| === markthomas is now known as markthomas|away | ||
| === trabbit_ is now known as trabbit | ||
| === G_ is now known as G | ||
| robert45 | thanks | 03:27 |
|---|---|---|
| robert45 | hi guys, my software RAID6 failed and Im unable to start it, can someone help me how to fix this? http://pastie.org/10406429 | 03:27 |
| sarnold | what "possibly out of date" mean? that sounds ominous | 03:30 |
| robert45 | sarnold thanks. well I dont have a clue what does that mean | 03:31 |
| sarnold | robert45: have you seen this yet? https://raid.wiki.kernel.org/index.php/RAID_Recovery | 03:34 |
| robert45 | sarnold In fact, yes. I tried the Event fix but didnt work, Im scary to try the recreate process since it may destroy the data | 03:35 |
| arooni | do you folks typically run a repository for .dotfile configurations like .vimrc .tmux.conf that you'd like to keep up to date on all the machines you use them on? | 04:36 |
| === kickinz1\afk is now known as kickinz1 | ||
| === Lcawte|Away is now known as Lcawte | ||
| === CiPi is now known as cipi | ||
| lordievader | Good morning | 07:26 |
| jamespage | zul, guh - I keep hitting this - https://bugs.launchpad.net/ubuntu/+source/lxd/+bug/1493797 | 10:47 |
| ubottu | Launchpad bug 1493797 in lxd (Ubuntu) "lxd unix socket not accessible reliably" [Undecided,New] | 10:47 |
| shoaib | hi | 11:09 |
| Beliq | My hosting provider recently changed their VPS service, double the performance and significantly cheaper. However, they still charge me for their old service. Didn't automatically scale me up nor offered transition. I am right to be disappointed ? | 11:17 |
| shoaib | what is the use of having server computers at home? | 11:18 |
| === MrPPS is now known as Guest80338 | ||
| zul | jamespage: i think you are hitting this https://github.com/lxc/lxd/commit/593b4c0e50e87189fbf8e658855d9fdad870eb50 | 11:25 |
| jamespage | zul, yeah - I saw that and I think you are correct - I've tried a deploy from git one and it appears OK | 11:25 |
| jamespage | maybe | 11:25 |
| zul | ack | 11:26 |
| jelly | shoaib: servers provide services. Sometimes you want to have some services available at all times, even in a home environment. For example, a file server to access media from various devices. | 11:26 |
| Teduardo | has anybody been able to get sendmail to work after the logjam fix? | 12:28 |
| jpds | Teduardo: https://twitter.com/sadserver/status/630745530100899840 | 12:30 |
| Teduardo | so the answer to my question i guess you are saying is no? | 12:35 |
| === dannf is now known as dannf2 | ||
| Aethenelle | anyone install a CAS server on ubuntu successfully? | 14:09 |
| Aethenelle | more precisely, when I run tomcat under systrace as root or tomcat8, the jaas config is read fine. when by systemd, it fails. I can't seem to get java.security.debug to work either. | 14:11 |
| Aethenelle | any ideas? | 14:11 |
| notafads | looking for help getting a raid 10 install up and running | 14:40 |
| notafads | can tip im also willing to pay if completed dont know if that changes anything | 14:41 |
| Pici | notafads: I'd start by looking at https://help.ubuntu.com/lts/serverguide/advanced-installation.html | 14:46 |
| notafads | well i got hardware raid available via http://www.newegg.com/Product/Product.aspx?Item=N82E16813182959 and no concept how to access the intel storage manager | 14:47 |
| pmatulis | notafads: what's the storage manager? anyway, use hardware raid or software raid, not onboard (fakeraid) | 15:03 |
| patdk-wk | heh? onboard fakeraid == software raid | 15:07 |
| notafads | ? | 15:09 |
| notafads | pmatulis, honestly i do not know i am having trouble locating it hence the issues | 15:25 |
| pmatulis | patdk-wk: you say that b/c mdadm now can interface with onboard raid? | 15:35 |
| patdk-wk | heh? | 15:36 |
| patdk-wk | there is no onboard raid | 15:36 |
| patdk-wk | dmraid is the software implementation that uses mdraid parts to do the work | 15:36 |
| patdk-wk | the only part that is onboard raid, is a simple helper to let the bios boot | 15:37 |
| pmatulis | onboard raid, i meant onboard raid controller | 15:39 |
| patdk-wk | I origionally used it, back on 10.04 | 15:44 |
| patdk-wk | that lasted a few months, dmraid just annoyed me to no end, due to it's write buffering | 15:45 |
| patdk-wk | and changed to mdraid | 15:45 |
