=== unrael__ is now known as unreal [05:28] jamespage: the right way is to turn around the server/client relationship [05:32] jamespage: qemu creates the socket OVS connects [05:32] no more needs the perm/user juggling [05:33] jamespage: see https://git.launchpad.net/~ubuntu-server/ubuntu/+source/dpdk-testing/tree/test.sh#n1056 and https://git.launchpad.net/~ubuntu-server/ubuntu/+source/dpdk-testing/tree/test.sh#n1043 for mode server (old) client (new) [05:34] the path is something qemu created after being set up via e.g. libvirt XML [06:14] Good morning === Casper26_ is now known as Casper26 === kklimonda_ is now known as kklimonda === coreycb_ is now known as coreycb === rsalveti_ is now known as rsalveti === fabiomirmar_ is now known as fabiomirmar === popey__ is now known as popey [08:22] cpaelzer: great - I'd got to that conclusion - I think we can just drop those params for openstack versions where ovs is the client to the libvirt created vhostuser socket [08:23] yes === lotuspsychje is now known as lotus|H3X === lotus|H3X is now known as lotuspsychje [12:57] Hi. I am facing an issue of systemd-resolved and dnsmasq probably looping and consuming 100% cpu, and would really appreciate some help! [12:58] test [12:58] http://paste.ubuntu.com/p/PH4MSBG9CK/ [12:58] ^^ /etc/resolv.conf [12:58] http://paste.ubuntu.com/p/NPWzVRR7mp/ is /etc/systemd/resolved.conf [12:59] http://paste.ubuntu.com/p/yZZ5FPGp7K/ is service systemd-resolved status [12:59] http://paste.ubuntu.com/p/dwf9nJnYbw/ is systemd-resolve --status [13:00] http://paste.ubuntu.com/p/X3kdGd4KJ8/ is `ip a` [13:00] https://github.com/iiab/iiab/issues/1747 is the bug report where we're tracking this [13:00] this is 18.04.2 LTS on an Intel NUC [13:01] m-anish: did you firmware upgrade your nuc to latest? [13:01] there is a bug about this, but it doesn't have a clear reproducer [13:02] 100% cpu looks like there is a query loop [13:02] I researched a bit on the internet, but couldn't find a proper answer, and I don't deeply understand how DNS resolution is working in ubuntu (given my system's config) [13:02] hmm let me check [13:02] ahasenack yeah, any way i can figure out what might be happening [13:02] tcpdump [13:03] any switches i should use with tcpdump? [13:03] it seems to happen sometimes with installations that were release upgraded from previous releases [13:03] this is a fresh install of 18.04 [13:03] if you don't know them, then tcpdump probably won't be useful to you [13:03] I'd suggest -i any -n -p port 53 [13:03] and brace for impact [13:03] if you want, i can let you ssh in via ngrok [13:03] nope [13:04] its awfully quiet `tcpdump -i any -n -p port 53` [13:05] wait [13:05] the race condition stopped itself :( [13:05] i mean im back to normal cpu usage [13:08] ill run that command if/when this happens next [13:08] ahasenack lotuspsychje thx [13:08] good luck [13:10] here we go again [13:11] https://paste.ubuntu.com/p/QhNfQF5h6Y/ [13:11] ahasenack ^^ is a sample [14:26] so, I have a performance issue with NFS mount, it seems that the sync option is slowing down writing a lot (unpacking a 13 GB gzip file takes up to 8 minutes with sync on but only a few seconds with async) [14:27] the async option, is it really that dangerous if the NFS mount is used mostly for reading? [14:32] its mostly used for Joomla websites, most writing is done when users upload images and other files and when the website admin updates the software [14:32] the software update is the 13 GB unzip [14:33] you should have separate exports then [14:33] I have a seperate export for each site [14:34] played around with rsize and wsize, no luck there [14:35] you should have one (maybe for each site) which mounts writable with async, and another (maybe for each site) which mounts read-only without async [14:35] hmmm, don't think sync/async matters for read, only for writing [14:38] and