=== PaganMin is now known as GitGud === GitGud is now known as GameOfTrolls === Kenrinx is now known as kenrin === cpaelzer_ is now known as cpaelzer === _degorenko|afk is now known as degorenko [09:27] Good morning. [09:29] good morning lordievader [09:30] jamespage: I was (trying to) start packaging the daily git of openvswitch for experiments with dpdk 16.04 [09:30] Hey cpaelzer [09:30] jamespage: but I found that the orig tarball matches non of the sources I checked, so I wonder what you are usually based on [09:31] jamespage: e.g. old 2.5 tarball doesn't match release tarball, the ones of the 2.5 git you had don't match (in content style) what one can find at https://github.com/openvswitch/ovs/tarball/master [09:31] jamespage: so I just wonder what your usual entry point is to work with [09:32] jamespage: any repo I could clone or something like it [09:32] jamespage: I'll start digging through the packaged readmes instead of just trying - maybe you documented it already :-) [09:32] cpaelzer, release tarballs from openvswitch.org (not github - checksum will never match) [09:33] cpaelzer, repo - https://code.launchpad.net/~ubuntu-server-dev/ubuntu/+source/openvswitch/+git/openvswitch [09:33] its git-buildpackage based [09:33] jamespage: something new to become desperate ... umm learn :-) - thanks [09:34] I came by gbp just this morning, good friday opportunity [09:34] leaves me in a shattered state to the Weekend to regenerate :-) [10:17] teward: see https://lists.ubuntu.com/archives/ubuntu-devel-discuss/2016-June/016654.html [10:18] teward: that's up to you I guess, but I suspect the sensible answer is "no". [10:19] teward: and a reasonable solution would be "ship it inside Ubuntu". [11:14] però è vero da oggi sono extracomunitari [11:22] jamespage: I tried to build the repo you referred to as is with cowbuilder as it is the default for bit-buildpackage but it fails [11:22] jamespage: is there more needed than "DIST=yakkety git-pbuilder create" followed by "DIST=yakkety git-buildpackage"? [11:22] what it runs into is: [11:22] /usr/include/linux/ethtool.h:1321:18: error: 'INT_MAX' undeclared (first use in this function) [11:22] return speed <= INT_MAX || speed == SPEED_UNKNOWN; [11:24] cpaelzer, build the source packaging using gbp-buildpackage -S [11:24] and then use whatever you like to build the resuting source package [11:25] ok usually sbuild and into adt - will try thanks [11:25] that's what I generally do [11:25] perfect, hat is all I wanted to know [11:25] jamespage++ [11:27] hello, I installed dnsmasq, but the /etc/dnsmasq.conf is ignored [11:27] It looks like NetworkManager starts the dnsqmasq service [11:27] right [11:27] and all settings are ignored [11:27] network manager controls dns masq by default [11:27] so /etc/dnsmasq.conf is ignored as that is not dnsmasq [11:27] ok, so I should uninstall network manager? [11:27] you'll find /etc/NetworkManager/dnsmasq.d is where the config is [11:27] ah, ok [11:28] hmm, there is no dnsmasq.d dir inside NetworkManager [11:31] ps -ef | grep dnsmasq will show you where it's looking at config wise [11:32] ok, brb, reboot [11:56] jamespage: one point for consistency, driving it in the background through a source packaging build and sbuild fails at exactly the same spot [11:57] jamespage: but I think I let that as it is until Monday, no need to focus too much today [11:57] rbasak: My first comment on that in my draft is to CC this to the server list because that's where the server team and other nginx-related things have landed in the past. My second comment is that ultimately they may wish to coordinate with, say, me or hte Server Team on the version bump coordination. That said, you're right, "Ship it inside Ubuntu" is a valid statement, but then we fall into the lack of a build system - they have to compile [11:57] it themselves, then ship the .so [11:57] which is *not* a sane way to go about it 'cause htey then need to support each individual arch [11:58] rbasak: it's also a further reason I"ve not merged from Debian - "experimental" dynamic module support is questionable at how it would be executed for third party binary packages === spammy is now known as Guest86283 === Guest86283 is now known as spammy [11:59] rbasak: and a two paragraph description of the dev cycle and that we can't guarantee a version will be in an Ubuntu release right until it's released [12:00] rbasak: my feeling is "We cannot reasonably commit during a development cycle to a release unless we are already tracking a Stable branch, which is partly dependent on Debian." [12:00] jamespage: I think that is my yakkety env - ust found bug 1592930 [12:00] bug 1592930 in linux (Ubuntu) "failures building userspace packages that include ethtool.h" [Medium,Triaged] https://launchpad.net/bugs/1592930 [12:00] because Debian's having of mainline was the evil problem we had the Xenial cycle [12:00] and