| === markthomas is now known as markthomas|away | ||
| === shauno_ is now known as shauno | ||
| === ilken is now known as ilk | ||
| _andy_ | Hi there! I'm having issues with my AD DC showing up in the network. | 03:42 |
|---|---|---|
| _andy_ | I am able to resolve NetBIOS to an IP address and vice versa, but when searching for computers over the network, the domain controller does not show up. | 03:42 |
| _andy_ | Hi there! I'm having issues with my active directory domain controller showing up on the list of NetBIOS computer on the network. | 03:50 |
| lordievader | Good morning. | 06:00 |
| === Lcawte|Away is now known as Lcawte | ||
| adsc | does ubuntu server automatically use kexec/ksplice technology or do we have to restart after a kernel update? | 08:27 |
| lordievader | For the time being you need to restart after a kernel update. | 08:27 |
| lordievader | Perhaps with Wily or 16.04 you no longer need to. | 08:28 |
| RoyK | adsc: http://www.zdnet.com/article/no-reboot-patching-comes-to-linux-4-0/ | 09:04 |
| === lukasa is now known as Guest4580 | ||
| jpds | adsc: You don't HAVE to restart for a kernel upgrade | 09:07 |
| RoyK | jpds: well, unless you have hotpatching, you'll still be running the same old kernel code :P | 09:11 |
| jpds | RoyK: Well, if the kernel upgrade in question doesn't touch any modules you use... | 09:12 |
| adsc | ok, thx...I was just confused because a colleague claimed that after a kernel patch he ran uname -a and it showed the new kernel already | 09:13 |
| adsc | which was surprising me, because I set up the system and haven't installed any ksplice or similar technology | 09:14 |
| adsc | but I guess he was just wrong, then | 09:14 |
| jpds | ksplice wouldn't even change what uname -a shows | 09:15 |
| jpds | adsc: You just see a load of "ksplice_*" entries in lsmod | 09:16 |
| mufloirc | hi guys | 09:16 |
| mufloirc | I'm having an issue with the maas region controller in Ubuntu 14.04 (maas-region-controller 1.8.0+bzr4001). Basically I'm trying to import the boot resouces with this command "sudo maas e4 boot-resources import" but it fails. | 09:18 |
| jpds | mufloirc: Tried importing from the web UI? | 09:19 |
| mufloirc | no, with the cli | 09:19 |
| jpds | mufloirc: Tried importing from the web UI? | 09:19 |
| mufloirc | looke like the the signature is not good | 09:20 |
| mufloirc | "gpg: Signature made Tue Sep 1 02:46:44 2015 CEST using RSA key ID 476CF100\ngpg: BAD signature from "Ubuntu Cloud Image Builder" | 09:20 |
| ivoks | mufloirc: using proxy? | 09:21 |
| mufloirc | no, I'm not using proxy | 09:21 |
| ivoks | Signature made Tue Sep 1 | 09:22 |
| ivoks | sounds like today :0 | 09:22 |
| RoyK | adsc: should work with linux 4.x | 09:22 |
| adsc | good to know, thanks | 09:25 |
| jpds | mufloirc: Yeah, apparently that broke today, they're working on it | 09:26 |
| mufloirc | thanks jpds | 09:27 |
| === lukasa_ is now known as lukasa | ||
| === mufloirc is now known as Muflo | ||
| rbasak | utlemming: could you take a look at bug 1489675 please? | 10:15 |
| ubottu | bug 1489675 in open-vm-tools (Ubuntu) "open-vm-tools-dkms 2:9.10.2-2822639-1ubuntu3: open-vm-tools kernel module failed to build [error: ‘new_sync_read’ undeclared here (not in a function)]" [High,Confirmed] https://launchpad.net/bugs/1489675 | 10:15 |
| === martins-afk is now known as martins | ||
| jamespage | morning smb - hows dpdk looking? | 10:41 |
| smb | jamespage, lonely | 10:41 |
| smb | jamespage, I filed bug 1487538 for ffe but I am not sure I have done the right steps to get MOTUs (or whomever) take notice | 10:42 |
| ubottu | bug 1487538 in Ubuntu "[needs-packaging] [FFE] Add dpdk to wily universe" [Wishlist,New] https://launchpad.net/bugs/1487538 | 10:42 |
| jamespage | smb, does not need a motu - either rbasak or myself can sponsor that | 10:54 |
| rbasak | jamespage: well, we're both motus :) | 10:55 |
| smb | jamespage, just that rbasak mentioned it needs probably acks before sponsoring | 10:55 |
