Performance isn't really necessary (low use, test/dev environment VMs)
[00:31] <RoyK> SLOG is very good for sync writes and VMs do a lot of that
[00:31] <Sachiru> Yeah, but I don't have the budget for an SSD
[00:31] <Sachiru> Sadly
[00:31] <RoyK> just get a cheap SSD
[00:31] <Sachiru> The VMs are just small AMP stacks
[00:32] <Sachiru> But sure, when I get into the money I'll add an SSD
[00:32] <RoyK> just trying to talk about how to make zfs better
[00:32] <RoyK> no offence
[00:32] <Sachiru> None taken
[00:32] <Sachiru> So you're sure that ZFS can fit into my needs?
[00:32] <Sachiru> Can I limit/throttle ARC on a linux system?
[00:32] <RoyK> yes
[00:33] <Sachiru> Cool.
[00:33] <RoyK> but I don't remember how
[00:33] <RoyK> try #zfsonlinux
[00:33] <RoyK> they'll know
[00:35] <Sachiru> Thanks!
[00:35] <RoyK> np :)
[00:35] <Sachiru> Originally I was planning LinuxRAID, but, lack of transparent filesystem compression.
[00:35] <RoyK> the sad thing about zfs is the lack of flexibility
[00:36] <RoyK> the good thing aboud MD is the revese
[00:36] <Sachiru> In that you can't define partitions as differing RAID levels, right?
[00:36] <RoyK> you can add and remove drives from an MD raid without issues
[00:36] <RoyK> you can't add another drive in a raidz on zfs
[00:37] <Sachiru> True
[00:37] <Sachiru> BTW
[00:37] <Gallomimia> reading all that made me hopeful for a decent answer to my problem of getting a system to boot with raid5, encrypted filesystem, and LVM. the root filesystem is on there. but in the mean time i've been tinkering with it and can't get past the language select screen on the 14.04 server installer cause the keyboard just dies
[00:37] <RoyK> that'll require block pointer rewrite, which has been thought to be in the works for 5+ years
[00:37] <Sachiru> Is it possible to run LVM on top of ZFS?
[00:37] <bekks> Sachiru: sure, but it makes no sense.
[00:37] <bekks> ZFS is a volume manager already.
[00:37] <Sachiru> Ok.
[00:37] <Gallomimia> i thought about zfs for this system but it didn't sound like something i wanted
[00:38] <Gallomimia> block pointer?
[00:38] <RoyK> Gallomimia: don't do it. boot the root fs from a pair of mirrored drives or even a tiny drive and use separate data drives
[00:38] <Sachiru> Given a choice between linuxRAID and ZOL, if the primary concern is stability and secondary performance, which would be the common recommendation?
[00:38] <Gallomimia> uh...
[00:38] <bekks> Sachiru: dont use ZOL for mirroring your boot devices.
[00:38] <Sachiru> @Roy: What I was planning was LinuxRAID two 8GB flash drives for boot/OS
[00:39] <bekks> Sachiru: USe it for your data stuff.
[00:39] <Sachiru> OS swap + VM storage on ZFS mirrored 2TB drives
[00:39] <Gallomimia> how much space does the "root" fs need? do i need to mount everything in the root dir from separate volumes?
[00:39] <Sachiru> Linux will boot even if it doesn't have swap right?
[00:39] <bekks> Sachiru: Use an OS which support that natively, then.
[00:39] <RoyK> Sachiru: sounds reasonable
[00:39] <Gallomimia> doesn't need swap
[00:39] <RoyK> Sachiru: just keep the data and the root separate
[00:40] <Sachiru> Or should I keep swap on root drive?
[00:40] <bekks> You need swap. :)
[00:40] <bekks> Sachiru: Keep it on the root disk.
[00:40] <RoyK> bekks: no, you don't
[00:40] <Gallomimia> yeah. YOU need swap
[00:40] <Sachiru> Ok
[00:40] <bekks> RoyK: For suspend2disk - you need swap, of course.
[00:40] <Sachiru> A minimal install of lubuntu shouldn't eat up 8GB of space on the flashdrive right?
[00:40] <RoyK> bekks: for a server?
[00:40] <Sachiru> Also no plans of suspending
[00:40] <Sachiru> For a server
[00:40] <Gallomimia> me either
[00:40] <Gallomimia> if i need to suspend, better to shut down
[00:41] <Sachiru> This is intended to be a KVM server. Lubuntu simply for the lightweight GUI
[00:41] <RoyK> Sachiru: then install ubuntu server without the gui
[00:41] <Sachiru> All it will do is start, stop and manage VMs. Everything else will be in a VM inside, and Lubuntu because I'm not yet that comfortable with configuring KVM from the command line
[00:41] <RoyK> no need for a gui on a server
[00:41] <bekks> RoyK: for various use cases, you actually need swap, on a server.
[00:42] <RoyK> bekks: no, you don't. you need swap when you get a memory leak, but the swap will fill up and it'll take longer to find out what happened
[00:42] <RoyK> bekks: because with swap, it'll just slow down, you can't log in blablabla
[00:42] <Gallomimia> my server's a desktop. but i want to do some server stuff on it. all im trying to do is boot up the rescue mode of the server install. my keyboard dies at that point. any ideas?
[00:42] <Sachiru> So what's the consensus? Swap on rootFS (LinuxRAID mirror on two 8GB flash drives)?
[00:42] <Sachiru> Or swap on ZFS data drive?
[00:42] <RoyK> bekks: without swap, it'll crash gently
[00:43] <RoyK> Sachiru: no
[00:43] <Gallomimia> dont put your swap on a usb drive
[00:43] <Gallomimia> wow.
[00:43] <bekks> You do. Like the requirements of Oracle RDBMS clearly state. So the general assumption "you do not need swap on a server" is wrong.
[00:43] <RoyK> geowany`work: no!
[00:43] <RoyK> Gallomimia: no§
[00:43] <Sachiru> RootFS would be on usb flash drives
[00:43] <Sachiru> So swap on ZFS data drive then?
[00:43] <Sachiru> Ok
[00:43] <RoyK> Sachiru: no
[00:43] <Sachiru> Where would I put swap then?
[00:43] <RoyK> Sachiru: keep the root and swap on the same media
[00:44] <Sachiru> But root is USB flash
[00:44] <Sachiru> Or should I discard swap altogether?