the poor timing of 1.10.x [12:01] rbasak: if I may ask, what're your thoughts? [12:03] teward: good idea to Cc the server list. Or even suggest to completely move the thread there. [12:04] teward: why do they have to compile it themselves? He said that it's open source. [12:04] ikonia, works [12:05] teward: I think we're beholden to upstream too much to make that sort of commitment. We're not in a position to keep it. [12:05] Aison: super [12:05] ps fax [12:06] teward: also, if a mere version bump is enough to change the signature, then that precludes us from doing micro release updates. I don't think we can commit to that. [12:06] rbasak: to compile a dynamic module you have to compile it alongside the nginx package and extract the built .so for it [12:06] next question: I search a tool/webapp for my NAS where I can connect with my mobile phone and push files to it [12:06] rbasak: which is a poor way to do it, but that's the existing system [12:07] pinba-engine-mysql is packaged and does something similar. [12:07] It is hacky, but it has been done. [12:07] rbasak: I think this needs further discussion [12:07] teward: agreed :) [12:07] In pinba's case, the security team kindly rebuild it every time MySQL is MRE bumped. [12:08] It's not ideal, because an FTBFS in pinba will potentially hold up a MySQL security update. I suspect that the security team would choose to break pinba in this case. [12:08] mhm [12:08] But that applies to any ABI break in any package receiving a security update in the general case. [12:09] rbasak: I think I also need clarifications on what they mean by signature - that is if they mean upstream 1.10.0 as a signature or 1.10.0-foo-bar-baz as the whole signature [12:09] rbasak: because we *will* have bugfixes with it being an LTS [12:09] maybe not whole version bumps [12:09] but potentially enough to break things [12:09] Yep [12:09] over the course of LTS [12:09] I don't think we should close the door to micro version bumps either though. [12:09] rbasak: can we loop in the SEcurity team (cc: sarnold) for opinions? [12:10] Sure [12:10] as well [12:10] rbasak: because if they say "A security release which is necessary will break this, so we can't make this kind of commitment" will trump both of our viewpoints [12:10] If we agree that we're not closing the door to micro version bumps, then your question is moot I think. [12:10] unless they want to do the hacky method [12:10] rbasak: I try and avoid micro version bumps [12:10] I do want to encourage them to maintain their module in universe or backports though, if that is possible. [12:11] because we're on stable, the only version bumps are security or major bugfixes [12:11] which we can reverse-apply as we already do with cherrypicked patches [12:11] rbasak: for an in-development release version pinning is out the door [12:11] and that's by the nature of the devel cycle [12:12] the only Ubuntu version bumps outside the cherrypicked patches would be to fix major issues in the packaging [12:12] which so far isn't the case [12:12] rbasak: you're right though, we can't close the door to micro-revisions [12:17] rbasak: perahps you should write the reply? [12:17] i've not had any coffee yet [12:18] so my replies are not well worded :/ [12:27] jamespage, beisner, horizon 2:9.0.1-0ubuntu2~cloud0 is ready to promote to mitaka-proposed when you get a moment === zul_ is now known as zul === alexisb-afk is now known as alexisb [14:52] rbasak: thanks for replying to that - far better than what I could have written [14:54] np === alexisb is now known as alexisb-afk [15:43] how to change the time of ask a password when use sudo command? === degorenko is now known as _degorenko|afk [16:15] jak2000, man sudo ... have a look at the -k option [16:34] ciao ragazzi, per caso c'è qualcuno che ne sa abbastanza di ubuntu server?... ho per sbaglio sovrascritto dei file in /srv, ho la lista e tutto... se qualcuno se ne intende e può dargli un attimo un'occhiate gliene sarei grato... [16:38] !it [16:38] Vai su #ubuntu-it se vuoi parlare in italiano, in questo canale usiamo solo l'inglese. Grazie! (per entrare, scrivi « /join #ubuntu-it » senza virgolette) [16:39] I made a mistake and I launched a sync amazon s3 in the wrong direction: aws s3 sync remote local [16:39] for the / srv, I immediately blocked the process, attach files that have changed. [16:39] Back up s3 was the previous day. [16:39] What can I do? If restart does not work nothing or no problem for this file? [16:39] I Use ubuntu 14.14 [16:39] This is list of file http://txt.do/53uip === alexisb-afk is now known as alexisb === devil is now known as Guest24542 [19:23] hey all good afternoon, I have a fresh installation of Ubuntu 16.04 LTS server running but it's consuming over 1GB of ram [19:23] anyone experienced this before [19:24] I found this bug but refers to the unity