|
Yes everyone wants it but it's never gonna happen. Arch comes the closest to solving it but if you can't stomach that distro you're pretty much SOL. Maybe dockerhub is the solution you want? 
|
#
?
May 29, 2019 15:32
|
|
|
|
| # ? Apr 23, 2024 10:51 |
|
|
Well that's.....a solution. Really though I haven't had a good opportunity to play with docker much and you did get to me to google it's performance impacts, pretty surprised at how lightweight it is for a lot of things. When I re-do my plex server I might do it with docker just to get some experience in it. Cool stuff.
|
|
#
?
May 29, 2019 16:15
|
|
|
How is stuff in the AUR audited? Is it very susceptible to malware sneaking in?
|
|
#
?
May 29, 2019 18:06
|
|
|
CFox posted:This has been my biggest irritation about linux really. I'd love to install everything from one central repository that is updated frequently and not have to worry about adding a dozen different PPAs but besides possibly the Arch User Repository I don't think there's anything like that in linux land and Arch just ain't my thing. That's what flathub.org is, it's just very limited in what it carries. It's great for stuff like Calibre which needs harmony between about 60 dependencies that your distro may upgrade at random times.
|
|
#
?
May 29, 2019 18:15
|
|
|
apropos man posted:How is stuff in the AUR audited? Is it very susceptible to malware sneaking in? In short, it's not. Malware has been found before. That's sort of the whole thing. Repo grooming isn't free, nor is it rewarding work. So you accept old packages and tighter security, or accept some risk and more user sources material.
|
|
#
?
May 29, 2019 19:14
|
|
|
Craptacular! posted:That's what flathub.org is, it's just very limited in what it carries. It's great for stuff like Calibre which needs harmony between about 60 dependencies that your distro may upgrade at random times.
|
|
#
?
May 29, 2019 21:20
|
|
|
Vulture Culture posted:Snaps would be another option. Fedora devs just twiched: Flatpak Same poo poo, different colour (kinda, almost, not quite, meh).
|
|
#
?
May 29, 2019 22:17
|
|
|
FreeBSD ports and Gentoo portage also comes fairly close - they are basically a collection of makefiles that specify what to fetch, how to compile it, and which other things in the system it depends on; there's also a way to add patch files if any are needed. (FreeBSD also provides binary packages of almost all of these, so the user experience is basically like apt unless you want to dig in.) The spec files that get turned into packages on other (Linux) distros are not that different - but something about the "easily accessible directory structure of makefiles"-approach seems to keep things fairly updated and inclusive. Of course, there are good reasons to not use FreeBSD or Gentoo (and some things just don't run on BSD) - but going from ports to redhat style "the default repositories has everything a conservative kernel developer from five years ago might need" is always a letdown. Computer viking fucked around with this message at 23:34 on May 29, 2019 |
|
#
?
May 29, 2019 23:28
|
|
|
Volguus posted:Fedora devs just twiched: Flatpak I just flattened and repaved my laptop a couple of days ago. The machine I'm typing from now. I'd been running Fedora 29 KDE on it and started about 18 months ago on Fedora 27 KDE. I really like KDE but it was time for a wipe and I fancied hopping back to GNOME for a bit. I use Telegram on the desktop a bit and I installed Telegram via Flatpak. Little did I know, but the official Telegram client relies on qt toolkit, which doesn't come as standard on GNOME, as it's GTK on GNOME. When I installed the Telegram Flatpak, it pulled about 400MB of KDE/QT libraries and a few seconds later I had the KDE version of Telegram, looking really nice, running sandboxed in GNOME. Without having to install loads of KDE dependencies on my fresh GNOME desktop. So I like Flatpaks. It's a cool idea.
|
|
#
?
May 29, 2019 23:49
|
|
|
Volguus posted:Fedora devs just twiched: Flatpak Flatpak isn't a repo, it's a packaging spec. Flathub is a repo of curated flatpak's, often maintained by the upstream authors but if it's not the community keeps things super up to date. That said, you don't need a repo to use flatpaks, you can compile one on your own if you wanted but it'd be up to you to maintain it if you wanted newer versions.
|
|
#
?
May 30, 2019 05:13
|
|
|
I'm have a long running process that has the possibility of using an absurd amount of memory just before finishing. I have a swap partition set up already, but I don't think I sized it large enough. Can I add a swap file in addition to the partition to supplement it(without rebooting/unmounting root partition) if I have plenty of free drive space?
|
|
#
?
May 30, 2019 14:45
|
|
|
peepsalot posted:I'm have a long running process that has the possibility of using an absurd amount of memory just before finishing. I have a swap partition set up already, but I don't think I sized it large enough. Can I add a swap file in addition to the partition to supplement it(without rebooting/unmounting root partition) if I have plenty of free drive space? Yes; you create the file (using fallocate or dd if=/dev/zero) and then mkswap and swapon it as normal. See the man pages for the latter commands for some cautions about using swap files on certain filesystems.
|
|
#
?
