[01:39] caliculk: hi, I don't remember it being excluded on purpose [01:40] --with-shared-modules=idmap_rid,idmap_ad,idmap_adex,idmap_hash,idmap_ldap,idmap_tdb2,vfs_dfs_samba4,auth_samba4 \ [01:41] kstenerud: have you tried the setup script with the packages from timo's ppa? [01:41] kstenerud: I got that thing to crash in many different ways, this is just one that we are trying to fix :) [01:58] ahasenack, is there any way (besides building from source) to possibly add it back in? [01:58] otherwise, ill open up a bug ticket because something is off [02:07] I think a bug is in order [02:07] and probably a debian one as well [02:08] as we take the samba package mostly from them, just adding some touches [02:09] I did a quick search in their bug database, and found nothing about zfs in the samba package [02:09] caliculk: can you elaborate on what starts working once this module is in place? [02:10] it's best to elaborate in the bug, though [02:12] ahasenack: zfs doesn't use the posix-ish acls that most of linux uses, they use the nfsv4 acls instead [02:12] ahasenack: this module looks like it's a way for samba to use the nfsv4 acls on zfs backed storage rather than the posix acls that might have worked elsewhere [02:20] sure, but i dont think that is any reason to exclude the module. it should be up to the sysadmin to enable or disabke the module in daily use [02:25] ive spent three days trying to figure this out so i am glad im not insane when looking at this and it wasnt some simple fix lol [02:25] sorry about grammar, dont have autocorrecton this device === ahasenack is now known as Guest4879 === kirkland is now known as Guest54596 === jrib is now known as Guest43012 === tyhicks is now known as Guest8831 [02:41] sarnold, and (i guess the other user netsplit) the zfsacl module is described here https://www.samba.org/samba/docs/current/man-html/vfs_zfsacl.8.html [02:42] but it is a eay to have acls properly implemented with zfs === Frickelpit is now known as Guest95045 === phunyguy is now known as Guest34406 [03:52] Is this: https://bugs.launchpad.net/ubuntu/+filebug supposed to create a bug or link to a wiki, because the "Report a bug" text is really deceiving and not cool if it is supposed to link to a wiki. [03:52] I am trying to launch this from here fwiw: https://bugs.launchpad.net/ubuntu [03:54] I keep getting forwarded to this (https://help.ubuntu.com/community/ReportingBugs) page when I click that +filebug link [04:31] caliculk: ugh :/ [04:31] I thought that thing had some intelligence to only redirect you once [04:31] Nope :/ just keeps doing it. [04:31] caliculk: how about this? https://bugs.launchpad.net/ubuntu/+source/samba/+filebug [04:31] Not sure if you are in #ubuntu but you can see my complaints there but... ultimately, it should be called "how to file a bug" not, "file a bug". [04:32] Yeah, I eventually found that... I filed the bug: https://bugs.launchpad.net/ubuntu/+source/samba/+bug/1788776 [04:32] Launchpad bug 1788776 in samba (Ubuntu) "Ubuntu 18.04 samba/samb-vfs-module not compiled/built --with-acl-support or zfsacl vfs module" [Undecided,New] [04:32] oh good [04:32] But it took me more then ten minutes, which is unreasonable. :D [04:33] There should also be a section in the wiki on how to file a bug via the web interface properly, not via the CLI interface. :) [04:33] beautiful bug report though :) [04:33] Yeah except the formatting... is there a way to put that into code blocks or something so it isn't just plain text? [04:34] nope :( [04:34] Well... thats not helpful either haha [04:39] https://bugs.launchpad.net/ubuntu/+filebug?no-redirect [04:39] there we go [04:39] I finally found the thing [04:41] See that took too long haha :D [04:42] YES [04:42] yes it did :) [04:42] For what it is worth, I didn't mention the lack of --with-acl-support in smbd -b, but that is also an issue, and I am not sure if that is why the zfsacl is not being included, but if it is supposed to be enabled by default (according to samba dev/helper - someone that wiki edit rights - it is) then I don't see why it is also being turned off when building the package. That also seems like a bug. [04:43] But, I put them in the same bug report because it seems like they are linked. [04:43] Which apparently, breaks rule #8 :D [04:44] caliculk: nono, this is perfect [04:44] caliculk: it *might8 be related, you spotted it, and mention it as a possibility [04:44] maybe it is, maybe it isn't, but if it's related it might save someone else an afternoon [04:44] I also need to stop my run on sentences in here. [04:45] (hopefully half an hour, but you know how things go) [04:45] Considering it took me three freaking days to finally realize that ubuntu was the cause. I even tried upgradeing to a PPA with 4.8.4 but that completely broke samba. :/ [04:46] Speaking of which, how can I close this? https://bugs.launchpad.net/ubuntu/+source/logwatch/+bug/1010625 Because... I don't think this is going to be worked on any longer. :D [04:46] Launchpad bug 1010625 in logwatch (Ubuntu) "named logs are not being reported in logwatch" [Medium,Confirmed] [04:47] caliculk: here we go, that #8 is supposed to stop bugs like this :) https://bugs.launchpad.net/bugs/1787564 [04:47] Launchpad bug 1785380 in vlc (Ubuntu) "duplicate for #1787564 The whole system freezes when using VLC in fullscreen mode, on X11" [Undecided,Confirmed] [04:48] Yeah... :) [04:48] I am surprised it isn't just closed. :) [07:33] Hi, I'm looking for someone who can delete a wiki page on the ubuntu wiki. This is my personal page and there is some personal data like name, email. I no longer have access to edit any page. https://wiki.ubuntu.com/Raybuntu [07:33] Thank you [08:19] hi, i've used this https://help.ubuntu.com/lts/serverguide/sssd-ad.html.en to get set up with active directory. samba and kerberos seem to be working fine but SSSD isn't updating /etc/passwd, can anyone help? [09:25] hi, can someone help me with active directory integration? === Guest4879 is now known as ahasenack [10:53] good morning [11:18] another victim of https://bugs.launchpad.net/subiquity/+bug/1783129 [11:19] Launchpad bug 1783129 in livecd-rootfs (Ubuntu) "Only "main" component enabled after install" [High,In progress] [11:19] sounds like feature, not a bug :) [11:21] let's call it a regression then :) [11:21] the other components are not even listed for you to uncomment [11:24] yeah I was just kidding a bit ;) [12:04] Hello there, I installed a basic bionic server with debootstrap for an nspawn container. Now I noticed that there are no manpages. Should I install the basic server task? Manpages seem a to be a basic thing, I thought even debootstrap would install that [12:06] Maybe most people would not use a container that way, but I was hoping to use the container pretty much as a normal system, for example with sshd, so for a normal working environment it would be nice to have the manpages [12:11] I'm receiving the error " multipathd: uevent trigger error" in syslog [12:11] any inputs on how to resolve it [12:26] * ahasenack doesn't know [12:33] l4m8d4: it's pulled in by ubuntu-standard. I usually add ubuntu-minimal and ubuntu-standard for debootstrap'd installations [12:36] l4m8d4: question is, do you really need that in the nspawn'd container [12:38] blackflow: I'd like to have it since I am often ssh'd into it and then in need of manpages for a program that is maybe only installed there [12:39] l4m8d4: you can just install "manpages" package then. looking at my installation scripts now, I'm actually not pulling in ubuntu-standard, only -minimal. -standard has some overhead I don't need. [12:45] cpaelzer: do you know if it's ok/common for a python package to depend on *both* python3.6 and python3.7? [12:48] ahasenack: doesn't sound right [12:53] blackflow: Okay, ubuntu-minimal was already installed by default, ubuntu-standard was not, it seems I will install ubuntu-standard, since it has the manpages, bash-completion, the relase upgrader and other things [12:53] Thanks for the usggestion [12:53] I see what you mean though, with things like telnet, rsync etc. that not everyone needs it has a fair amount of "bloat" in it [12:54] right. [12:56] Or stuff like ntfs-3g or hdparm, which im definitely not gonna use, but yeah, I guess 100mb more or less won't matter [12:58] those are recommended though, not hard deps [13:01] Right, I would install it that way, though, because I want some of the recommended ones^^ [13:01] you can always pull in those directly. you don't have to install ubuntu-* metapackages [13:11] Guys, question. with