| rbasak | smb, jamespage: my only concern really is that we don't create a future ABI nightmare, so I wanted to get ABI agreement from upstream before uploading. | 10:56 |
| rbasak | smb: it needs an FFe ack from a release team member before upload | 10:56 |
| rbasak | Sorry I've been slow with getting the upstream communication sorted. My todo list is still too long :( | 10:56 |
| rbasak | I'm catching up with maintenance and sponsoring today, then have to get some juju sponsorship done. But we can sync when arges is up maybe. | 10:57 |
| jamespage | rbasak, smb: a sync sounds like a good idea | 10:57 |
| smb | jamespage, rbasak Yeah. Not sure timewise. Tuesdays are meetingful. | 10:59 |
| smb | But maybe we can squeeze something in after either your or our irc meeting | 10:59 |
| jamespage | rbasak, fwiw I'm also participating in a thread on the ovs ML re DPDK which is pertinent to this conversation. | 11:17 |
| rbasak | jamespage: thanks. We'll sync as soon as we can. | 11:23 |
| smb | rbasak, jamespage, would you have time in about 1.25hrs? (for the sync) | 12:23 |
| rbasak | smb, jamespage, arges: 1400 UTC? Yes for me. I've sent an invite as a starting point - we can move. | 12:29 |
| jamespage | rbasak, +1 | 12:30 |
| rbasak | smb: on a separate note, who looks after HWE related DKMS issues? I triaged https://bugs.launchpad.net/ubuntu/+source/iscsitarget/+bug/1483415 | 12:30 |
| ubottu | Launchpad bug 1483415 in iscsitarget (Ubuntu) "iscsitarget-dkms fails to build on 14.04" [Undecided,Confirmed] | 12:30 |
| smb | rbasak, jamespage, arges +1 (for meeting) | 12:31 |
| smb | rbasak, I might. It a bit of a shared duty | 12:31 |
| jamespage | rbasak, urgh - could we bump it 30 mins | 12:31 |
| smb | if it is shortened to 30mins as well... | 12:33 |
| rbasak | smb: do you know of any testing we do for HWE+DKMS? It seems sub-optimal to me that we have to be reactive on this. | 12:33 |
| rbasak | jamespage: I have no objection but looks like +30 minutes would clash for smb and arges | 12:34 |
| smb | rbasak, there is testing usually. think something broke for a bit recently | 12:34 |
| arges | +30m is fine too. smb we can move our mtg? | 12:36 |
| smb | arges, maybe a bit. thought there is other things following up. But potentially we don't need the full hour and can dynamically start | 12:38 |
| rbasak | smb, jamespage, arges: OK, moved. | 12:40 |
| smb | rbasak, So we checked and that should be fixed in the current update of the hwe kernel (3.19.0-26) | 12:50 |
| rbasak | smb: I'm sorry, it isn't hwe at all. He's using Wily. | 12:57 |
| rbasak | Err | 12:57 |
| * rbasak is very confused | 12:57 | |
| smb | rbasak, in the bug report its 3.19 | 12:57 |
| smb | thats not wily | 12:57 |
| rbasak | I think I'm looking at the wrong bug | 12:57 |
| rbasak | Sorry, triaging too many bugs at once | 12:58 |
| rbasak | smb: you're right. Thanks for the comment. | 12:58 |
| smb | rbasak, no worries. yw. Its unfortunate the kernel on the iso has that flaw... and people tend to not update | 12:59 |
| smoser | rbasak, have you done any thing other than triage https://bugs.launchpad.net/ubuntu/+source/keepalived/+bug/1481337 | 14:32 |
| ubottu | Launchpad bug 1481337 in keepalived (Ubuntu Vivid) "keepalived makes a floating IP available on more than one host after configuration reload" [Medium,Confirmed] | 14:32 |
| samba35 | is there any way to change log level for dmesg | 15:07 |
| === martins is now known as martins-afk | ||
| smb | rbasak, OK, let me know how the updated paragraph for dpdk sounds to you | 15:41 |
| === caitanya is now known as neurotus | ||
| === neurotus is now known as Guest84381 | ||
| jamespage | coreycb, is troveclient blocking all of the other openstack packages via heat failing its build/dep8 tests? | 15:53 |
| coreycb | jamespage, yes | 15:53 |
| coreycb | jamespage, everything should be freed up once the troveclient FFE gets through and we can sync the new version | 15:54 |