i'd recommend unpacking archives on the nfs server, not on a mounted export [14:39] well, the update is done through the Joomla web interface, don't have an option there I think [14:40] what I am mostly interested in is if async is active, what happens if a write is taking place and the link gets disrupted somehow === tinwood_ is now known as tinwood === MassDebates_ is now known as MassDebates === andol_ is now known as andol === SoniEx2 is now known as Soni [19:42] Hi, [19:43] ho [19:43] Can we run GUI applications in Ubuntu server , and access it via ssh graphically. [19:43] ? [19:43] yes [19:44] works well - I'm using for instance virt-manager that way [19:44] How? I tried installing xorg , but failed to do so [19:44] virt-manager is virtualization things. [19:45] just install something, like x11-apps - log out and in again to make sure x11-forward works and start xeyes or something [19:45] bipul: it's still just a normal x application [19:46] Yes, an IDE type application like IntelliJ [19:46] should work [19:46] just try xeyes first [19:46] it's a neat little thing for testing x [19:46] I tried installing xauth , xorg and openbox. [19:46] no need for xorg [19:46] Then? [19:47] that's the x server - you only need an x server on your client [19:47] server/client in x land is a bit opposite of the way others name it [19:47] bipul: what sort of client are you running? [19:47] I am using Ubuntu , inside virtualbox. [19:48] inside virtualbox Ubuntu server, where i would like to install IntelliJ [19:48] just install x11-apps and log out and in again and check if $DISPLAY is set [19:48] Oh let me try [19:49] you may have to ssh -X to enable x11 forwarding, but it should be enabled by default, both on client and server [19:50] yes, [19:52] RoyK, Thanks for your wonderful suggestion. [19:53] works? [19:54] I am trying [19:54] check the DISPLAY variable if it doesn't work - "echo $DISPLAY" [19:55] When you say x11 apps? then what specific package should i install ? [19:55] apt-cache search x11 list me many packages [19:56] TJ-: so your patch is now in Proposed for nginx heere in Eoan. PROBLEM: chromium being snapped now broke autopkgtests for something and those autopkgtests block nginx. unrelated blockage but still [19:56] Laney and others're looking into that one [19:56] bipul: x11-apps [19:56] yes x11-aps [19:56] yes x11-apps [19:58] so - now log out and in again and check if DISPLAY is set [20:00] sure i will let you know :) [20:00] or just try to run xeyes first [20:00] you'll see [20:00] or get an error message [20:02] teward: I'm not fixing Chromium! [20:02] TJ-: no you aren't :P [20:03] lol I wouldn't expect you to xD [20:05] Okay, echo $DISPLAY does not works. [20:05] no output? [20:05] Yes no output. [20:06] ok - try ssh -X [20:08] It says "/usr/bin/xauth: file /home/ubuntu/.Xauthority does not exist." [20:08] log out and in again [20:08] this usually happens the first time you login with x11 forward [20:08] I think we need to give small "x" [20:09] -x disables x11 forward [20:09] -X enables it [20:09] see man ssh [20:09] WoW ! It works [20:09] Thank you very much. :) [20:10] then set ForwardX11 yes in .ssh/config [20:10] so you won't need to use -X next time [20:11] okay :) [20:11] or just globally in /etc/ssh/ssh_config [20:11] on the client, that is [20:12] Remember that "ForwardX11 yes" potentially allows attacks from the server [20:12] You might want to set it for the one server [20:12] lordcirth: my guess is his server is behind nat anyway, so it won't be much of a problem [20:13] The problem is if he sets it globally, then ssh's to a different server later [20:14] I know [20:14] bipul: check "man ssh_config" - you can add separate blocks for each server [20:18] lordcirth: btw, isn't ForwardX11Trusted needed for the server to gain too much access? [20:19] as in ssh -Y === hays_ is now known as hays