[00:44] <RoyK> then set vm.swappiness = 1
[00:44] <Gallomimia> listening to oracle for advice on running open source software sounds like listening to the advice of the surgeon general telling you cars are perfectly safe, cause they cause more business for surgeons than any other market
[00:44] <RoyK> so linux won't swap at all unless it really has to
[00:44] <Sachiru> Ok
[00:45] <Sachiru> Is it required for swap on Linux to be the same size as RAM?
[00:45] <RoyK> if you have the root on usb sticks, set vm.swappiness=1
[00:45] <RoyK> no
[00:45] <Sachiru> Or can I have 2GB swap with 8GB RAM?
[00:45] <RoyK> just make a small thing like 128MB
[00:45] <Sachiru> Ok
[00:45] <RoyK> if you set vm.swappiness=1, it won't use swap
[00:45] <bekks> Gallomimia: do you actually earn money with running Oracle RDBMS? If not, I gently discard your post.
[00:46] <RoyK> running oracle on ubuntu?
[00:46] <RoyK> that's not supported afaik
[00:46] <RoyK> so you won't make much money ;)
[00:46] <Gallomimia> true enough, but can we agree that oracle or its software is not anywhere close to this topic?
[00:46] <bekks> RoyK: It isnt. It just tried to point out that the general assumption "you do not need swap on a server" is wrong.
[00:47] <Sachiru> Ok
[00:47] <RoyK> bekks: you don't
[00:47] <Sachiru> Thanks for the advice
[00:47] <bekks> RoyK: I did, read the backlog please.
[00:47] <Gallomimia> i ran servers with no swap for awhile. i eventually put swap in. and the problems of memory leaks were as the rest of the channel mentioned
[00:47] <RoyK> bekks: if you need swap on a server, you have too little memory
[00:48] <bekks> RoyK: If you do not read the requirements section, no one can help you.
[00:48] <Sachiru> I'll do  / and 256MB SWAP on USB flash drive with VM.swappiness=10% (for ZRAM), then /home and /VM-Storage on the ZFS data disks
[00:48] <RoyK> bekks: I know linux uses swap a lot to make room for active memory. it's true. but it's also true that some versions of the kernel abuse this rather a lot, and start swapping out even with lots of gigs of memory available
[00:48] <Gallomimia> the only time i needed the swap, was when there was a memory leak. and only then just so i could kill the server between levels and not in the middle of a game :P
[00:48] <bekks> RoyK: which doesnt change the point that the general assumption is wrong.
[00:49] <Gallomimia> anyway, in ubuntu it's possible to have more than one swap file
[00:49] <RoyK> bekks: I've been using linux for a few years, almost 20, I know a bit about how it handles memory
[00:49] <Gallomimia> on different volumes if desired
[00:50] <bekks> RoyK: I am, too. But still there is software out there that requires swap, no matter wether it is used or not. So again, the general assumption is wrong.
[00:50] <lorenz> TJ-: So, Memtest is now running, after one pass no errors. But I'm letting it run. What about the system?
[00:51] <Gallomimia> again you're correct in a general sense, but this guy told us his entire usecase and it doesn't actually need swap. aside from the fact that he's using up most of his memory in vm's
[00:51] <RoyK> bekks: no, software doesn't require swap. all memory usage is virtual. if the memory consumption is higher than what linux has, it swaps out. if vm.swappiness is set high, it swaps out early to free rarely-used pages for those who use them
[00:51] <RoyK> bekks: but no software require swap
[00:51] <bekks> RoyK: So did you read the Oracle RDBMS requirements section, actually?
[00:52] <Gallomimia> 1. are we using oracle rdbms?
[00:52] <RoyK> bekks: I don't run oracle, so no
[00:52] <RoyK> bekks: I use open source software only
[00:52] <bekks> I can safely consider this discussion to be pointless then.
[00:52] <lorenz> I can confirm that, on many of my PCs and server I don't have swap set up and everything worked fine. If you have enough RAM it doesn't matter.
[00:52] <Sachiru> @lorenz: One pass isn't sufficient in my experience
[00:52] <TJ-> lorenz: how do you mean?
[00:53] <Sachiru> I've seen several systems where it required two or three passes to catch intermittent errors
[00:53] <RoyK> bekks: when was oracle a part of this discussion before you introduced it?
[00:53] <Gallomimia> 2. just because this one piece of software says it needs swap space doesn't mean all or any software actually needs swap space. it's much more likely that the software at hand is poorly written and abuses its privilege of having swap space.
[00:53] <lorenz> TJ-: Was a response to the swap discussion.
[00:53] <Sachiru> My current practice is to run memtest for at least 12 hours (three passes) before declaring it "semi-stable"
[00:53] <Sachiru> Unless it's ECC RAM, in which case two passes is enough.
[00:53] <Gallomimia> 3. no more oracle in FOSS discussions please
[00:54] <lorenz> Sachiru: Yes, I'll let it run.
[00:54] <bekks> RoyK: It never was - again, I proved that the general assumption is wrong. However, I'm not going to discuss this any further, since you dont want to see the difference between the "Sachiru use case" and "the general assumption".
[00:54] <TJ-> lorenz: I wasn't reading it :)
[00:54] <lorenz> TJ-: What about the system? I can give you full change lists, but I've already looked through them and there's nothing interesting there.
[00:55] <RoyK> bekks: there's a little difference between running linux systems for 10+ years in production and "assumptions". I know linux. I know how it works. It doesn't need swap until the midden hits the windmill
[00:55] <RoyK> bekks: it's not assumptions - it's what I've learned over the years
[00:56] <TJ-> lorenz: I'm at a loss, really. Aside from memory corruption being flushed to disk I can't imagine what could cause those symptoms
[00:56] <bekks> And you havent touched Oracle yet and you are assuming how it should work. I'm off of this pointless discussion.
[00:56] <RoyK> bekks: so please, tell me why we should have a ton of unused gigabytes of swap on our servers ;)
[00:56] <RoyK> bekks: I Don't Use Oracle. Got it?
[00:56] <bekks> So please just read the requirements section where this point is explained.
[00:57] <bekks> And I am off of this discussion. Got it?
[00:57] <RoyK> are we discussing oracle or linux?
[00:57] <RoyK> bekks?
[00:57] <lorenz> bekks: Noone here cares about Oracle requirements. It is simply not relevant.