| jamespage | coreycb, as that's such a blocker, I'd be tempted to sync it without the FFe right now | 15:56 |
| jamespage | coreycb, I just found: | 15:57 |
| jamespage | http://tarballs.openstack.org/python-troveclient/python-troveclient-master.tar.gz#egg=python-troveclient | 15:57 |
| jamespage | in the test-requirements of openstack-trove | 15:57 |
| jamespage | urgh | 15:57 |
| coreycb | jamespage, yeah.. wasn't sure if I should open a bug on that or what | 15:57 |
| coreycb | jamespage, fine by me if we can sync it without an ffe | 15:57 |
| jamespage | coreycb, I was trying to see the heat dependency on it - I'm assuming its not expressed correctly upstream | 15:58 |
| === gaughen_ is now known as gaughen | ||
| coreycb | jamespage, I opened a bug for trove's test-requirements.txt | 16:04 |
| coreycb | jamespage, shall I go ahead and sync troveclient? | 16:05 |
| jamespage | coreycb, wait a sec | 16:05 |
| === markthomas|away is now known as markthomas | ||
| smoser | rbasak, have you done any thing other than triage https://bugs.launchpad.net/ubuntu/+source/keepalived/+bug/1481337 . wondering if you have thoguht on how hard to push on a fix for that. | 16:17 |
| ubottu | Launchpad bug 1481337 in keepalived (Ubuntu Vivid) "keepalived makes a floating IP available on more than one host after configuration reload" [Medium,Confirmed] | 16:17 |
| rbasak | smoser: no, only triaging the bugs that I'm assigning - not looking deeply into anything. | 16:18 |
| rbasak | smoser: I'm fine with it if you think it's more effort to SRU than it's worth, which it sounds like it is now that you've looked. | 16:18 |
| smoser | well, its definitely not "just cherry pick patch" | 16:18 |
| rbasak | smoser: maybe we can drop the SRU bug task importance to Low and I'll take it off my radar | 16:18 |
| dork | yikes | 16:23 |
| === kickinz1 is now known as kickinz1|afk | ||
| jamespage | coreycb, I'll sync it over - you'll get kudos and I'll get any release team flak | 16:27 |
| coreycb | jamespage, ok thanks very much. definitely need to move forward on liberty so I think it's worth it. | 16:28 |
| jamespage | coreycb, if its the fix for a blocking issue, I think we can skip the FFe tbh | 16:28 |
| coreycb | jamespage, good point | 16:28 |
| jamespage | coreycb, ok synced - you can probably rebuild heat now | 16:34 |
| jamespage | well in a bit | 16:34 |
| jamespage | once its published | 16:34 |
| coreycb | jamespage, ok will do | 16:34 |
| jamespage | coreycb, urgh - its stuck in NEW | 16:34 |
| coreycb | jamespage, oh gosh, because of the epoch? | 16:35 |
| jamespage | coreycb, no cause of the python3 support | 16:35 |
| jamespage | coreycb, accepted | 16:38 |
| coreycb | jamespage, awesome | 16:39 |
| Kalimero | i want to include my .htaccess to darkstats webui but i dont know where the wbfiles laying. any hints? | 16:45 |
| smb | rbasak, [repost] OK, let me know how the updated paragraph for dpdk sounds to you | 16:50 |
| teward | i assume i missed the server meeting. | 16:56 |
| njbair | where would the relevant logs be for the "waiting for network configuration" startup delay? | 17:04 |
| teward | njbair: nowhere, my guess is you have an 'auto eth0' or something in there that's holding up network booting, either by DHCP wait, or otherwise | 17:05 |
| teward | (maybe in the boot logs, but if it's just waiting for configuration by DHCP, etc, it might not be listed) | 17:05 |
| njbair | both of my interfaces are configured statically | 17:06 |
| njbair | don't I want the auto eth0 line to bring up the interface automatically? | 17:07 |
| teward | let it error out then check network information and see which one failed to 'configure' or rise up. If neither have no config, then i'm not sure how you'll check... (most of mine come up instantly, and they're static configured too) | 17:08 |
| Wicaeed | LVM/Virtualization question: Is there any difference, from an LVM standpoint, of adding an additional virtual disk to a VM vs. giving expanded capacity to an already existing virtual disk? | 18:33 |