desktop:https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1572801 [19:24] Launchpad bug 1572801 in linux (Ubuntu Xenial) "Ubuntu 16.04 Unity desktop uses much more ram than Ubuntu 15.10" [High,Confirmed] [19:24] wondering if it's related [19:26] it's nearly useless to talk about "ram consumed" without far far more data [19:27] linux by default aims to use nearly all the memory nearly all the time, because unused memory is wasted memory [19:27] so it caches all that it can [19:28] hmm yeah I know.. but the box shows 1.3 GB of used memory [19:28] not in cache/buffers [19:28] vanilla install [19:28] http://www.linuxatemyram.com/ [19:28] that seems high to me [19:29] jge_: you've got a few different tools to use to investigate; you can use e.g. top's M sort-by-memory command to find out which processes are using the most -- pay more attention to the RSS column than the others [19:30] jge_: since processes share memory of shared libraries they have in common, it's also worth using the smem tool to investigate which processes have 'unique' memory allocated; smem allocates parts of shared memory to every process that's using it to try to give a different view -- it's not more or less accurate, but it may be more useful if you're looking for services to stop [19:32] got it [19:34] strange, since I've already looked into htop, sort by memory, and looked at the RES column, stopped all services I suspect were using it and still pegged at 1.3 [19:35] I'm going to spin up a new vm, fresh install same specs and see if it happens again [19:35] this time without installing anything [20:01] jge_: you wouldn't have X running on that "server"? [20:05] hi all [20:06] i got issue with ssh key that whenever i rebooted the server, it refused my key, but once i had logged in using password from the console of the server, it is ok again [20:07] jelly: nope [20:17] nme_: sounds a bit like you may have used home directory encryption [20:17] nme_: .. then your ~/.ssh/authorized_keys might not be available to the server until after you've authenticated and the filesystem is decrypted [20:18] yeah, i got encryption enabled [20:18] anyway to have encryption on the home directory and at the same time ssh is limited to keys only ? [20:20] i've never investigated but there's probably some mechanism to have the authorized_keys file visible when you're not logged in.. maybe in the underlying mount? [20:21] probably have to update it during login and when not logged in [20:21] oh, no it won't work [20:21] you just cannot use keys at all [20:21] http://unix.stackexchange.com/questions/174427/encfs-automount-and-authorized-keys [20:21] cause you need your password to unlock the encryption [20:22] https://help.ubuntu.com/community/EncryptedHome [20:22] might help, dunno if it's a real work around, or usable work around [20:22] to symlink it outside the space [20:43] i tried to put in under / directly [20:43] not working too :( [20:46] thanks for the suggestion [20:46] will try it tomorrow, too late here, going to sleep [20:46] thanks guys [20:56] coreycb can you review and push my ci fixes for: lp:~ddellav/ubuntu/+source/aodh, lp:~ddellav/ubuntu/+source/keystone, lp:~ddellav/ubuntu/+source/neutron please? [20:56] ddellav, sure [20:57] ddellav, those all for newton? [20:57] coreycb yes [21:06] ddellav, I've pushed those, thanks [21:13] coreycb ty, I will trigger the rebuilds to make sure [22:36] so installed a fresh copy of ubuntu server 16.04 and same problem, memory used reported to be 1.3G out of 2G [22:37] when I do a top, look at the RES column nothing in particular stands out that could be the cause of this high memory usage [22:38] this is vmware virtual machine [22:38] so strange.. [22:44] jge_: could you pastebin the top output sorted by memory? I'm curious now :) [22:47] sarnold: sure thing, I just rebooted the box so it went down to about 70MB but it will creep up back again [22:47] want me to paste that output now or wait till it goes up [22:48] ooh good idea, maybe a before/after would be useful [22:48] ok let me get a paste going, one sec [22:49] sarnold: ok, here it is: http://pastie.org/pastes/10889454/text === Bae is now known as GameOfTrolls [23:37] jge_: did you get an after? [23:39] I am new to unbuntu server - what are some areas to work with [23:46] mgriffin and sarnold: just went back to 1.3GB again, here's the paste: http://pastie.org/10889473 [23:48] jge_: zounds. [23:48] jge_: hmm. that looks like it's sorted by pid, not by memory use [23:49] yeah I just did a top -n 1 :( [23:49] let me get a paste going sorted by memory [23:51] should I do sorted by RES or %MEM [23:51] RES is probably most useful [23:52] roger [23:54] ok here's the new one [23:54] http://pastie.org/10889475 [23:57] have to step out for some time, be back later - leaving computer on so please pm or mention if anything :D [23:58] jge_: does df show /dev/shm usage? [23:59] (or any tmpfs)