[00:57]  * RoyK diverts bekks to #ubuntu-offtopic
[00:59] <RoyK> Sachiru: no, it's not *needed* to have swap on linux
[00:59] <RoyK> but a small section of swap is always good
[00:59] <RoyK> say, 1GB or 500MB or so
[01:00] <RoyK> so that linux can swap out things not in use there and use the remaining memory for something useful
[01:01] <lorenz> TJ-: I currently have a very similar system, which (after an upgrade from 12.04 to 14.04) runs for around 2 months. No problems there. My newer host (which is the one that fails) has a newer i7 4770, while the other host has the predecessor i7 3770. Any known issues with Haswell CPUs on Linux?
[01:01] <RoyK> Sachiru: if you have a fast drive and little memory, it may be good to keep a 1-2GB swap space and setting vm.swappiness=100
[01:01] <RoyK> Sachiru: if you have sufficient memory, it's no need
[01:02] <lorenz> RoyK: vm.swappiness=100 just kills the SSD because of the writes
[01:02] <RoyK> Sachiru: if you have low memory and *slow* swapspace, set vm.swappiness=1 or so
[01:02] <RoyK> lorenz: no, it won't swap that much
[01:03] <RoyK> lorenz: it only swaps when it finds it necessary
[01:03] <lorenz> RoyK: I thought at 100 it would swap almost everything
[01:03] <RoyK> lorenz: and most good SSDs these days can handle rather a lot of writes, even cheap MLCs
[01:03] <pakcjo> hello
[01:03] <RoyK> lorenz: no
[01:04] <RoyK> lorenz: it's just an index - at 100 it'll try to work out which pages to swap out earlier. the default is 60
[01:04] <lorenz> RoyK: Ok, something learned :)
[01:05] <RoyK> :)
[01:08] <Sachiru> Well
[01:08] <Sachiru> I was planning on using ZRAM
[01:08] <Sachiru> (compressed swap in RAM)
[01:08] <Sachiru> I guess I can just use ZRAM and no swap
[01:08] <Sachiru> Then vm.swappiness = 100%
[01:08] <Sachiru> So that it uses compressed swap in RAM always (essentially a dirty way of compressing RAM transparently)
[01:09] <lorenz> TJ-: Added files since install http://paste.ubuntu.com/7540516/ , Modified files since install http://paste.ubuntu.com/7540519/  (Excluded /usr/share/man, /var, /usr/share/doc)
[01:11] <TJ-> lorenz: As I said the other day; I blame one of those PPAs!
[01:14] <Sachiru> One more query:
[01:15] <lorenz> TJ-: Well Owncloud is certainly not the problem (Just a web application), Elasticsearch and Percona XtraDB are just database servers, so they aren't even started at the time it fails. That leaves Suricata and Ksplice. I already removed Ksplice because it integrates into initramfs and others and is therefore a candidate for the failure, but even after a complete uninstall it didn't work better. Suricada does not really interact with
[01:15] <lorenz> the kernel (besides NFQUEUE), but I'm trying to remove it and see what happens.
[01:15] <Sachiru> As I said, rootFS shall reside on mirrored 8GB flash drives (Sandisk Cruzers). Should I max out the 8GB, or just partition 7GB and leave 1GB for free space (garbage collection, trim and all that).
[01:15] <Sachiru> Or do USB drives not have TRIM?
[01:17] <lorenz> @Sachiru It depends on the USB adapter and the drive.
[01:17] <Sachiru> So, google the drive and research controller if it implements wear leveling
[01:18] <Sachiru> If yes, leave slack for overprovisioning, if no, allocate all and hope that OS is sane enough to wear-level on its own?
[01:18] <patdk-lap> you do know it's rare for a usb flash drive to do garbage collection or trim?
[01:18] <patdk-lap> and I don't know of any that do it with <128gigs of flash
[01:19] <lorenz> @Sachiru I would assign the full 8GB, even that is barely enough for many applications
[01:19] <Sachiru> No intended applications
[01:19] <lorenz> patdk-lap: I use a lot of Kingston KC300 60GB SSDs and they do TRIM
[01:19] <Sachiru> This is intended to be VM host
[01:20] <Sachiru> Flash drives only host boot data for OS
[01:20] <Sachiru> Everything else is stored onto ZFS data drives
[01:20] <patdk-lap> lorenz, that is a usb stick?
[01:20] <lorenz> patdk-lap: No, but still flash
[01:20] <Sachiru> Also a quick google shows that Sandisk Cruzers have very basic wear leveling (no trim or garbage collection but better than nothing yes?
[01:20] <patdk-lap> I said flash usb
[01:21] <patdk-lap> sure, you can get sata flash at 4gigs that do trim
[01:21] <patdk-lap> but not usb
[01:21] <Sachiru> Also, query: RAID mirror of 2 flash drives where the /boot and / partition resides ensures that the system can boot from either of the two drives, correct?
[01:21] <patdk-lap> Sachiru, nope
[01:22] <Sachiru> Ok
[01:22] <patdk-lap> you have to make sure grub is installed to both
[01:22] <Sachiru> Yes
[01:22] <Sachiru> And that too
[01:22] <patdk-lap> and that bios is set to boot both
[01:22] <Sachiru> Forgot to say that, sorry
[01:22] <patdk-lap> and hope the first one doesn't fail in a half state
[01:22] <davidbowlby> what's the best way to backup my old vmware ubuntu server instance and restore it to my openstack instance?  I'm not looking to do the full vmdk move because it takes too much space.  I've been playing with tar to zip the files and restore, but I'm obviously overwriting stuff that the cloud image uses to mount the boot volume.  Any help would be appreciated.  :)
[01:22] <patdk-lap> !best
[01:22] <patdk-lap> stupid bot
[01:24] <davidbowlby> I've used the VMDK transfer method for my win2k8r2 VMs, but I figured since both the source and destination are running the same release of ubuntu I could probably do some form of backup/restore
[01:24] <lorenz> TJ-: Checked memtest again, still no errors at 3 passes.
[01:24] <davidbowlby> it would save me a ton of space because I have quite a few ubuntu VMs in my old environment
[01:24] <Sachiru> @patdk-lap
[01:25] <Sachiru> What would you recommend then?
[01:25] <Sachiru> I want to make it so that the base OS can boot from either of the two drives
[01:25] <patdk-lap> you can't
[01:25] <Sachiru> So that if one fails I can pull it out
[01:25] <Sachiru> And put the other in
[01:25] <lorenz> Sachiru: BTRFS can do that (I think)
[01:25] <patdk-lap> you just have to hope the first one when it fails, fails badly enough, it boots the second
[01:25] <Sachiru> Ah
[01:25] <Sachiru> Hmm
[01:25] <patdk-lap> heh? btrfs seems to do a lot of things it can't
[01:26] <lorenz> patdk-lap: Well it has checksumming and integrated raid, so it would detect where the faulty data is.