| pmatulis | patdk-wk: why did you say "onboard fakeraid == software raid" ? | 15:49 |
| patdk-wk | cause there is no raid support with fakeraid | 15:50 |
| patdk-wk | it DEPENDS on and uses software raid | 15:50 |
| patdk-wk | fakeriad is raid boot support only, not raid at all | 15:50 |
| patdk-wk | just a boot helper | 15:50 |
| pmatulis | ok, by 'software raid' i meant 'mdraid (mdadm)' | 15:50 |
| notafads | the motherboard i linked | 15:50 |
| patdk-wk | mdraid is software raid, but your left to having the bios boot a usable disk yourself | 15:50 |
| notafads | does do hardware raid? | 15:50 |
| pmatulis | patdk-wk: right, so we agree ;) | 15:51 |
| patdk-wk | nothing is worse than a disk the motherboard thinks it can boot, and doesn't work :) | 15:51 |
| pmatulis | notafads: so i reiterate, use mdraid (madam) or a real hardware raid controller | 15:52 |
| notafads | pmatulis, with ubuntu install it wont do raid 1+0 | 15:53 |
| notafads | natively | 15:53 |
| pmatulis | notafads: your fakeraid/motherboard won't give you the raid you want? | 15:54 |
| notafads | pmatulis, i cant seem to even figure it out on the mobo i cant find smart storage in the bios only have the option to turn raid feature on for esatas and satas | 15:54 |
| notafads | sorry i am quiet new to this kinda everything fell into my lap | 15:55 |
| pmatulis | notafads: just another reason to disable it then | 15:55 |
| notafads | generally i am use to seeing a ctrl+S for storage options | 15:55 |
| notafads | on boot | 15:55 |
| pmatulis | notafads: probably for hardware raid | 15:55 |
| notafads | but that option is not there | 15:55 |
| pmatulis | notafads: that's b/c you don't have h/w raid! | 15:56 |
| notafads | http://www.newegg.com/Product/Product.aspx?Item=N82E16813182959 b-but it says sata raid 0/1/5/10 | 15:56 |
| notafads | ;_; ugh | 15:56 |
| sarnold | most of those "sata raid" things are thin shims toget a boot loader running | 15:57 |
| sarnold | it's best to not trust them much | 15:57 |
| sarnold | oh, heh, Isee pmatulis and patdk-wk have you well covered :) | 15:57 |
| quantic | notafads: there's a difference between software/BIOS RAID (aka FakeRAID) and an actual hardware RAID controller. | 15:57 |
| notafads | quantic, i am aware of that much but thank you | 15:57 |
| quantic | notafads: then you're aware of the problems inherent in FakeRAID. Cool, I won't say anything more. | 15:58 |
| patdk-wk | you do not want to use motherboard supported raid, unless there is a VERY VERY specific reason why, unless it's a real onboard raid card, like an lsi | 15:58 |
| patdk-wk | and even then, those normally don't have options for a battery, or write cache, so are painfully slow in raid mode also | 15:59 |
| notafads | quantic, no i know there is a difference not that i fully understand the issues | 15:59 |
| quantic | notafads: ahh, gotcha. Misunderstood. | 15:59 |
| notafads | quantic, sorry if my tone came off frustrated it is because i am just not at you guys just the issue in general. | 16:00 |
| quantic | notafads: FakeRAID solutions generally just offload the RAID calculations to the CPU, and don't have any "real" RAID features, like write-caching, battery backup, etc. | 16:00 |
| notafads | im trying to take advantage of an opportunity dropped in my lap and all i have done is continually hit walls. | 16:00 |
| quantic | notafads: They're also somewhat proprietary. You won't be able to start the RAID set in any motherboard other than the exact model you're working with, for the most part. | 16:00 |
| quantic | notafads: What are you trying to do? | 16:01 |
| notafads | oh ok | 16:01 |
| notafads | build a raid 10 in ubuntu install | 16:01 |
| quantic | notafads: Oh, so exactly what I do. | 16:01 |
| notafads | lol | 16:01 |
| quantic | notafads: Don't use the motherboard RAID. Disable it. Use mdadm, works great. | 16:01 |
| notafads | thats software? | 16:01 |
| quantic | notafads: How many drives and how big? | 16:01 |
| notafads | 4x 3tb | 16:01 |
| notafads | hot swappable | 16:01 |
| notafads | so i can quickly replace as needed and have a led indicator when i have issues | 16:02 |
| quantic | notafads: Big, fat drives. Okay, you'll have a few gotchas. 1. You MUST use GPT. Disks are too big for MBR partitioning. | 16:02 |
