[01:20] <CarlenWhite> Is the ubuntu repository acting slow for anyone else?
[01:23] <sarnold> CarlenWhite: hrm, I don't see anything particularly surprising on our bandwitdh pages..
[01:23] <sarnold> CarlenWhite: what specifically is going slow?
[01:26] <CarlenWhite> Trying to install a package but there might be some interference outside the scope of my desktop and router. Wasn't expecting slow network and browsing normally seems fine. I'll have to poke around.
[01:29] <CarlenWhite> I guess the Chromecast is active...?
[01:31] <sarnold> netflix apparently says 4k 60hz video takes ~25Mbps
[01:32] <sarnold> so depending uponm what you're doing with chromecast, maybe it is or maybe it isn't a factor..
[01:32] <CarlenWhite> We rural. Max we get here is 15Mbps down over DSL.
[01:34] <sarnold> aha, then chromecasting something might have a very real impact on what you can get from the ubuntu archive servers
[01:34] <CarlenWhite> I think my QoS setting might be acting weird. A bolt on page for DD-WRT is reporting 500KiB as 30% used capacity.
[01:36] <CarlenWhite> Actually after I said that it's down 130KiB. I'm getting conflicting reports on different pages of my router.
[01:39] <CarlenWhite> As much as I love DD-WRT, it really could use a built in method to help figure out which device is being overzealous.
[01:39] <CarlenWhite> Then again, any router would be better with that information.
[01:41] <sarnold> I'm surprised ddwrt doesn't already have that
[01:41] <sarnold> it almost seems like the first thing that would have been implemented :) hehe
[01:43] <CarlenWhite> You can bolt something on but it hasn't been maintain for a few years and takes ~30s until it collects enough information to show usage and refreshes slowly. Not exactly ideal for finding bursts of when slowdowns happen.
[01:44]  * CarlenWhite is apparently at war with the various internet connected devices in his home.
[01:47] <CarlenWhite> Testing upload speed.................................................. Upload: 0.00 Mbit/s
[01:47] <CarlenWhite> That isn't good.
[01:49] <sarnold> that's enough to make me think it's busted entirely :)
[01:49] <sarnold> try https://fast.com/# -- after the primary test you can click on the upload portion to see your upload speed
[01:50] <CarlenWhite> I guess the speedtest-cli is being weird.
[01:52] <CarlenWhite> Oh I guess I could also open the Home app and see that one of the Chromecasts is indeed being used for Netflix.
[01:52] <CarlenWhite> That could've been a more sane course of action.
[01:52] <CarlenWhite> Probably made my sister's viewing experience become a mess of pixels from trying to figure out what was going on.
[01:56] <CarlenWhite> Well good knowing the QoS settings are doing it's job. It has effectively made my network go at a snails pace for anything that isn't video streaming.
[01:58] <sarnold> CarlenWhite: hah, poor sister.. "why does it look like 64x48 at 2hz?!?" :)
[01:58] <CarlenWhite> Well actually if it worked correctly, she should notice little quality loss.
[11:09] <zzarr> hello!
[11:10] <zzarr> I have installed Ubuntu server on a machine with LVM
[11:12] <zzarr> how do I resize the root partition?
[11:16] <blackflow> zzarr: using your filesystem specific tools (eg. resize2fs for ext*), offline (not booted into that root). don't forget to resize the LV too, after you've resized teh filesystem. Of course, back up all the data first.
[11:17] <zzarr> I meant to resize the LVM partiton
[11:20] <blackflow> zzarr: you need to specify what you mean as "LVM partition"?  The underlying pv? The vg? the LV for the root fs?
[11:21] <zzarr> ohh, sorry, I'm new to LVM's
[11:21] <zzarr> the LV
[11:22] <zzarr> I wish to make /dev/mapper/ubuntu--vg-ubuntu--lv bigger
[11:22] <blackflow> so then exactly what I said :) you first need to resize the fs, then the LV under it
[11:22] <blackflow> oh bigger, then you enlarge the LV first, then the fs on it. To enalrge the LV look into lvextend command
[11:23] <zzarr> I resize the fs first?