[01:26] <patdk-lap> the only way to solve that fully, is to use a real raid card
[01:26] <Sachiru> Is there an easy way to clone a USB drive?
[01:26] <patdk-lap> lorenz, and at that point? your boot already failed
[01:26] <patdk-lap> and it won't boot the second disk
[01:26] <patdk-lap> so no, btrfs didn't fix it
[01:26] <Sachiru> I'm thinking just use one flash drive, do weekly clone
[01:26] <Sachiru> To another
[01:27] <patdk-lap> it just didn't run the corruption, at best
[01:27] <lorenz> Sachiru: A simple dd would work
[01:27] <Sachiru> In case primary fails just unplug it and plug in the latest clone
[01:27] <Sachiru> Does that work?
[01:27] <patdk-lap> why wouldn't it?
[01:27] <Sachiru> Ok
[01:27] <Sachiru> If I do LinuxRAID (2-way mirror) of the flash drives then unplug one
[01:28] <Sachiru> If I plug it back in later could I use MDADM to rebuild/re-clone the drive?
[01:28] <lorenz> yes, that would work
[01:28] <lorenz> but it is a lot of work
[01:28] <patdk-lap> just keep bitmaps turned on
[01:28] <Sachiru> I think I'd go with LinuxRAID then
[01:29] <Sachiru> LinuxRAID doesn't do "rebuild the moment the drive is plugged back in" right?
[01:29] <lorenz> No
[01:29] <patdk-lap> make sure mdadm is setup correctly to boot in degraded mode
[01:29] <lorenz> At least not in default config
[01:29] <Sachiru> Ok
[01:29] <Sachiru> Thanks
[01:29] <Sachiru> Now I know what to do. Thanks all! You've been such a great help
[01:34] <lorenz> TJ-: I left my VM running and it has made it to the promt where it says that /tmp, /boot/efi and /home aren't ready
[01:34] <lorenz> TJ-: Would point to a udev problem
[01:35] <TJ-> lorenz: Can you checksum the boot files (vmlinuz, initrd.img) and compare them against the originals before the problem?
[01:35] <lorenz> Has anyone of you had udev problems lately?
[01:35] <lorenz> TJ-: Well, there are no originals (at least no this time)
[01:36] <lorenz> TJ-: That system was installed, booted and hung
[01:36] <TJ-> lorenz: You have pretty amazing Linux-killing superpowers!
[01:36] <davidbowlby> could I just boot my vmware ubuntu server and use rsync in some way to transfer its files to the openstack ubuntu server?
[01:37] <davidbowlby> I have several servers for different purposes and don't want to lose their configs or data.
[01:37] <lorenz> TJ-: Yay! But I never asked for it :D
[01:43] <lorenz> TJ-: initrd.img is (nearly) the same as on my local host
[01:44] <lorenz> TJ-: Nope, my diff call was not recursive.
[01:49] <lorenz> TJ-: The diffs are ok, I also tried to boot it with my local initrd. It booted the kernel, but didn't do more than the old initrd.
[01:49] <TJ-> lorenz: something very weird then, it must be disk image corruption somewhere
[01:50] <lorenz> TJ-: BTRFS -- So no disk corruption possible
[01:54] <lorenz> TJ-: Booting with host vmlinuz and initrd works, but doesn't make it any better
[02:04] <lorenz> TJ-: I got something! I removed all udev and plymouth upstart jobs and now it boots quite quickly to the point where it fails to mount /etc/fstab volumes. Then I started a maintaince shell, which surprisingly worked and now I got the log message that mountall failed (but after 19 seconds, not the usual 300)
[02:06] <lorenz> TJ-: That's also the first boot which logged into kern.log!!
[02:08] <TJ-> how about starting udev now, manually
[02:09] <lorenz> TJ-: Ok, done that. I'll try now to continue the boot process.
[02:10] <lorenz> TJ-: Mountall seems to fail now.
[02:12] <Sachiru> Anyone know how to configure memory limits for ZRAM on ubuntu? There's nothing in the official documentations and if there's anything on the forums about it 2 hours of googling apparently isn't enough.
[02:12] <lorenz> /etc/init/zram-config.conf
[02:12] <lorenz> probably there (I can't check that because I don't have ZRAM)
[02:15] <Sachiru> Nope, no such file, and if it did exist it's empty (not even a comment)
[02:41] <Gaba1> after I install ubuntu server from USB it will not find the HDD to boot. I try ubuntu 12 and that works fine. anyone have any ideas?
[03:18] <Gaba1> hello, I am having an issue where I create a USB boot drive and when the installation gets to GRUB there is an error and then when I go to complete the installation all I see is a little white blinking underscore
[03:19] <cfhowlett> Gaba1 verify your ubuntu ISO and the boot USB
[03:19] <cfhowlett> !md5sum
[03:25] <Tohsh> I have a clean install of server 14.04 on an intel i5 3rd gen based machine. I am unable to boot normally without the nomodeset flag set on the kernel line in grub. It will attempt to boot and then freeze when adding swap. This has never been an issue in the past on previous releases for me. I have verified my UUIDs match in fstab and blkid. Any idea?
[07:48] <RoyK> beamed-up-zombie-day!
[07:48] <sarnold> RoyK: do you have the day off?
[07:49] <RoyK> yep
[07:49] <sarnold> nice
[10:35] <Tazmain> hi all how do I upload to a ftp that uses a username and password ? And where is the rsync.conf file in ubuntu I cannot find the default one.
[10:37] <ikonia> you create the rsync.conf
[10:37] <bekks> Tazmain: Use a ftp client. And you can provide the credentials on the rsync commandline.
[10:38] <ikonia> and you upload to an ftp site that needs a username and password by puting in the username/password when prompted by the client
[10:40] <Tazmain> so do I set the location in rsync as the ftp client ? or ftp url ?
[10:41] <bekks> rsync has no gui, it is a command line utility.
[10:41] <Tazmain> that I know
[10:42] <Tazmain> you need to say in the config the location you are backing up to
[10:42] <bekks> You dont need any config, you need the command line only.
[10:43] <Tazmain> I want to automate it, doesn't it require a config then ?