| quantic | notafads: Create the following partitions in the installer: a. 1MB - reserved BIOS boot partition. 2. 1G - RAID partition. 3. The rest of the disk, RAID partition. | 16:02 |
| quantic | notafads: On each disk. | 16:02 |
| patdk-wk | if you want the led's to work, you will want hardware raid | 16:03 |
| notafads | leds are essential so its dummy proof for the most part on my behalf | 16:03 |
| patdk-wk | you can make led's work with software raid, but it is a lot of work, and different on every system | 16:04 |
| notafads | quantic, your saying software raid? | 16:04 |
| quantic | notafads: Yup. | 16:04 |
| notafads | patdk-wk, i will have to go with a hardware raid controller i need those leds | 16:04 |
| quantic | notafads: If you want hardware raid, buy a raid controller. If you want to use what you have and make a functional system, use mdadm. | 16:04 |
| notafads | quantic, ^ | 16:04 |
| quantic | notafads: With basic configuration, no, you don't. Mine emails me if a drive fails. | 16:05 |
| quantic | but hey, its your system. | 16:05 |
| notafads | quantic, thats actually pretty cool | 16:05 |
| patdk-wk | depends on the usage | 16:05 |
| patdk-wk | emailing is needed either way :) | 16:06 |
| patdk-wk | but not having someone pull the wrong disk when actually at the rack, is helpful | 16:06 |
| notafads | patdk-wk, thats what i am afraid of | 16:06 |
| === markthomas|away is now known as markthomas | ||
| notafads | im kinda being forced into being a system admin on this server and i got someone to help me set it up and everything however i do need to get around this hurdle first | 16:07 |
| pmatulis | notafads: get a real raid controller then | 16:07 |
| patdk-wk | now for the next question :) | 16:07 |
| patdk-wk | what kind of performance do you need? | 16:07 |
| pmatulis | heh heh | 16:08 |
| patdk-wk | will you need a real raid card, with bbwc, or just a hba that *kindof* does raid | 16:08 |
| patdk-wk | bbwc makes writes very fast :) | 16:09 |
| patdk-wk | but your talking like $250 vs $750 | 16:09 |
| notafads | thats way to much | 16:10 |
| patdk-wk | both? | 16:11 |
| patdk-wk | really though, software raid is going be better than the lower cost option | 16:12 |
| patdk-wk | but led's will have to be done manually, if it is even possible | 16:12 |
| patdk-wk | possible depends on the hardware | 16:13 |
| patdk-wk | normally the best idea for that, is just to pound the disks, and remove the one that isn't blinking :) | 16:13 |
| sarnold | haha :) | 16:14 |
| patdk-wk | it works on kids too, remove the one that doesn't yell | 16:15 |
| sarnold | hahahaha | 16:16 |
| pmatulis | patdk-wk: i would prolly try to take a pulse but you're business ;) | 16:29 |
| pmatulis | *your | 16:50 |
| === suresh is now known as Guest49556 | ||
| blizzow | I have a server with 256GB RAM, I'd like to set aside 32GB of RAM as a ramdisk mounted at /myramdisk on every boot. How would I do that in trusty server or vivid server? | 17:56 |
| tarpman | blizzow: a line in /etc/fstab of type tmpfs with a size option | 17:59 |
| jrwren | blizzow: you CAN do that, but likely you don't need to. linux bufcache is REALLY good. | 17:59 |
| jrwren | blizzow: anyway, we call it tmpfs. files in it may get swapped out. | 18:00 |
| blizzow | Here's the kicker, I have no swapfile. | 18:06 |
| jrwren | blizzow: should be fine still. | 18:09 |
| === Lcawte is now known as Lcawte|Away | ||
| RoyK | blizzow: really, use swap, it'll help you use physical memory better | 18:11 |
| RoyK | blizzow: not sure how much with a quarter of a terabyte of RAM, but it all depends on the load | 18:11 |
| blizzow | RoyK: The server hosts a bunch of VMs. The out of memory killer kills a VM once in a while, and I believe it's because there is no swap. The problem is, I have to be absolutely sure the VM processes do NOT hit our swapfile. | 18:13 |
| RoyK | blizzow: the linux memory manager is very good at spotting pages not in use and swapping those out - better allocate a truckload of swap and you'll se it's using a small amount of it and leaving the left RAM usable for something useful | 18:16 |
| RoyK | blizzow: linux swapped out whole processes sometime in the mid ninetees - it doesn't do that anymore | 18:17 |
| RoyK | blizzow: why don't you want pages of the VMs swapped out, btw? | 18:18 |