[11:23] <blackflow> you enlarge the LV first, then the fs
[11:23] <blackflow> you shrink the fs first, then the LV
[11:23] <zzarr> that's what I thought
[11:23] <zzarr> can I list GV's and LV's?
[11:24] <blackflow> sure. lvscan, vgscan.
[11:25] <zzarr> nice, that's my biggest problem, not knowing the names
[11:25] <blackflow> there are plenty of LVM primers online, if you need one to get through all the concepts and commands ;)
[11:27] <zzarr> yes, I have been reading the official LVM guide, but I felt that I didn't get the hang of it
[11:28] <zzarr> I just realized the machine I got have 4 GB RAM, not 2 GB as I thought
[11:29] <zzarr> success, the root partition is now 60GB instead of 4GB
[11:29] <zzarr> nice, thank you
[11:31] <zzarr> fs resized as well, it's easy when I understand :)
[11:32] <blackflow> zzarr: if you're still learning these concepts, I wholeheartedly suggest you to look into ZFS. it's so advanced and integrates filesystem, raid and volume management, into one technology. it is a pooled FS, so it behaves like LVM when it comes to provisioning PVs.
[11:34] <zzarr> ZFS is nice, but I don't have any interesting hardware to use it on
[11:38] <blackflow> zzarr: you have a server, right? :)
[11:38] <zzarr> yepp, but only one drive (240GB SSD)
[11:39] <blackflow> so? you won't need the raid capabilities, but there's still: pooled fs, snapshots, data checksums, compression.
[11:39] <blackflow> although... zfs on linux still doesn't support TRIM, if that's important to you, so that might be a deal breaker.
[11:40] <zzarr> wouldn't that require a reinstall?
[11:40] <blackflow> it would, yes. plus the installer can't do zfs, so you'd have to debootstrap it from liveusb
[11:40] <zzarr> I need trim
[11:41] <zzarr> I see, I might buy some other hardware to do that with
[11:41] <blackflow> are you sure? modern SSDs do wear leveling just fine. TRIM was important in the past, not so much today.  but okay, if you don't want ZFS, I'm not trying to coerce you :)  just saying ZFS is worth all the trouble to get it under your (important) data ;)
[11:42] <zzarr> by the way, is single system image possible in a cluster of Ubuntu server machines?
[11:43] <zzarr> I'm planing on building a Ceph SAN later
[11:44] <blackflow> not sure what you're asking
[11:44] <blackflow> one SAN-based rootfs shared by multiple compute nodes?
[11:46] <zzarr> yes it will be shared among many nodes
[11:48] <zzarr> I have one of these http://www.banana-pi.org/r2.html it will make my SAN look like a NAS from the outside
[12:00] <blackflow> well... not sure. Me and Ceph are not uh.... on good terms :)
[12:01] <blackflow> *Ceph and I
[12:05] <zzarr> I see, I have not tested it yet, but I have a friend that works with it and he can help me
[12:06] <zzarr> not sure if you answered my question about single system image
[12:06] <zzarr> is it possible in Ubuntu server?
[12:19] <blackflow> zzarr: oh I answered, I'm not sure because I'm not good with Ceph. We don't talk. :)
[12:20] <blackflow> otherwise I don't see why not. if by system image you mean the rootfs, I suppose it can be easily shared in read only mode.
[12:20] <zzarr> single system image have nothin with ceph to do
[12:21] <blackflow> well there are a lot of aspects to SSI, and I'm guessing you're asking about the rootfs (since you mentioned ceph - which is storage, and not, say, IPC shared in teh cluster)
[12:22] <zzarr> single system image is a type of cluster
[12:23] <blackflow> yeah, I know. But I don't know how well ubuntu supports it. as for shared rootfs, which I initially commented for, it should be doable.
[12:23] <zzarr> I see
[12:27] <zzarr> single system image is more then just sharing filesystem, it's a cluster which if you log in to it your shell ends up on one node and if you start an application it might end up on another depending on load
[14:33] <jak2000> hi all
[14:33] <jak2000> anyone here have experience with mysql?