[10:43] <bekks> No.
[10:43] <bekks> It requires a command line.
[10:45] <Tazmain> really ? I am on the rsnapshot howto page and it says 4.1 create the config file. But okay if i just require command line I can make a bash script. So in the command I will also have to put the username and password then ?
[10:47] <Tazmain> bekks, if you run rsync as a deamon doesn't that require a config. Just trying to clear my confusion
[10:54] <bekks> Tazmain: you dont run the client as a daemon.
[10:54] <bekks> Tazmain: you are connecting using the _client_ to the ftp site, which doesnt require a config but a command line only.
[10:58] <Tazmain> I meant rsync as a daemon
[11:03] <bekks> You dont use the rsync client as a daemon.
[11:03] <bekks> And you do not configure the server to connect anywhere, thats what the client does.
[11:04] <Tazmain> Oh I understand
[11:04] <bekks> I told you three times now ;)
[11:06] <Tazmain> bekks so I could go rsync backup /var/www/stuff ftp ftp.location.com/var ?
[11:12] <bekks> "/var/www/stuff ftp ftp.location.com/var" doesnt make much sense. you can use rsync to transfer data to an ftp location, yes.
[11:16] <Tazmain> bekks okay do I need to mount the ftp location or how would I do the command? Could you give me an example if that isn't too much of a bother
[11:21] <bekks> you dont need to mount the ftp location but you specify the credentials and the transfer protocol on command line. "man rsync" shows you how to do it, and you want to use the -e option.
[11:58] <Tazmain> bekks, thank you very much I will have a look at that otherwise I will use curlftpfs and just rsync to that. Only thing I still want to ask is if I want rsyn to run everyday do I need to make it a daemon or just a cron task
[12:02] <sarnold> cron task is probably fine
[12:02] <sarnold> daemon mode rsync is nice if you have many clients all reading (or writing) to the same places
[12:20] <Tazmain> sarnold, so would you recommend cron ?
[12:20] <sarnold> Tazmain: yes
[12:37] <bekks> Tazmain: the daemon is for creating an rsync server, nothing else.
[13:56] <foolhardy> I cannot seem to mount a "linux" partition: http://pastebin.com/DpJt5fx0
[13:56] <foolhardy> any advice
[13:56] <foolhardy> ?
[13:57] <sarnold> foolhardy: "Linux" is just a partition type, it doesn't tell you what is actually on that block device
[13:57] <foolhardy> how can I find that out?
[13:58] <sarnold> foolhardy: try file -s /dev/sdb1
[13:58] <bekks> sudo blkid
[13:58] <foolhardy> /dev/sdb1: sticky x86 boot sector, code offset 0x0
[13:59] <foolhardy> sudo blkid did not return anything regarding this new disk
[13:59] <foolhardy> only the original os disk
[14:00] <bekks> So there is no filesystem found.
[14:02] <sarnold> you could try leaving off the -t <foo> and see if mount can just figure it out
[14:03] <sarnold> if not, try ext2, ext3, exg4, btrfs, zfs, ntfs, ntfs-3g, vfat, etc..
[14:07] <bekks>  -t auto
[14:17] <moparisthebest> can I get do-release-upgrade to just download all needed packages and not change anything else?
[14:18] <moparisthebest> because I have to do the upgrade remotely over a slow connection, so I'd like to have it download everything while I don't have to watch it, then just apply things later
[14:23] <rbasak> moparisthebest: back up /etc/sources.list and /etc/sources.list.d/. Then replace all mentions of "saucy" with "trusty" (for example). Then run "apt-get update", then "apt-get --download-only dist-upgrade".
[14:23] <rbasak> Then change the files back. Correction: /etc/apt/sources.list and /etc/apt/sources.list.d/
[14:23] <rbasak> Then you'll have everything downloaded in /var/cache/apt/archives/, and a later upgrade will use those files instead of downloading against, except if there is anything new since you did that.
[14:24] <rbasak> instead of downloading again
[14:24] <moparisthebest> ah yea, guess that should work
[14:24] <moparisthebest> yea it should get the vast majority for me, thanks!
[14:58] <pakcjo> Hello, I think kerberos is starting before slapd and it should be the other way around, how can I confirm/fix this?
[14:59] <sarnold> pakcjo: check if they have old-style init.d scripts or upstart scripts, that'll decide how you fix it..
[15:00] <pakcjo> looks like upstart
[15:01] <pakcjo> sarnold: at least for slapd: http://pastie.org/9235435
[15:06] <moparisthebest> hehe ok rbasak: Calculating upgrade... Failed The following packages have unmet dependencies: *snip* E: Error, pkgProblemResolver::Resolve generated breaks, this may be caused by held packages.
[15:06] <moparisthebest> so is there a way around that?
[15:06] <rbasak> I don't think that should happen unless you've got third party packages installed.
[15:06] <rbasak> If that's right then you'll need to resolve those yourself.
[15:06] <moparisthebest> well, this is a lucid 10.04 box
[15:07] <moparisthebest> it worked for downloading precise 12.04 packages
[15:07] <moparisthebest> but failed with that when I tried trusty 14.04 :(
[15:08] <genii> Might want to ppa-purge and try again
[15:09] <moparisthebest> there is nothing in /etc/apt/sources.list.d/ never was
[15:11] <genii> moparisthebest: When you say "this is a lucid 10.04 box " ... "it worked for dowloading precise 12.04 packages" ... Do you mean you upgraded 10.04 to 12.04 or that you installed 12.04 packages onto a 10.04 system?
[15:11] <moparisthebest> well aptitude is getting the majority of them, that'll be good enough
[15:12] <moparisthebest> genii:  no I'm trying to pre-download most of the packages I'll need for do-release-upgrade
[15:12] <moparisthebest> I'll be upgrading to precise, then immediatly to trusty
[15:12] <moparisthebest> so I changed my sources.list to precise, apt-get update, apt-get -d dist-upgrade
[15:12] <rbasak> Oh, I see. I'm not aware of any way to get the trusty package downloads without upgrading to precise first.
[15:12] <moparisthebest> which worked, downloaded packages and did nothing else
[15:12] <moparisthebest> and then changed sources.list to trusty and did the same which broke a little
[15:12] <rbasak> I would strongly suggest redeploying trusty rather than upgrading all the way from lucid.
[15:13] <rbasak> Use configuration management, and test first.
[15:13] <moparisthebest> you think so?