| Wicaeed | Oh, right, when you add capacity to an existing virtual disk fdisk doesn't see the additional capacity until you reboot. Herp | 18:34 |
| === markthomas is now known as markthomas|away | ||
| RoyK | Wicaeed: iirc echo '- - -' > /sys/class/scsi_host/host0/scan | 18:59 |
| bekks | Ah, good old SCSI reset :) | 19:00 |
| RoyK | Wicaeed: have used that several times with vmware guests ;) | 19:01 |
| Wicaeed | orly | 19:01 |
| patdk-wk | doesn't work on pvscsi :) | 19:23 |
| patdk-wk | you have to use something else to reset it | 19:23 |
| === tomaw_ is now known as tomaw | ||
| Tangurin | Hi! aynone online? | 20:22 |
| Tangurin | anyone* | 20:22 |
| bekks | Tangurin: Look at "/names". | 20:23 |
| === markthomas|away is now known as markthomas | ||
| Tangurin | Ahh! I ask because this weird server crash just happened to me! I was logged in to the site I am building, in the administration. When I press edit (an AJAX request is sent) the server crash! the Apache2 crash I guess! how is this possible? a site crash apache2? | 20:24 |
| bekks | Look at the logs. | 20:24 |
| Tangurin | what log should I look in bekks I am looking in? | 20:28 |
| bekks | Tangurin: /var/log/ | 20:28 |
| Tangurin | bekks: okey but there is 1000 logs, where should I start :) | 20:29 |
| Tangurin | bekks: when I press the save button in my administration, all my sites at the server goes offline and I get kicked from SSH! where is the logs for that, where do you think? :) | 20:30 |
| bekks | Tangurin: As I already told you. | 20:30 |
| Tangurin | bekks: you told me /var/log/ | 20:32 |
| bekks | I did. | 20:33 |
| bekks | Whats unclear with that? :) | 20:33 |
| Tangurin | I don't know where to look :/ bekks , This is so weird, I press save, an AJAX request is sent, and directly the server goes down | 20:35 |
| bekks | Tangurin: So look at the logs in /var/log/ | 20:36 |
| Tangurin | bekks: okey I will try find something | 20:36 |
| bekks | You will find logs in there. | 20:37 |
| Tangurin | bekks: I find them but I don't know which one to look for or what to find! | 20:38 |
| bekks | So look at the logs which are younger than your crash. And no need to use exclamation marks all the time. | 20:39 |
| JaguarDown | Hi all newbie here. Upon trying to restart the services clamav-daemon and clamav-milter, the daemon "clamd" appears to successfully stop but upon attempting to start it tries to change ownership of two files but says operation not permitted. Then clamav-milter stops but can't remove the PID file, tries to chown same two files, and kill the clamav-milter process, and set the GID but can't due to "operation not permitted" | 21:18 |
| JaguarDown | http://pastebin.com/8qmSMLwM | 21:18 |
| JaguarDown | Formatting got weird, not sure why. I am sure I have file permissions wrong somewhere but don't know where. I am following a tutorial to set up email and have rechecked everything exactly. | 21:18 |
| JaguarDown | Just wondering if anybody had any experience with this. | 21:18 |
| JaguarDown | This is the page of the tutorial I am currently on http://arstechnica.com/business/2014/03/taking-e-mail-back-part-3-fortifying-your-box-against-spammers/2/ | 21:26 |
| poorvikrampandit | Hi, just had a pretty major emergency where my Ubuntu server (running on a VMWare ESX vm with 2048 MB RAM, 1 CPU allocated) became completely unresponsive. | 21:28 |
| bekks | poorvikrampandit: Add more CPU, more RAM. | 21:28 |
| tgm4883 | +1 | 21:28 |
| poorvikrampandit | Is it bad practice to run a memory & CPU-intensive application (image cropping server) on the same system as Postgresql and web application? I'm worried about Postgres becoming inconsistent (the OS killed the application, but I'm sure postgres could get killed too) | 21:28 |
| bekks | poorvikrampandit: It is a bad practice to grant less resources than you need. | 21:28 |
| tgm4883 | than you need on a regular basis | 21:28 |