May 30, 2019 15:37
|
|
|
I think there's another way now of creating a swap file a little more elegantly, but I'm not certain. I have a USB stick with just one big swap partition on it. It's for when I'm using an SBC and have an "Oh poo poo!" moment when something balloons out. I shove the stick in, check it's the device I think it is then activate the swap partition.
|
|
#
?
May 30, 2019 21:55
|
|
|
The big calculation server at work has a 512GB Samsung consumer ssd jimmied into a 3.5" slot with folded pieces of the box it came in. We needed more than the 256GB of RAM in it for an analysis (some CgH/mRNA expression clustering IIRC), so I dropped by an electronics store on the way home from work and stuffed it in there the next day. Worked fine. Not a hotplug, though; I had to power down. A fast usb 3.1 drive would have been a smarter choice in that regard. (HPE hardware is basically permanently one boardroom martini/coke line away from moving to their own proprietary connectors for everything. Hotplug a drive without the appropriate, DRMed, super expensive, "smart" tray, and the machine spins every fan to 100% until you power it down. This sounds like something between an angry vacuum cleaner and an air raid siren. )
|
|
#
?
May 31, 2019 01:44
|
|
|
Oh poo poo. Scientific linux is going bye byes too.
|
|
#
?
May 31, 2019 03:54
|
|
|
General_Failure posted:Oh poo poo. Scientific linux is going bye byes too. There's no point to it, it's identical to centos except for a couple packages used to brand it. They had the same debate with rhel7 but this time opted to retire the distro.
|
|
#
?
May 31, 2019 04:00
|
|
|
Computer viking posted:(HPE hardware is basically permanently one boardroom martini/coke line away from moving to their own proprietary connectors for everything. Hotplug a drive without the appropriate, DRMed, super expensive, "smart" tray, and the machine spins every fan to 100% until you power it down. This sounds like something between an angry vacuum cleaner and an air raid siren. ) This is funny cause
|
|
#
?
May 31, 2019 04:02
|
|
|
CFox posted:This has been my biggest irritation about linux really. I'd love to install everything from one central repository that is updated frequently and not have to worry about adding a dozen different PPAs but besides possibly the Arch User Repository I don't think there's anything like that in linux land and Arch just ain't my thing. Surely everyone else wants this too right? I feel like this should be a very high priority for all distros. Basically just do what google did with the android play store and I'd be happy. Do other distros have anything similar to the openSUSE software search page? https://software.opensuse.org/explore It's pretty handy to be able to search on there and often find what I need available for a one-click install.
|
|
#
?
May 31, 2019 06:55
|
|
|
SoftNum posted:This is funny cause Omigod. More business for them I guess?
|
|
#
?
May 31, 2019 08:35
|
|
|
CaptainSarcastic posted:Do other distros have anything similar to the openSUSE software search page? https://software.opensuse.org/explore Yea I've found something similar for debian and fedora I think.
|
|
#
?
May 31, 2019 13:23
|
|
|
CaptainSarcastic posted:Do other distros have anything similar to the openSUSE software search page? https://software.opensuse.org/explore http://packages.ubuntu.com https://apps.fedoraproject.org/packages A whole lot of distros: https://pkgs.org
|
|
#
?
Jun 1, 2019 04:32
|
|
|
I'm about to create a new LVM VG on a boot drive for hosting some VM's in some LVs. The drive is 500GB and about 200GB full of other OS stuff. Is it worth doing a "dd if=/dev/zero of=fatfile bs=1G count=300" and deleting it before provisioning the volume group, so that I have a nice contiguous piece of space for running my LVM partitions on? The drive is an SSD.
|
|
#
?
Jun 6, 2019 21:02
|
|
|
apropos man posted:I'm about to create a new LVM VG on a boot drive for hosting some VM's in some LVs.
|
|
#
?
Jun 6, 2019 21:10
|
|
|
I want to leave the OS intact. The usual trick of dd'ing a load of zeros into a massive file, then deleting it is what I'm after. Then, once the huge file is deleted you basically have a contiguous stretch of zeros after the OS files. I think what I'm doing is a waste of time, since the SSD controller probably positions blocks all over the place. I'm nearly finished dd'ing now, anyway. So I've probably wasted 20 mins but, hey ho.
|
|
#
?
Jun 6, 2019 21:30
|
|
|
Are you planning to resize the OS partition and then create a new partition for the LVM VG? A partition can only belong to one volume group, every new VG will require a new partition. That dd-operation would not guarantee that you have contiguous zeros after the OS files. First 10 GB of the fatfile could be between some existing files followed by OS files, then there could be 50GB more of the fatfile again followed by some other files. The files on the drive may be spread around everywhere logically. Below that is all the weirdness the SSD does.
|
|
#
?