[15:13] <moparisthebest> I actually have a server I originally had 6.X on that currently has 12.04
[15:14] <moparisthebest> I've never upgraded twice in a row like this, but hopefully it won't have any issues
[15:14] <rbasak> The general approach is much more reliable and reproducible. Know what you have deployed, rather than carrying issues forward.
[15:14] <moparisthebest> yea makes sense, what configuration management do you use?
[15:14] <rbasak> We do support upgrading, but the nature of server distributions is that they involve local customisations, and upgrade scripts cannot always carry that forward.
[15:14] <moparisthebest> all I've ever used is backups and such :)
[15:24] <davidbowlby> This sounds like you should walk through it a couple times.  I suggest booting up a virtual machine running 10.04 and attempt the upgrades a couple times before playing with a production system.
[15:24] <davidbowlby> you can hose a VM over and over ;)
[15:27] <moparisthebest> hmm, never saw the --sandbox option for do-release-upgrade before
[15:27] <moparisthebest> that should be pretty safe?
[15:28] <moparisthebest> -s, --sandbox         Test upgrade with a sandbox aufs overlay
[15:29] <jpds> moparisthebest: Safe as an upgrade gets.
[15:32] <jpds> moparisthebest: https://wiki.ubuntu.com/AufsBasedUpgrades
[15:33] <moparisthebest> ah thanks I was trying to find more info about it
[15:33] <davidbowlby> it wouldn't take you long to run through it in virtualbox
[15:34] <moparisthebest> well except I don't have another exact copy of this machine in a VM
[15:34] <davidbowlby> you don't need an exact copy to get the procedure down
[15:34] <davidbowlby> I'm talking about nailing down how you're going to do it
[15:34] <davidbowlby> the rest is just data
[15:35] <moparisthebest> I've upgraded 10.04 to 12.04 on multiple desktops/laptops/servers before, it was just around the time 12.04.1 was released, so, awhile ago :)
[15:35] <davidbowlby> not saying there can't be issues with packages you have installed
[15:35] <davidbowlby> but you can do a dpkg list to find out what's on it
[15:35] <davidbowlby> but you're not talking 10.04 to 12.04
[15:35] <davidbowlby> you're talking 10.04 to 12.04 to 14.04
[15:35] <moparisthebest> I've also upgraded from 12.04 to 14.04
[15:36] <davidbowlby> but you're asking about the double upgrade
[15:36] <moparisthebest> the same computers that *used* to run 10.04, and 8.04 before that
[15:36] <bekks> davidbowlby: 10.04 to 12.04 to 14.04 includes 10.04 to 12.04.
[15:36] <davidbowlby> personally I'd want to walk through it on something I don't care about first
[15:36] <moparisthebest> I'd think it'd be about the same
[15:36] <davidbowlby> thinking and knowing are two different things
[15:36] <moparisthebest> yea the only thing that's giving me pause is I have to drive 10 miles if it won't reboot or something hehe
[15:37] <davidbowlby> exactly, so why not be sure about it
[15:37] <davidbowlby> create a 10.04 vm
[15:37] <davidbowlby> with the same packages installed
[15:37] <davidbowlby> (you can get a list from dpkg)
[15:37] <davidbowlby> and then try it
[15:37] <davidbowlby> if it goes well, you have a procedure
[15:37] <davidbowlby> if it breaks, you didn't break a server 10 miles away
[15:38] <davidbowlby> 10 miles lol, try 1600 miles away
[15:38] <bekks> moparisthebest: Just use do-release-upgrade -s
[15:38] <moparisthebest> yea it's a good idea, just not sure if I want to try it or just screw it and roll the dice, like you said, it's not like it's 1600 miles away or more like the rest of my servers
[15:39] <davidbowlby> I'd do the VM and document a procedure in confluence for future use
[15:39] <davidbowlby> but that's me :D
[15:39] <moparisthebest> well and this is my last computer on 10.04 so I don't ever need to do it agai
[15:39] <davidbowlby> ah
[15:39] <davidbowlby> but you can always share it online for other suckers in your position
[15:40] <moparisthebest> hmm yea
[15:40] <davidbowlby> #contribute
[15:40] <davidbowlby> lol
[15:54] <MTughan> I'm trying to provision an Ubuntu 14.04 image on an OpenStack installation and am switching to using WaitConditions. I get a URL from the wait condition and install the heat-cfntools package (to get cfn-signal), but running cfn-signal returns a Python import error: "No module named heat_cfntools.cfntools". I can't find the module anywhere on the system.
[15:54] <MTughan> Am I just missing a package that needs to be installed? Is something missing from the package? Something else entirely going on?
[16:08] <fridaynext> anybody here use phpvirtualbox on ubuntu server? any comments on performance?
[16:08] <fridaynext> i'm currently remoting into KVM from a VM on my OSX laptop, and that's too kludgey for me. I want to simplify.
[16:09] <fridaynext> I already have OpenVPN installed, so it seems phpvirtualbox would be a perfect companion to access my media server's VMs from anywhere.
[16:24] <davidbowlby> MTughan, did you install heat-api-cfn  ?
[16:25] <MTughan> davidbowlby: No, because I think that's the server component.
[16:25] <davidbowlby> I think heat-engine is
[16:25] <MTughan> Looking at its dependencies, you may be right.
[16:26] <davidbowlby> https://packages.debian.org/hu/sid/web/heat-api-cfn
[16:26] <MTughan> Yeah, the name suggests that it's the API endpoint, but maybe it's not.
[16:26] <davidbowlby> I'd say based on your error message, it's a good thing to try ;)
[16:27] <MTughan> It's weird though, looking at the OpenStack git repo for heat_cfntools, the module's in that repo. http://git.openstack.org/cgit/openstack/heat-cfntools/tree/
[16:27] <davidbowlby> This package contains the CloudFormation (CFN) API.
[16:27] <davidbowlby> I assume you installed heat-api
[16:27] <davidbowlby> that error is specific to cfn though, so I think apt-get install heat-api-cfn will fix it
[16:28] <MTughan> If the packages were available, perhaps.
[16:28] <MTughan> E: Failed to fetch http://nova.clouds.archive.ubuntu.com/ubuntu/pool/main/h/heat/python-heat_2014.1~rc2-0ubuntu3_all.deb  404  Not Found [IP: 91.189.88.153 80]
[16:28] <davidbowlby> hmm interesting
[16:28] <davidbowlby> sudo apt-get update
[16:28] <MTughan> I'm trying an update now.