[14:34] <blackflow> jak2000: maybe, ask and wait if someone might know. there's also #mysql, if your problem is not ubuntu specific.
[14:34] <jak2000> sorry and escuse me:    always that i restart my server with: shutdown -r now      after restarted if i try access to mysql i get the error: "ERROR 2002 (HY000): Can't connect to local MySQL server through socket '/var/run/mysqld/mysqld.sock' (2)" and i need type: sudo mysqld_safe --tc-heuristic-recover=COMMIT    because i cant connect immediatly, why? (need stop the service before the
[14:34] <jak2000> restart?) any advice?
[14:36] <blackflow> jak2000: is the mysql service running, on boot? which ubuntu is this?
[14:45] <rbasak> jak2000: also, how did you install MySQL? Are you using the package shipped with Ubuntu, or are you using a third party package?
[14:48] <compdoc> an update once broke my mysql. had to mess with the .conf files and put them in the right places
[14:50] <jak2000> installed: apt get install mysql
[14:51] <jak2000> how to check i if is running on boot?
[14:57] <blackflow> jak2000: systemctl status mysql.service     (or whichever the service unit name is)          (I'm asuming Ubuntu 16.04 or newer)
[15:03] <jak2020> blackflow thanks
[15:06] <sandstrom> I installed the package `1.14.0-0ubuntu1` a few weeks ago on 18.04, worked smoothly. Then I spun up a new Ubuntu 18.04 machine today and tried the same thing. Didn't work, because of this error:
[15:06] <sandstrom> The following packages have unmet dependencies:
[15:06] <sandstrom>  nginx-extras : Depends: libnginx-mod-http-auth-pam (= 1.14.0-0ubuntu1) but 1.14.0-0ubuntu1.1 is to be installed
[15:06] <sandstrom> […}
[15:07] <rbasak> jak2020: if you can reproduce, please could you show us (pastebin) /var/log/mysql/error.log straight after a reboot, and straight after a manual start attempt following the reboot? Please identify the time of the reboot so we know where that happened in the log.
[15:07] <sandstrom> Anyone know what has caused this? Seems like the dependencies are somehow pointing at the wrong thing now?
[15:07] <rbasak> jak2020: the "pastebinit" command is handy :)
[15:08] <mybalzitch> rbasak: I do enjoy that command with a few arguments
[15:08] <rbasak> sandstrom: have you tried an "apt update"?
[15:08] <sandstrom> Yes
[15:09] <sandstrom> It works if I change to the package name with extra `.1` at the end
[15:09] <sandstrom> "1.14.0-0ubuntu1.1"
[15:09] <rbasak> Are you confusing package names with package versions?
[15:09] <rbasak> I don't follow what you're doing exactly.
[15:09] <sandstrom> Perhaps :)
[15:09] <rbasak> What command are you typing?
[15:09] <sandstrom> `sudo apt install nginx-extras=1.14.0-0ubuntu1`
[15:09] <rbasak> Why are you forcing the version string there?
[15:10] <rbasak> Why not just "sudo apt install nginx-extras"?
[15:10] <sandstrom> To avoid updates that we aren't aware of
[15:10] <rbasak> Your problem is that you've already picked up something that wants 1.14.0-0ubuntu1.1 - probably you already have an update that you aren't aware of.
[15:10] <sandstrom> This is for a provisioning script, so if we are spinning up many machines over the course of 1-2 months we want them all to run the same version
[15:10] <rbasak> Don't use apt then.
[15:11] <rbasak> If you want them all the same, get all the debs and throw them at dpkg -i directly.
[15:11] <rbasak> Or manage your own repo
[15:11] <sandstrom> I thought apt could handle version-pinning, but perhaps I've misunderstood something
[15:12] <rbasak> It can but only if you don't give it an impossible to fulfil request, which is what you've managed to do.
[15:12] <sandstrom> Hehe :)
[15:12] <rbasak> Generally apt tries to keep you up to date.