| blizzow | Because it's super hard to tell if a VM is having performance issues because the hypervisor has swapped out part of the VM process, or there is a problem in the underlying VM. | 18:19 |
| jrwren | blizzow: you can get dynamic swap instead of using swap partition by installing swapspace package. | 18:19 |
| RoyK | blizzow: let linux do the job, please. you can adjust vm.swappiness (how rapidly swap is used) with sysctl | 18:19 |
| RoyK | blizzow: default is 60 (out of 100). set it to 1 if you're very nervous. probably the default is good. it's always better for a VM to be slow than to be dead | 18:20 |
| jrwren | there are good tools which will tell you swapped out pages for a process, if you need. | 18:20 |
| RoyK | jrwren: you rarely need those | 18:21 |
| RoyK | people are too afraid of swapping - it's not like win95 swaps :P | 18:21 |
| jrwren | RoyK: maybe when you want to tell if a VM is having performance issues because its swapped out. | 18:22 |
| RoyK | jrwren: then you'll see that on the i/o numbers on the host - it's that simple | 18:22 |
| RoyK | jrwren: and a VM is *never* swapped out - just the pages it didn't use much | 18:23 |
| RoyK | jrwren: if you have too little memory - get more - don't stop swap - it's a lifesaver | 18:23 |
| jrwren | RoyK: how can I tell that a VM is having performance issues because pages of it are swapped? | 18:24 |
| ciscam | Using virtualbox 4.3 two of my ubuntu servers are behaving differently. one is a default installation via .iso and the other is an openvpn-as prepared .vhd. Now when I suspend both of them, on resume the default ubuntu server (15.04) has the correct time in a few seconds and the openvpn-as prepped server (14.04.1 LTS) continues at the time when it was suspended, until I 'ntpdate time.nist.gov' it | 18:24 |
| RoyK | jrwren: you'll see it by the swap usage on the host | 18:24 |
| blizzow | if one VM is busy hitting I/O and another VM has chunks that are swapped. It's camouflaged. | 18:25 |
| jrwren | RoyK: so you are syaing don't swap? | 18:25 |
| jrwren | RoyK: so why have a pagefile? | 18:25 |
| RoyK | jrwren: please - it's been 24 years of development on linux and everyone in the kernel developer gang says "use swap!" | 18:25 |
| RoyK | jrwren: you may want to try without it, but you may as well trust the experts | 18:25 |
| RoyK | jrwren: I'm saying swap is nice, because it sorts out the bits of allocated memory not in use are swapped out for the rest of the memory to be used efficiently | 18:26 |
| RoyK | jrwren: just trust those nerds, please | 18:26 |
| ciscam | What I found out is that ntp is not installed on the default ubuntu server and the command hwclock returns that it's not accessible via known means. Does somebody know how it synchronizes the time anyways? The openvpn-as servers ntp is configured with 'tinker panic 0' and hwclock returns the time when it was suspended, identical to date. the time only syncs with a manual ntpdate. | 18:27 |
| RoyK | ciscam: why virtualbox? | 18:28 |
| jrwren | RoyK: i'm only asking questions. I can either keep asking becuase I don't understand you, or I can stop. What do you suggest? | 18:28 |
| ciscam | it's available as a preconfigured jail for freenas | 18:28 |
| ciscam | neither of those servers has the guestadditions installed | 18:29 |
| jrwren | blizzow: I'm curious what solution you use. Let me know, please? | 18:33 |
| RoyK | jrwren: just use swap, ok? | 18:37 |
| jrwren | RoyK: *sigh* | 18:37 |
| jrwren | RoyK: it depends. ;] | 18:37 |
| RoyK | no, really, it doesn't | 18:37 |
| RoyK | setup swap and set swappiness = 1, and it'll only use swap when it's critical | 18:37 |
| jrwren | RoyK: i'm ok with no swap on some small VMs and certain other applications. | 18:38 |
| jrwren | RoyK: i'm ok with no swap on rpi | 18:38 |
| RoyK | well, justr trying to help, I've been using Linux for >20 years, you're on your own | 18:39 |
| jrwren | RoyK: i'm almost always leary of absolute prescriptions. | 18:39 |
| jrwren | RoyK: >20? wow! | 18:39 |
| jrwren | RoyK: Any advice on monitoring an applications pages being swapped out? | 18:40 |
| blizzow | jrwren: I tried to mount -t tmpfs -o size=2048M tmpfs /swapspace Then I made a swapfile in there and when I tried to swapon -a /swapspace/swapfile, I get "swapon: /swapspace/swapfile: swapon failed: Invalid argument" | 18:40 |