[16:28] <davidbowlby> ;)
[16:28] <davidbowlby> gmta
[16:29] <MTughan> Normally, I have cloud-config do an update on boot. But I think that was skipped this time because of an issue I had while testing.
[16:29] <MTughan> Yeah, now it's working.
[16:30] <davidbowlby> sweet, I just started playing with OpenStack last week
[16:31] <davidbowlby> got tired of esxi hardware support continuing to dwindle
[16:33] <davidbowlby> migrated my mysql server last night using rsync, that was nerve racking, but went well
[16:33] <davidbowlby> luckily I was smart and used DNS in my environment lol
[16:36] <MTughan> davidbowlby: Neither heat-api nor heat-api-cfn fixed it.
[16:37] <MTughan> I had thought because I didn't do an update that heat-cfntools might not've been up-to-date, but upgrade didn't touch it or fix it.
[17:00] <lordievader> Good evening.
[18:03] <pakcjo> which is the init.d script for dhcpd?
[18:05] <PryMar56>  dhcp3-server
[18:07] <pmatulis> pakcjo: if you want something lighter and progressive you may consider dnsmasq.  it doesn't have all features though
[18:09] <pakcjo> PryMar56: I don't have that (Ubuntu 12.04.4 LTS)
[18:10] <pakcjo> I have isc-dhcp-server and isc-dhcp-server6
[18:10] <PryMar56> pakcjo, I checked lucid
[18:11] <pakcjo> I'm asking because it seems that there was a change, /etc/dhcp instead of /etc/dhcp3/
[18:12] <pakcjo> is there a better tool than update-rc.d to see which init.d scripts are enable?
[18:12] <PryMar56> pakcjo, ls -al /etc/rcS.d  ; ls -al /etc/rc2.d/
[18:13] <PryMar56> if you see symlinks to /etc/init.d/ the service is setup
[18:14] <pakcjo> hmm interesting, I have a S07dhcp3-server but it is not in /etc/init.d/
[18:15] <pakcjo> so, is this a new upgrade bug?
[18:15] <PryMar56> pakcjo, that is a dead symlink
[18:15] <PryMar56> its harmless
[18:16] <pakcjo> PryMar56: yes, but I don't have anything to start dhcp with...
[18:17] <PryMar56> pakcjo, run update-rc.d against the new isc-dhcp...
[18:18] <pakcjo> PryMar56: which? isc-dhcp-server or isc-dhcp-server6? why I have 2?
[18:19] <PryMar56> do you want to hand out ipv6 addresses? I don't
[18:19] <pmatulis> pakcjo: there may be an upstart job
[18:20] <pmatulis> /etc/init
[18:20] <pakcjo> oh 6 is for ipv6, thanks ;)
[18:20] <pakcjo> pmatulis: oh yes, there is :)
[18:23] <timhansen> good afternoon
[18:23] <pakcjo> hello timhansen
[18:25] <timhansen> i just created a new 14.04 server on linode. installed lamp, and setup a virtualhost. added the entry into my local hosts file, but when i go to the fake domain, i’m getting the index view of the html directory (i removed the default ubuntu index.html)
[18:25] <timhansen> here’s the output of my virtualhost: https://gist.github.com/willc0de4food/cc014d072cf42eb538dc
[18:26] <timhansen> any ideas as to what i did wrong?
[18:30] <pakcjo> is there a way to set order of init.d scripts?
[18:35] <pakcjo> if I manually change the symbolic lick from S06 to S05 will that work?
[18:35] <pakcjo> (is that the correct way of doing it?)
[18:38] <pmatulis> pakcjo: is this kerberos & slapd again? normally those links should not be changed
[18:38] <pakcjo> pmatulis: yes
[18:39] <pmatulis> timhansen: remove 'Indexes'?
[18:39] <pakcjo> I have to start them manually after a reboot... I was thinking of adding them to /etc/rc.local but there should be a better way...
[18:40] <timhansen> i figured it out. i’m absent-minded & didn’t name the vhost with .conf -.-
[18:44] <pmatulis> pakcjo: doesn't kerberos only need access to slapd when it receives a request?  why would it depend on it to start up?  are you also storing kerberos/kdc config information in slapd?
[18:45] <pakcjo> pmatulis: yes, ldap is used as db for kerberos
[18:45] <pmatulis> pakcjo: so what is the current problem?
[18:45] <pakcjo> pmatulis: kerberos tries to start, fails with couldn't connect to ldap server, slapd starts
[18:46] <pakcjo> i need slapd to start before kerberos
[18:46] <pmatulis> pakcjo: i wonder why?  it must do a check when it starts
[18:46] <pakcjo> a check?
[18:47] <pmatulis> pakcjo: yeah, it obviously does a check of some sort
[18:47] <pmatulis> pakcjo: would be great if you could pastebin some logs from kerberos
[18:48] <pmatulis> pakcjo: but it's weird that such a problem has not already been identified
[18:48] <pakcjo> there isn't much: kadmind[1333](Error): Can't contact LDAP server while initializing, aborting
[18:49] <pakcjo> https://bugs.launchpad.net/ubuntu/+source/krb5/+bug/652433
[18:49] <pmatulis> ah
[18:49] <pakcjo> from 2010 XD XD XD
[18:57] <pmatulis> pakcjo: that bug explains things fairly well.  were you not able to get a workaround out of it besides manually changing the symlinks?
[19:00] <pakcjo> well the bug report suggest to change order
[19:03] <pakcjo> and rc.local XD
[19:03] <pakcjo> another reason to get rid of slapd
[19:17] <sixBB> hello.  I am exeperiencing problems with installing/upgradiing udev.  Thie configure process hangs.  I've googled and not tound anything pearticularly helpful
[19:17] <sixBB> any suggestions?
[20:26] <pakcjo> what is the correct way to start/stop services in /etc/init/?
[20:29] <Patrickdk> service xxxx (start|stop)
[20:30] <pakcjo> thasnk Patrickdk
[20:31] <pakcjo> so start xxxx and stop xxxx is for something else?
[20:31] <pakcjo> or they are aliases?
[20:31] <Patrickdk> dunno
[20:31] <lordievader> pakcjo: There are many ways, using initctl is another way ;)
[20:32] <Patrickdk> hmm, start/stop is aliased to initctl
[20:32] <lordievader> Ah, that is new to me. :)
[20:33] <pakcjo> thanks
[20:33] <MTughan> pakcjo: service using start and stop in the script for start and stop.