| tgm4883 | But yea, more resources! | 21:29 |
| bekks | +1 | 21:29 |
| poorvikrampandit | bekks: How do I convince my boss the problem is RAM? | 21:32 |
| poorvikrampandit | err, resources. | 21:33 |
| bekks | Repeat the sentence to him. | 21:33 |
| bekks | Last time we ran database hosts with 2GB RAM was like... 2000. | 21:33 |
| poorvikrampandit | So does it sound like we are doing too much with too little, and not that our image cropping service is using too many resources? | 21:33 |
| bekks | You are doing to much with too little. | 21:34 |
| tgm4883 | poorvikrampandit: you could watch the resources and then see what it does when you put load on it | 21:34 |
| tgm4883 | but I'd add more ram and a second CPU | 21:34 |
| bekks | At least. | 21:34 |
| tgm4883 | sorry, yes. At least that. Depending on how many users you intend to have | 21:34 |
| poorvikrampandit | You are being very helpful and I admit I am terribly out of my league. How much should I ask for? | 21:35 |
| tgm4883 | I don't build VMs with less than 1 CPU anymore | 21:35 |
| mybalzitch | how could you have less than 1 cpu in a vm | 21:35 |
| bekks | poorvikrampandit: Double the resources. | 21:35 |
| tgm4883 | poorvikrampandit: what does your host machine have for resources? | 21:35 |
| bekks | poorvikrampandit: If the problem persists, analyze resource usage to see wether you need even more RAM or CPU or both. | 21:35 |
| tgm4883 | mybalzitch: heh, I meant 2 CPUs | 21:35 |
| tgm4883 | mybalzitch: although I suppose if you are oversubscribed enough it's like having half a CPU | 21:36 |
| poorvikrampandit | bekks: That sounds doable if I kick off the other VMs running on the same machine (the host has 6GB and 2 3 GHz cores) | 21:36 |
| tgm4883 | poorvikrampandit: how many users are you planning on? | 21:36 |
| tgm4883 | are we talking 10, or 1000 | 21:37 |
| bekks | poorvikrampandit: Oh geez. Replace the host with current system. | 21:37 |
| poorvikrampandit | tgm4883: Around 200 users with current hardware budget. | 21:40 |
| poorvikrampandit | What I'm most worried about is this image cropper causes the kernel to go OOM on Postgres and data becomes inconsistent. Does moving the image cropper to its own VM sound like a way of insuring against this? It would at least guarantee the web app stays on, | 21:42 |
| bekks | poorvikrampandit: Then get a server with enough RAM. | 21:43 |
| bekks | poorvikrampandit: And split services into separate VMs. | 21:43 |
| tgm4883 | Yes, separate VMs would help here | 21:44 |
| poorvikrampandit | Okay, so if you had a psql & redis, a web application (Node), image cropper (Golang), page rasterizer (another troublesome service written in go that thumbnails web pages), plus various other little go services that are more benign, how would you split them into VMs? I'm thinking what I can do is put the data-critical stuff (DBs, web app, services that aren't resource-intensive) on one VM, and the 2 resource-intensive s | 21:48 |
| bekks | One critical service - one vm. A bunch of logical dependant uncritical services - one vm. | 21:49 |
| poorvikrampandit | So I should split into two VMs even if I can't get a faster machine? | 21:52 |
| poorvikrampandit | err, more RAM | 21:52 |
| bekks | Ermm - no. | 21:53 |
| bekks | If you cant get more CPU/RAM on your current host, get new hardware. | 21:53 |
| poorvikrampandit | What if I'm stuck with the hardware I have, but I can kick the other VMs off and then have the full 6 GB to play with? | 21:54 |
| bekks | You wont get more CPU then, since you have to leave one physical CPU for your host. | 21:54 |
| bekks | And then - you have one CPU only for your VM. | 21:54 |
| JaguarDown | Not that anyone was listening but please disregard my last question. The problem lied with not using sudo after the && operator. The tutorial's commands were being carried out in a root shell whereas I was playing it safe and using sudo for everything. Thanks anyway though. | 21:54 |