| jrwren | blizzow: a swap file on tmpfs? WTF are you doing? don't do that. | 18:41 |
| blizzow | I wanted to use a ramdisk for swap. :( | 18:41 |
| jrwren | blizzow: it literally makes zero sense. :) | 18:41 |
| tarpman | o_O | 18:42 |
| tarpman | blizzow: what's the difference between a swap file backed by RAM, and no swap at all? | 18:42 |
| tarpman | blizzow: suggestion: apt-get install zram-config | 18:42 |
| RoyK | blizzow: that's nice, if you're low on RAM, send the data through several layers of complexity to RAM :D | 18:42 |
| blizzow | I'm trying to avoid VM processes from EVER touching disk based swap. This is so I NEVER have to worry about a swap as a variable for VM performance. A problem arises in the hypervisor activates an OOM killer and kills VMs when there is no swap given to the hypervisor. | 18:45 |
| sarnold | blizzow: sounds like what you really want is to make sure the sum of all memory on all VMs on the system is ~two gigabytes less than the amount of RAM installed in the machine | 18:52 |
| blizzow | I'm trying to eliminate hypervisor swap access as a variable for VM performance. That's all. | 18:52 |
| blizzow | sarnold: That's part of the problem. Having no swap seems to activate an out of memory killer even when the sum of all VMs on the system uses 50GB less RAM than is available to the system. | 18:54 |
| blizzow | I'd be fine without the ability to overcommit RAM to virtual machines, but only being able to commit <60% of RAM to VMs seems wrong. | 18:56 |
| jrwren | blizzow: you can set swapiness per cgroup. start each VM in a cgroup with swapiness 1 | 18:56 |
| sarnold | blizzow: that's gotta be wrong somewhere. please file a bug. | 18:59 |
| sarnold | blizzow: with that kind of memory I wonder about numa memory placement but .. even that shouldn't invoke the OOM killer. | 19:00 |
| blizzow | sarnold: When the OOM killer gets invoked, in NUMA node 0, the largest page the kernel has available is 8kb. | 19:02 |
| sarnold | blizzow: ouch. :/ | 19:02 |
| blizzow | cgroups/cgmanager documentation is worse than trying to read stereo instructions from Ikea. How do I ensure that any qemu-kvm processes have swappiness set to 0 or 1? | 19:26 |
| Doyle | Do values set in /etc/default get loaded on service reload, or just service start/restart? | 19:27 |
| Doyle | ./etc/default/whatever | 19:28 |
| sarnold | Doyle: hopefully all three | 19:28 |
| Doyle | I suspected. Thanks sarnold | 19:29 |
| Aethenelle | anyone know how I can get java's securitymanager to actually output debugging info under ubuntu's install of tomcat? | 19:35 |
| === markthomas is now known as markthomas|away | ||
| === xachet_ is now known as xachet_away | ||
| === Lcawte|Away is now known as Lcawte | ||
| iurygregory | Hello jamespage, I'm making puppet modules for openstack and I have a pythonlib called pysaml2 (which is installed using pip) it is possible to have a repository to be used apt-get? | 21:23 |
| === markthomas|away is now known as markthomas | ||
| === xachet_away is now known as xachet_ | ||
| Kallis | hi there, i was wondering if anyone could help me please, I am running this command to copy from a server to a local directory rsync -r -t -v --progress -s /run/user/1000/gvfs/smb-share:server=192.168.15.101,share=ac /tage-data/ac when the items copy they are dropped into a directory on the local system called cmb-share:server=192.168.15.101,share=ac but i want the files to overwrite or resume from the remote ac | 22:41 |
| Kallis | folder to the local ac folder | 22:41 |
| sarnold | Kallis: try adding a / to the end of both the source and destination | 22:42 |
| Kallis | ok i will try that now | 22:42 |
| sarnold | Kallis: /run/user/1000/gvfs/smb-share:server=192.168.15.101,share=ac/ /tage-data/ac/ | 22:43 |
| Kallis | that did it | 22:43 |
| Kallis | derp | 22:43 |
| Kallis | been playing for hours lol | 22:43 |
| sarnold | rsync is brutal that way :) | 22:44 |
| Kallis | thank you again :D | 22:44 |
| sarnold | or maybe _subtle_ is better description... | 22:44 |
| Kallis | heh | 22:44 |
| sarnold | anyway, I know i'v elost hours to that one too. hehe. | 22:44 |
| Kallis | i don't feel quite so derp now lol | 22:44 |
| Kallis | how to make acls work and i can go to bed :D | 22:45 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!