[20:33] <MTughan> And status and reload for those two as well, apparently.
[20:34] <MTughan> All of which are aliased to initctl by softlinks.
[20:35] <MTughan> "restart" is just accomplished by calling stop followed immediately by start.
[20:35] <Patrickdk> ah, service is just a script, that runs init.d or initctl for init
[20:35] <MTughan> Exactly.
[20:35] <MTughan> Shell script, even. Thought it might be Python, but I guess that'd be overkill.
[20:36] <Patrickdk> since upstart, I switched to using service for everything
[20:36] <MTughan> (sorry, I've been doing a lot of Python lately)
[20:36] <Patrickdk> I've been doing lots of lua scripts lately :)
[20:36] <Patrickdk> very low latency :)
[20:36] <MTughan> Yeah, service should be your goto for service-y stuff (no pun intended).
[20:47] <pakcjo> nested db in ldap is a bad idea right? is there any documentation about it?
[20:48] <pmatulis> pakcjo: it makes things unecessarily complex
[20:49] <pakcjo> pmatulis: do you have any link about where I can read about it? or a guide to switch/port to hmm "lineal?" model?
[20:49] <pakcjo> flat model
[20:50] <pmatulis> pakcjo: no, but IMO it should just be a matter of changing the slapd config so it points to the new location
[20:50] <pmatulis> pakcjo: i idle in #openldap , you might get better opinion there
[20:50] <pakcjo> pmatulis: it's nested vs flat or is there a better terminology for not nested?
[20:50] <pakcjo> thanks
[20:51] <pmatulis> pakcjo: i don't think 'nested' is a great term
[20:51] <pakcjo> :)
[20:51] <pmatulis> pakcjo: it usually implies one thing working within another.  but in this case, it's just a matter of directories.  a location and not something in the logical/working aspect
[20:52] <pakcjo> pmatulis: thanks
[20:52] <pmatulis> ex: nested kvm.  a hypervisor which is actually a guest of another hypervisor
[20:56] <pmatulis> pakcjo: i know i have at least one cleanup/re-install shell script that would wreak havoc if i had a database located within the directory of another database ;)
[20:57] <pakcjo> :)
[21:14] <__dan__> hi guys, hope you're all well, been having consistent hangs with btrfs defrag on ubuntu 14.04 stock kernel, was wondering if there is a guide to updating the kernel to latest and also updating the btrfs supporting programs? any guidance appreciated, cheers :)
[21:20] <Patrickdk> don't use btrfs till it's stable?
[21:25] <__dan__> when will that piece of string reach it's end?
[21:27] <__dan__> also, I'm not using the more cutting edge stuff like dedupe or parity raid, I'm comparing between ZFS on Linux and btrfs on the same hardware and would like to evaluate btrfs latest, preferably in a safe non-hacky manner I was hoping to get some help with that
[21:27] <Patrickdk> dunno, watch btrfs.wiki.kernel.org
[21:27] <__dan__> uh huh
[21:27] <Patrickdk> they don't say btrfs is stable yet, at all
[21:27] <Patrickdk> only that the ondisk format is stable, not that the code/runtime is stable
[21:28] <__dan__> well, I believe the man page etc was changed last year to reflect it's relative stability from where it was when that man page was originally written
[21:28] <Patrickdk> well, consider the wiki was updated less than a month ago
[21:28] <Patrickdk> and says NOT STABLE
[21:28] <Patrickdk> I would assume it's not stable
[21:29] <__dan__> the situation is a little more complicated than you seem to be giving credit
[21:29] <rberg_> __dan__:  check here for builds of upstream kernel releases.. http://kernel.ubuntu.com/~kernel-ppa/mainline/
[21:29] <Patrickdk> how is it more complicated?
[21:29] <__dan__> because some parts of the codebase are more mature than others
[21:30] <__dan__> hi rberg_ thank you I will take a look
[21:30] <rberg_> as for userland you may need to build it yourself.
[21:31]  * Patrickdk just hopes he can use btrfs, without dedup/compression/snapshots/...
[21:31] <Patrickdk> just to store files
[21:31] <Patrickdk> without it exploding and becoming completely corrupted
[21:31] <__dan__> ah yes I saw this earlier, I'm running trusty and would like to keep it that way, is it OK to run the utopic kernels? and is there a way to set it up to update automatically? (ie. an apt repo that has latest 3.14 for instance?)
[21:31] <Patrickdk> had it do that way too many times
[21:32] <rberg_> heh the way btrfs and zfs do snapshots is REALLY nice..
[21:32]  * __dan__ nods :) I'm a FreeBSD refugee and used to running ZFS but ZFS on Linux is highly not recommended for 32-bit machines - guess what I'm running :/
[21:33] <__dan__> to be fair it's not the best idea on 32-bit FreeBSD either but I ran it with no issues whatsoever after the initial tune-up period
[21:33] <Patrickdk> there are still 32bit systems?
[21:33]  * __dan__ rolls eyes
[21:34] <Patrickdk> thought they died out with pentium D cpu's
[21:40] <metho> anyone successfuly migrated bind to windoz server?
[21:48] <__dan__> metho: I'm no DNS expert but can't you just configure windows as secondary and have it pull the zonefiles from bind primary?
[21:49] <__dan__> can't say I know anyone crazy enough to migrate DNS to windows though sorry :) i dare to even ask your use case
[21:51] <qman__> Migrating a DNS zone to MS DNS isn't hard, but it's not an ubuntu question
[21:52] <__dan__> maybe on some level he came here so we could talk him out of it :)
[21:54] <qman__> As for migrating _from_ MS DNS _to_ bind, making bind the secondary, transferring the zone, then reconfiguring as a master is a good method
[23:05] <Joe_knock> Hello there
[23:06] <Joe_knock> Whenever I SSH into my remote server, I get a small introduction about the system, how do I make this introduction appear whenever I want?
[23:16] <lutostag> Joe_knock: I believe you are talking about motd (message of the day), which is created by the scripts that live in /etc/update-motd.d/
[23:16] <Joe_knock> lutostag: is there a simple command to make it display? Or were u telling me by saying motd?
[23:20] <Jordan_U> Joe_knock: cat /etc/motd
[23:22] <Joe_knock> Jordan_U: Thank you
[23:23] <Jordan_U> Joe_knock: You're welcome.