apt-daily.timer/service now being systemd timer/service, what's recommended for mailing lists of available upgrades, still apticron? [13:57] ahasenack: The problem is that if any function returns a status other than OK, it calls abort(). In this case, if it can't find any one of the config files it looks for, it crashes. There might be other ways in which the init function returns a bad status as well :/ [13:58] So we can't be sure what is triggering for him specifically. I ca nask him what his syslog says, which could give a clue (it does spit out syslogs for configs it couldn't find) [13:59] Also, calling named-pkcs11 without the -f -u bind gets "further" in that it seems to generate some configs that it wouldn't with those arguments present, such that it gets a little further if you call it alone first (get a crash) and then with the arguments (crash in a different place) [14:01] sounds fragile :( [14:01] Yeah :( [14:02] Every RUNTIME_CHECK() and PK11_FATALCHECK() in the code is a return code test that calls abort() if it doesn't like it. And there are LOTS of those [14:03] security by fear [14:03] assert all the things [14:03] grep -r RUNTIME_CHECK *|wc -l [14:03] 733 [14:04] I'm surprised this thing runs at all [14:04] so, you couldn't get it to finish setup with or without the patch, is that the tl;dr? [14:04] yes [14:04] ok, do you have a ppa of your own? [14:04] because I think I'm triggering a different abort() than he was [14:04] for this? [14:04] yup [14:04] https://launchpad.net/~kstenerud/+archive/ubuntu/bind9-rtld-deepbind-1769440 [14:04] I'd like to give it a try, since I (think) reproduced the bug a while ago [14:04] but I also saw it crashing all over the place when things were a bit different [14:05] kstenerud: would you like to chance pace to some C coding? It's a backport of a patch [14:06] sure :) [14:06] but it involves talloc() [14:06] although the patch isn't about talloc() [14:06] the bad news is that freeipa is involved again [14:06] lol [14:06] but I can set a server up where it works, I think I used centos [14:07] it's the client that broke [14:07] sssd in this case [14:07] https://bugs.launchpad.net/ubuntu/+source/sssd/+bug/1775636 see what you think [14:07] Launchpad bug 1775636 in sssd (Ubuntu) "sss_ssh_authorizedkeys fails with: Error looking up public keys when client cert present in IPA" [Undecided,Confirmed] [14:07] the upstream issue is https://pagure.io/SSSD/sssd/issue/2977 [14:13] ahasenack: so basically this patch? https://pagure.io/SSSD/sssd/c/60787fb44924e84a0c7ddfe9d5e62e64ea1edcd1 [14:13] kstenerud: yes [14:14] kstenerud: iirc parse_cert_verify_opts() changed [14:15] Did you repro it as per https://bugzilla.redhat.com/show_bug.cgi?id=1372042 [14:15] bugzilla.redhat.com bug 1372042 in sssd "ssh with Smartcards - skip invalid certificates" [Medium,Closed: currentrelease] [14:15] yes [14:15] ok [14:15] but I used centos or fedora to install freeipa [14:16] centos7, I think I still have the vm [14:16] We should really start doing ansible scripts for repro cases :P [14:17] comment #8 shows the results, not how I did the setup [14:17] (in the lp bug, not the rh one) [14:42] if I want to remove custom services on Ubuntu 14, what do you guys think I should do? [14:43] I was looking at running: [14:43] 1) sudo service stop [14:44] 2) sudo update-rc.d -f remove [14:44] 3) rm /etc/init.d/ [14:44] 4) rm -R /var/lib/ [14:44] am I missing something else? [14:45] DammitJim: depends which service it is, but 1) followed by 2) would work for system V services [14:45] DammitJim: for upstart managed services, this would be a little different [14:45] yeah, this is not systemd [14:45] Ubuntu 14.04 LTS [14:45] DammitJim: correct, 14.04 used upstart [14:46] s/used/uses/ [14:46] right... so, are you saying I need to do something different because it's upstart? [14:49] DammitJim: uninstalling those services with apt will probably do this easy for you [14:49] DammitJim: and keep in mind that 14.04 won't be supported much longer - it'd be a good idea to upgrade soon [14:50] RoyK, they are custom [14:50] like I created multiple instances of tomcat [14:51] RoyK, yes, that's part of the upgrade... I don't want to have to worry about these services that have been phased out [14:51] when I go to 18.04 LTS [14:51] DammitJim: for upstart driven service, you'd do: sudo service $SVC stop; echo manual | sudo tee -a /etc/init/$SVC.override [14:51] also, keep in mind that it may be quicker to just install 18.04 on a new machine and then migrate the stuff fro the old one [14:51] ah, interesting... what does the echo manual do? [14:51] quicker and indeed cleaner [14:52] RoyK, yes! [14:52] sdeziel, I was going to just delete that file in /etc/init.d [14:53] wait, /etc/init ? let me look [14:53] that service doesn't exist in /etc/init for some reason [14:54] DammitJim: with sysv scripts, /etc/init.d is just a place to put the scripts and then have the system symlink to them from /etc/rcX.d where X is the run level (normally 2 on 14.04 IIRC) [14:54] oh ok, so I probably need to remove those symlinks [14:54] yes [14:55] update-rc.d does just that [14:55] yeah, that's what I thought [14:55] ok, thanks for your help guys [14:55] I think I have a plan to execute [14:58] ap_if.ifconfig() [14:58] oops [15:01] this is what I mean by ipa being nitpicky [15:01] ipapython.admintool: ERROR DNS zone example.com. already exists in DNS and is handled by server(s): a.iana-servers.net., b.iana-servers.net. [15:01] what a pain [15:03] * ahasenack uses --allow-zone-overlap [15:12] kstenerud: ugh, I think the freeipa version changed too much in cosmic since that bug was opened [15:12] the bind bug, I mean [15:13] it just configured apache with an encrypted private key, so apache won't start because it doesn't know the passphpase, nor does it prompt for it [15:15] did something happen to feature freeze? [15:16] dpb1: probably a better question for #ubuntu-release [15:17] * dpb1 nods [15:28] blackflow: Thanks for your suggestions, I installed ubuntu-standard, I think it's an OK compromise for me for now [15:28] l4m8d4: yw === jelly-home is now known as jelly [16:47] back [17:02] kstenerud: heh, about the apache startup issue with freeipa: https://bugzilla.redhat.com/show_bug.cgi?id=1591703#c3 [17:02] bugzilla.redhat.com bug 1591703 in freeipa "ipa-server-install fails to start the httpd server during installation and ends with error." [Unspecified,New] [17:03] * ahasenack troubleshoots further [17:04] lovely [17:21] ahasenack, you ended up disconnecting yesterday after the netsplit, to sum up what the zfsacl module does, it provides interoperability between acls on linux, zfs, and nsf. ZFS uses nsfv4 acls not posix type stuff. But I ultimately believe it to be a bug, so: https://bugs.launchpad.net/ubuntu/+source/samba/+bug/1788776 [17:21] Launchpad bug 1788776 in samba (Ubuntu) "Ubuntu 18.04 samba/samb-vfs-module not compiled/built --with-acl-support or zfsacl vfs module" [Undecided,New] [17:21] I saw it, thanks [17:22] so even when you run zfs set acltype=posixacl you need that vfs module? [17:22] Thats what I had when running freenas, and the vfs module was enabled. [17:23] You can easily still access shares when not running the vfs module, but it isn't full compatibility. [17:23] even with acltype=posixacl? [17:25] I believe so, because extended attributes would still apply: "To obtain the best performance when setting posixacl users are strongly encouraged to set the xattr=sa property" [17:27] that's ok, it's a performance tip [17:27] I'm trying to understand if a configuration change, like setting acltype=posixacl, isn't enough to fix the issue, or if a new module is really needed [17:27] I also don't understand how nfsv4acls intersect with samba, but I can read about that [17:28] kstenerud: ok, fixed. freeipa really wants $(hostname) (and not just $(hostname -f) to return the fqdn [17:28] it used to check for that iirc, but anyway [17:28] That I can't answer completely, I am not knowledgeable enough to answer that. :/ I just know from the documentation I have been reading, it appears that the module is necessary/useful for ZFS, but, by default it is not available (afaict) [17:29] so you haven't seen issues yet without it? I assume you have a samba deployment exporting a share that sits on zfs [17:30] Correct, it's just whether I enable or disable that module it becomes an issue. Granted, I haven't used the share that much since I don't want to write or do anything with it until I had gotten to the bottom of why the module was missing first. [17:30] I still have an