[15:12] <sdeziel> with security patches you know :)
[15:12] <rbasak> If you don't want to be up to date against what you're pointing it to then you don't need apt.
[15:12] <sandstrom> I'm fine with allowing minor updates, but would like to ensure that the major isn't bumped
[15:12] <rbasak> See https://wiki.ubuntu.com/StableReleaseUpdates
[15:12] <rbasak> That's what we try to do for stable releases.
[15:13] <rbasak> You can see actual updates as they are published here: https://lists.ubuntu.com/archives/bionic-changes/
[15:13] <rbasak> Or you can enable the security pocket only, disabling the updates pocket. Then you'll get security updates only.
[15:13] <blackflow> apt-get changelog <packagename> is very valuable too. I use that before running upgrades
[15:14] <rbasak> (though security updates are based on the latest SRUs, so if there's a security update in a package that has been SRU'd, you'll end up with SRU + security update)
[15:14] <sandstrom> Thanks for explaining. My primarily knowledge about dependency management and repositories is node packages and ruby gems, where it's fairly straight-forward with version-pinning.
[15:14] <rbasak> It's also fundamentally broken :)
[15:14] <blackflow> sandstrom: yeah but you can't do that with a integrated system like Ubuntu Linux distro is. can't pick'n'choose package versions.
[15:14] <rbasak> The problem comes when you have a complex dependency tree with versioned dependencies
[15:14] <sdeziel> I'd recommend setting up apt-listchanges, will show you the changelog related to the update you are about to pull and ask you to proceed or not
[15:15] <blackflow> if you need that, a custom repo and build from srcdebs is your only sane choice
[15:15] <rbasak> You absolutely can pick'n'choose package versions, but you'll find it extremely difficult to not end up in an impossible situation.
[15:15] <sandstrom> Hehe :) Great input, this is helpful
[15:15] <sandstrom> I'll look at disabling the updates pocket and then drop the versions altogether.
[15:16] <blackflow> rbasak: sure you can but only from list of versions in the repos, not the entire version range form upstrea,
[15:16] <blackflow> *upstram
[15:16] <blackflow> ohgodawfulkeyboardthing
[15:16] <sandstrom> Then we'll just rely on the stable to be just that. And hope that machines provisioned over the next year will work even though they may use slightly different versions of nginx (and other packages)
[15:17] <blackflow> sandstrom: why wouldn't they?
[15:18] <blackflow> the "slightly different" is mostly about security and bugfixes. the base version of a package is baked in for the duration of LTS, withsome rare exceptions
[15:18] <blackflow> but anyway, at any point in time, all your installations would have the same version (if you keep them all up to date)
[15:20] <blackflow> you should really trust the S in LTS, it helps a lot in maintaining the packages in a stable fashion, so your own testing should be minimized (compared to running everything yourself straight from the upstream)
[15:22] <sandstrom> blackflow makes sense!
[15:22] <sandstrom> We're very happy with Ubuntu, it's just that some concepts around versioning strings for the packages are new to me.
[15:22] <sandstrom> It's different from what I'm used to with ruby/node/rust
[15:23] <rbasak> We're much more rigid than the language-specific module communites on stable meaning stable.
[15:23] <blackflow> ruby/node/rust don't have that versioning scheme, because they're the origin
[15:23] <rbasak> We also welcome participation in testing updates before they're released.
[15:24] <rbasak> They appear in the proposed pocket, and if it breaks you and you tell us, we won't release the update.
[15:24] <rbasak> See https://wiki.ubuntu.com/StableReleaseUpdates for all the details.
[15:27] <jak2020> thanks.... all for help
[15:27] <jak2020> rbtask, when have the problem i check the log file, thanks
[17:35] <teward> ahasenack: if you're around do you have the bug number or link for the repositories enablement problem?
[17:35] <teward> (that was present in the ISOs)
[17:45] <powersj> teward, https://bugs.launchpad.net/subiquity/+bug/1783129 that one?
[18:07] <teward> powersj: yep that one, thanks