| poorvikrampandit | Well, the machine has 2 cores -- we just only have one allocated right now to this service | 21:55 |
| bekks | poorvikrampandit: two cores - one for the host, one for the vm. no more cores left. | 21:55 |
| bekks | poorvikrampandit: your hardware just has too less resources. | 21:56 |
| poorvikrampandit | The VM can't use both cores if I tell VMware to let it use both? | 21:56 |
| bekks | Do NOT do that. Listen again: always leave ONE entire CPU for the host. | 21:56 |
| sarnold | JaguarDown: woot :) | 21:56 |
| poorvikrampandit | bekks: I'm writing that down. | 21:56 |
| sarnold | bekks: that's good idea. | 21:57 |
| JaguarDown | Let me rephrase that, it sounded rude: Not that anyone was available to answer my question is what I meant. | 21:57 |
| bekks | poorvikrampandit: Engrave it over your door. Write it down the hallway. Print it on all flags inside and outside your office. | 21:57 |
| poorvikrampandit | Hmm, what do I do then if I can't get better hardware? | 21:57 |
| bekks | poorvikrampandit: Cry, or change your employer. | 21:57 |
| bekks | poorvikrampandit: if you cant get more resources, the situation will not change. | 21:58 |
| sarnold | poorvikrampandit: maybe you can rent better hardware from hetzner or amazon or digital ocean? | 21:58 |
| sarnold | poorvikrampandit: for some stupid reason managers love renting things at twice the price of buying things. | 21:58 |
| bekks | sarnold: :D | 21:58 |
| poorvikrampandit | bekks: I can't ameliorate by going from 2048 to the full 6GB? | 21:58 |
| poorvikrampandit | Yeah, the hardware is rented. | 21:59 |
| dasjoe | sarnold: I tried to rent out some storage servers, they preferred to buy them :( | 21:59 |
| bekks | poorvikrampandit: Using 6GB RAM will leave no RAM for your host. VMware ESXi has a requirement of at least 2GB for the host. | 21:59 |
| sarnold | dasjoe: ha! :) | 21:59 |
| sarnold | dasjoe: ohhhh, I understand now. boo. :( sorry to hear it. | 22:00 |
| poorvikrampandit | Okay, what do I do to keep this thing on until the new server comes in? | 22:00 |
| dasjoe | sarnold: still a good deal, but renting them out would've created passive income | 22:00 |
| bekks | poorvikrampandit: Nothing. Hope that your problem will not occur again. | 22:00 |
| sarnold | dasjoe: hey does that mean you're selling pre-made boxes now? :) | 22:00 |
| poorvikrampandit | I may be reshuffling the chairs on the titanic's deck, but what if I just move the offending service to one of the other VMs (shutting down what's on that VM)? Then at least ImageCropService can't bring down the web app & Postgresql. | 22:02 |
| jelly | poorvikrampandit: since you're using ESX as host, you should be able to see some performance graphs in the fat client of the flash web client | 22:03 |
| jelly | and maybe deduce what the bottleneck was, this time | 22:03 |
| poorvikrampandit | jelly: Good idea | 22:04 |
| dasjoe | sarnold: I'm still thinking about doing it at a larger scale, yes. This box was built for a regular client, have some blinkenlights during a scrub: http://i.imgur.com/f7Xl3nf.gifv | 22:04 |
| tgm4883 | bekks: 2GB? I don't think ESXi uses that much by default. I'm not even sure you can force it that high | 22:05 |
| sarnold | dasjoe: so very coool :) | 22:05 |
| jelly | tgm4883: heh, that "much" | 22:05 |
| tgm4883 | jelly: that's a significant amount for something running nothing else but a hypervisor | 22:06 |
| poorvikrampandit | It might be possible to move ImageCropper to EC2 if I can get a static IP. | 22:06 |
| jelly | tgm4883: it's not much slack space for a meager 72GB RAM host | 22:07 |
| tgm4883 | jelly: It's possible I'm thinking about something else. I thought there was a setting in ESXI to set the memory that could be used by ESXi, and I thought it defaulted to 768MB and maxxed at around 1.5GB | 22:10 |
| poorvikrampandit | Should I look at moving the CPU-intensive services to EC2 as an option? I doubt I can get any more RAM. | 22:13 |