entire backup to transfer back on the zfs filesystem that hasn't been completed yet either until I could confirm 100% [17:31] ok, getting an actual scenario where the module fixes a problem will help to determine how useful it is [17:31] oh, samba migrated [17:31] happy dance [17:32] This bug was fixed in the package samba - 2:4.8.4+dfsg-2ubuntu1 [17:32] nice! [17:34] Hm, I tried upgrading to 4.8.4 via a PPA (since 4.8.4 is not officially available on 18.04.1, and I don't feel comfortable installing samba from source), did not run into much luck getting it to start. [17:35] Unless I missed something, I couldn't find it in a specific distro list on any of the additional options "verses" available. [17:36] it just landed in cosmic, not 18.04.x [17:36] bionic won't get it, since it's a new major version [17:36] Yeah :/ [17:37] I guess it could be available in backports eventually [17:37] But... backports [17:41] I guess, best thing to do would be to file an issue here then? https://launchpad.net/bionic-backports ? [17:44] you can try, I don't know how responsive backports are [17:45] I dealt with one last year, and had to convert it to an SRU instead, which was a lot of work, but was done at least [17:45] for cosmic, there may still be time. We just entered feature freeze (a few hours ago), so a new module would be a new feature and would have to be explained carefully [17:45] that's why I need that scenario of what this module is fixing [17:45] to be able to get a "feature freeze exception", if it's worth it [17:46] specially since it would add to our delta with debian, who is not shipping that module either as far as I can tell [17:47] Well, unfortunately, I can't get someone in #samba to explain to me in a more context aware scenario (posted the bug report there asking for guidance). I have reached out to some other people I know that are running ZFS on Ubuntu, to see if they could assist as well. But haven't heard back yet. [17:47] ok [17:51] yeah, samba in debian doesn't have a zfs module either, as expected [17:51] jsut checked [17:53] kstenerud: ok, hm, the freeipa installation finished out of the box now, using cosmic's bind9 [17:54] I'll comment in the bug [17:54] bionic next [17:54] but good that it looks fixed in cosmic [17:55] ah, [17:55] not for long [17:55] Aug 24 17:31:41 cosmic-freeipa named-pkcs11[6550]: ../../../lib/dns-pkcs11/view.c:968: REQUIRE(view->zonetable != ((void *)0)) failed, back trace [17:55] heh [17:55] same cr*p [17:55] oh well, can test the fix [17:56] kstenerud: btw, you skipped a release number again in your versioning of your bind9 package in the ppa [17:57] you have -3ubuntu3~ppa1 when the cosmic package is 3ubuntu1 [17:59] I have to go for now, but I will update you on if I hear back from anyone about the module and it's specific benefits. My environment may just not be sufficient (use ubuntu-server in a one person apartment) to test whether the acl has a huge advantage with that enabled. [18:01] ok, thanks for checking [18:01] I'm a zfs user myself, and I also have a home-built NAS with samba on top of zfs [18:02] but just one user [18:02] that doesn't really exercise ACLs :) [18:03] The best I can tell it MAY help with Volume Shadow Service, but, yeah... sounds good. [18:09] kstenerud: yeah, the patch worked [18:09] no crash this time [19:13] ahasenack: Can you tell me how you configured the vm running freeipa? I can't seem to get it onto any accessible network :/ [19:14] kstenerud: the vm itself is using the default libvirt network, 192.168.122.0/24 [19:14] it's on virbr0 iirc [19:14] hmm yeah that's the same I'm running. Odd [19:14] kstenerud: it's a nat'ed network, and it has dhcp [19:14] ok [19:14] then /etc/hostname has an fqdn, with 3 "pieces" [19:14] I called it cosmic-freeipa.example.com [19:14] I also added that name to /etc/hosts with the real ip the vm got, not 127.0.x.x [19:15] then I called ipa-server-install with that dns option to allow zone overlap, you can find it with --help [19:15] "hostname" has to return the fqdn (that's how freeipa likes it), as does "hostname -f" [19:15] oh, and I rebooted after fixing the hostname like this [19:16] My eth0 on the vm isn't getting an IP address [19:16] that would be a problem :) [19:16] check /etc/netplan/* [19:16] ah, that's the other thing I did, I changed that to static, just to avoid surprises [19:16] but I used the IP I got via dhcp the first time [19:16] It's centos (for freeipa) so no netplan [19:16] check if your libvirt network has dhcp enabled [19:17] in virt-manager, go to edit -> connection details -> network [19:17] It's a headless install [19:17] but on your laptop/desktop? [19:17] oh hang on [19:17] where is the vm running? [19:18] I have it under virt-manager. There's no edit menu [19:18] it's in virt-manager's window, not the vm window [19:19] ok so virtual network default has a dhcp range [19:19] 192.168.122.x [19:20] well, try just running dhclient as root in the vm, see if it grabs an ip [19:20] if it does, then it's a centos configuration issue [19:20] also, check the nic in the vm window [19:21] kstenerud: do that first actually ^ [19:21] see if it's connected to the network you expect [19:21] oh hah ok running dhclient manually works :P [19:21] guess the minimal install is REALLY minimal [19:21] surprising, but could be [19:25] Hello [19:26] I am (possibly) having some RAM difficulties [19:26] I am running a home server with a quad core 3rd gen intel cpu with 4gb of ram [19:27] My 'physical memory usage' is quite frequently very high and I feel like downloads and bash speed over ssh slows down [19:27] the thing is, my 'real memory' is under control', surprisingly under 1gb of use [19:28] what does free -m show [19:28] !pastebinit | Jester316 [19:28] Jester316: pastebinit is the command-line equivalent of !pastebin - Command output, or other text can be redirected to pastebinit, which then reports an URL containing the output - To use pastebinit, install the « pastebinit » package from a package manager - Simple usage: command | pastebinit [19:29] dpb1, https://pastebin.com/WyxHPLuk [19:30] Right now it's alright [19:31] I'm not actually doing much, but yesterday when I was downloading and streaming a movie over local network, the stream would shut off, download speed was lower than usual and bash over ssh was sluggish [19:32] I did have a lot of cache at that point and clearing it improved everything, as well as rebooting, which I did the first time round [19:33] dpb1, https://i.imgur.com/QDqzk6W.png [19:33] that looks like a totally underutilized memory. [19:34] hmmm [19:34] physical mem will always be near fully used. unused physical memory is wasted memory. [19:34] see https://www.linuxatemyram.com/ [19:35] the caches are volatile. if an app allocates memory for its used, the kernel will happily drop the caches and give that space to the app. also, streaming issues you mention, including laggy ssh, is not indicative of memory issues, but of network issues. [19:36] alright, thanks blackflow [19:36] that was a concise and informative link, good read [19:37] I'll come back when/if I experience sluggishness again without a reason [19:37] good day! [19:37] good luck === terminal1 is now known as terminalator === saint_ is now known as saint_away [22:42] ahasenack, the person that I was talking to got back to me. As far as I can tell the biggest benefit is to emulate the Windows permission set properly with the vfs module and the --with-acl-support options. [22:43] That seems to be the biggest downside (so for a person in a 100+ company, it might make more sense) but for a one person home use, obviously it would not. However, if I were to share this out with my roommates, then it would make sense to want to include the zfsacl permission set. [22:45] Basically, it just provides proper windows support with zfs. [22:46] I am not sure if that is reason enough to ask for a feature freeze exception though [22:52] it's probably useful to know if including it can somehow break existing systems [22:53] (that's usually an important factor for freeze exceptions) [22:54] I don't believe it would, but I am not sure in the slightest on where to begin on testing that to get this included,. [22:55] Though the --with-acl-support option on Samba is not specific to ZFS, it doesn't allow samba to play nicely with ext4 or other filesystems either. [22:56] Basiclly, if I use FreeNAS (where this module is working and existing), the biggest plus is that Samba can then expose those shares with ACLs that Windows understands without any additional work required.