| poorvikrampandit | (beyond the physical 6GB) | 22:14 |
| jelly | you should investigate what really happened, first | 22:16 |
| tgm4883 | poorvikrampandit: I think you should probably look at how much that would cost, vs buying more ram | 22:17 |
| YamakasY | hi guys | 22:18 |
| YamakasY | anyone a clue why the mysql.log files keep empty when I enable logging ? | 22:18 |
| poorvikrampandit | Am I going to need to double my RAM&CPU again when I go from 4 users to 8 users, or is this more to do with a baseline cost to get the databases (postgres, redis) & services up and running? | 22:18 |
| dasjoe | It might be more cost efficient to "upgrade" this servers resources by buying a newer server instead | 22:19 |
| tgm4883 | poorvikrampandit: you shouldn't need to double at that point. I think you're under the baseline just for running that | 22:19 |
| prudentmav | how can I have a new /var/www/$user/ created when I run adduser? | 22:22 |
| prudentmav | I want a public and logs dir in the user dir as well | 22:22 |
| tgm4883 | prudentmav: you need to tell it to create the home directory and where it should be located | 22:22 |
| sarnold | prudentmav: look for /usr/local/sbin/adduser.local in adduser(8) | 22:22 |
| poorvikrampandit | tgm4883: How do I know what the baseline is? | 22:29 |
| === Lcawte is now known as Lcawte|Away | ||
| sarnold | to a first approximation, add gobs of swap space, load it all, and see how deep into swap you are | 22:32 |
| sarnold | that's only a first-approxmiation because it ignores the performance improvement you might get if more data fits into the kernel's block, page, and dentry caches -- most of which will be dropped under memory pressure | 22:33 |
| sarnold | but if you're five gigs deep into swap, adding six gigs more RAM would help. if you ten gigs deep, adding twelve would help. etc. | 22:34 |
| shauno | is it typical behaviour for LDS to run many copies of the same processes? in particular, 6 copies of process-usns and 8 copies of process-alerts looks out of place to my eye | 22:34 |
| sarnold | shauno: that may just be how they are multithreaded; check pstree output, do they all originate from the same parent process? if so, it's likely running as designed. if not, itm ight be worth further investigation. | 22:36 |
| shauno | well, my attention was drawn to it because landscape keeps alerting it's "exceeding run timeout". which is why it's getting an odd look | 22:38 |
| sarnold | sounds like a well-deserved odd look :) | 22:39 |
| shauno | if I understand pstree correctly, this looks like 8 seperate copies where the 2nd is launched before the first dies? http://paste.ubuntu.com/12249134/ | 22:39 |
| jelly | or it's just some of the cron jobs never die, and they don't use locking to avoid running in parallel either | 22:40 |
| jelly | (which in turn points to a possible different underlying issue) | 22:41 |
| sarnold | yeah, I'm with jelly. please file a bug, make sure to include that output... | 22:41 |
| jelly | update_security has nothing to do with LDS, does it? | 22:42 |
| sarnold | those ought to have some locking around them, but there's probably something else broken to make the first one not die.. once the first one doesn't die, all bets are off.. | 22:42 |
| jelly | and yet there are seven of those as well | 22:42 |
| jelly | if the kernel or filesystem is confused, all kinds of hangs like that can happen. It's a nice way to find bugs that only happen in extreme conditions, but I'd bet those jobs are just collateral damage | 22:44 |
| sarnold | check dmesg, there may be IO errors logged htere | 22:45 |
| shauno | almost nothing in dmesg, my only even since boot (~18 hours) was [30155.163525] perf interrupt took too long (2501 > 2500), lowering kernel.perf_event_max_sample_rate to 50000 | 22:47 |
| poorvikrampandit | Okay, I wasn't able to get any more RAM, and couldn't get rid of all the other VMs, but I did negotiate moving the CPU/RAM intensive processes (ImageCrop.go and PageRasterizer.go) to their own instance, so the data-critical web app & postgres are no longer in danger of OOM. | 22:53 |
| poorvikrampandit | Production will use two of the VMs now instead of one: 2 GB VM for mission critical, 1.5 GB VM for ImageProxy & PageRasterizer. | 22:54 |
| poorvikrampandit | Maybe this buys time? | 22:54 |
| poorvikrampandit | My boss will turn into Alec Baldwin in Glengarry/Glenross if this goes down again. | 22:55 |
| sarnold | more memory shouldn't be too bad, something like $310 USD for another 32 gigs of ddr4.. http://www.crucial.com/usa/en/poweredge-t630/CT6236763 | 22:59 |
| dasjoe | He shouldn't skimp on hardware, then | 22:59 |
| poorvikrampandit | It's rented hardware. (Unless they'll let me into their datacenter pop in another SIMM.) | 23:02 |
| sarnold | 140EUR/mo can get you 128 gb of memory: https://www.hetzner.de/hosting/produkte_rootserver/px121ssd -- maybe they can be more competitive. or maybe you can move to hetzner. :) | 23:04 |
| RevertToType | i cannot get sound out of my ubuntu-server install | 23:04 |
| RevertToType | alsa nope, pulseaudio nope | 23:05 |
| RevertToType | def recognizes the soundcard as "there" | 23:05 |
| RevertToType | aplay /testwavs plays them but no sound through the speakers, alsa-mixer is wrenched up all the way, no bios issues, no acpi volume button messing me up | 23:06 |
| sarnold | RevertToType: check permissions on the /dev/ nodes for the audio devices. iirc consolekit or something similar normally chowns those to logged in users when doing lightdm or whatever, but those pathways may not be used in a server install | 23:06 |
| sarnold | RevertToType: also, note that alsa starts muted; perhaps it needs to be unmuted? | 23:06 |
| RevertToType | wouldn't alsa-mixer set that? | 23:07 |
| sarnold | it might, that might be the MM at the bottom of the sliders.. | 23:07 |
| sarnold | but it might also let you slide volumes around withuot actually unmuting, too | 23:07 |
| poorvikrampandit | sarnold: Are hetzner reputable? | 23:10 |
| poorvikrampandit | We've suffered from a lot of bozo hosts (like one in New York State that let the servers overheat and had bad RAM)? | 23:11 |
| sarnold | poorvikrampandit: I think they have a bit of reputation as a 'bargain' outfit, but I haven't heard any specific complaints about services or customer service | 23:11 |
| poorvikrampandit | Hmm, what does bargain mean? | 23:13 |
| sarnold | cheap, thrifty | 23:15 |
| Tangurin | Hi! I use Laravel on my ubuntu server but something is really wrong and the server crash and I get logged out from SSH and all sites on the server goes down when I trying to do a simple ajax request at the page! the /var/log/apache2/error.. doesn't tell me anything, where can I look for more information? | 23:22 |
| RevertToType | thank you sarnold | 23:22 |
| RevertToType | now to figure out why my networking periodically drops | 23:22 |
| sarnold | RevertToType: got it? :) what was it? :) | 23:22 |
| RevertToType | it was re-muting despite alsactl saving | 23:22 |
| sarnold | Tangurin: check all the logs | 23:22 |
| sarnold | RevertToType: oof. :/ | 23:22 |
| RevertToType | but i already have a bashlogin script so it's all good | 23:23 |
| Tangurin | sarnold: what shall I look for, I can't see anything wrong | 23:23 |
| RevertToType | hrm... my wpa-2 connection every once in a while drops... then i need to ifdown/ifup and re-dhclient ... anything offhand a common ish? | 23:24 |
| sarnold | Tangurin: out-of-memory killer, segmentation faults, kernel panics, block io errors, etc etc.. | 23:25 |
| Tangurin | sarnold: thanks | 23:25 |
| Tangurin | sarnold: init: failsafe main process (766) killed by TERM signal is this something? | 23:26 |
| RoyK | if you see massive errors on a server (or even desktop) it's usually something bad in the hardware. try a memory test first. also, a bad PSU can make your day longer, in bad terms | 23:31 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!