Jun 6, 2019 22:18
|
|
|
I finished the fatfile, shrunk the original CentOS LVM root partition because I wanted to create a set of LV's with specific names for virtual machines. Guess what? I forgot to mount the root partition from a live OS and instead I reduced the root LVM it while it was running. I ended up with a segfault and ELF errors. What a dumbass. So I'm sat here, at almost midnight, reinstalling RHEL 8 developer edition. I did manage to qemu-img convert my important VM's first and put them on a separate RAID array, so at least I only have to import my VM's once I get a brand new OS running again. I expected to be in bed by now. I get what you're saying about the fatfile not being exactly contiguous. I think it's only useful if you're doing LVM snapshot backups and want to GZIP them. So creating a load of contiguous zeros is pointless unless you intend to dd the image and zip it up for storage.
|
|
#
?
Jun 6, 2019 23:41
|
|
|
apropos man posted:I finished the fatfile, shrunk the original CentOS LVM root partition because I wanted to create a set of LV's with specific names for virtual machines. Did you also resize the filesystem or just the LVM volume? Are you using ext4 instead of the standard XFS?
|
|
#
?
Jun 7, 2019 00:48
|
|
|
I think I was using xfs. That's default in CentOS 7 CLI installer, isn't it? Anyway, it's blown away now, as am I. It's 2:30am and I have a new base install to finish off tomorrow. Zzzzzz
|
|
#
?
Jun 7, 2019 02:23
|
|
|
You cannot shrink XFS.
|
|
#
?
Jun 7, 2019 07:02
|
|
Associate Christ
|
Good to know. I'm gonna start from scratch tonight. I have a CentOS 7 iso and the new RHEL 8 developers (free edition) iso. Hmm. First time I'll have used RHEL at home, so I think I'll give that a go. I imaged my VM's before I shitted everything up last night, so it's not a bad thing to have a fresh host and I'll just import the VM's into RHEL.
|
|
#
?
Jun 7, 2019 15:08
|
|
|
Two things to note. As mentioned, XFS can't be shrinked, it can only grow. More importantly, you can't just shrink a LVM volume, you need to shrink the file system first. If you had used something like ext4 shrinking would have been possible. I haven't tried it myself, but easiest method is probably the command 'lvreduce --resizefs' with supported filesystems. Since shrinking is tricky but growing is trivial, our approach at work with servers is to install them with a bunch of small LVM volumes, just 1 - 5 GB in size, and then lvextend them when necessary or create new volumes.
|
|
#
?
Jun 7, 2019 18:42
|
|
|
Yeah. Some good points. Since I usually plug a cheap GPU into my headless server and then remove it after I've installed the OS, I just adjusted the LVM's in anaconda this time round. I've just installed CentOS 7 with root and swap in LVMS. But this time I specified a smaller size than 'all available space' for root and left a big chunk, which I then stuck in the same VG and called it 'virtual_hosts'. So I now have this: code:I went back to CentOS because I forgot last night that ZFS support isn't yet available as a kernel mod for RHEL8 and I like ZFS too much to do without it. I'm busy putting the pieces in place now and will shortly be importing my VM's. e: actually, I think I should blitz that last LVM and split the single 'virtual_hosts' LVM into multiples for each VM. That'll be even tidier. apropos man fucked around with this message at 19:40 on Jun 7, 2019 |
|
#
?
Jun 7, 2019 19:18
|
|
|
apropos man posted:
You would suffer from disk space inefficiency unless you know exactly how much space each VM requires and they won't be increasing. You would need to give each VM free space instead of shared free space. All VMs on a same volume with thin disks will give you the most from your storage space. Unless you are giving the LVM volumes directly to the VMs as raw disks.
|
|
#
?
Jun 7, 2019 21:57
|
|
|
Yup. I'm doing the raw disks thing. Storage is cheap and the VM's I use are mostly static in size: a 20GB VM for admin of my personal stuff, a 40GB torrent box, a 10GB DNS and a 10GB bastion host. The only reason the torrent box and bastion are that size is if I want to shuttle files to and from home when I'm out. The torrent box links to an internal samba ZFS tank and the personal files one links to a similar NFS share. It's 40GB for the torrent box in case I'm leeching 3 or 4 movies and then they get moved over to the ZFS tank, too, once leeching is done. It's easier to script a snapshot partition on a cronjob, dd the contents over to the ZFS tank and then lvremove the snapshot with statically sized LV's. These are gonna be on a weekly refresh at, like 3am on a Sunday morning. Then they can be backed up further from software RAID, if necessary. I'm almost done with the LVs/VM's now. Just gotta reinstate my samba and NFS shares in the morning. In addition to this there's the docker containers that need putting up tomorrow. But it's getting late and when the eyelids start closing mistakes tend to be made. One more VM to bring to life and then sleep.
|
|
#
?
Jun 8, 2019 00:26
|
|
|
Is there a clever way to upgrade PHP and PHP-FPM from 7.2 to 7.3, but retain all your configuration modifications without redoing it all?
|
|
#
?
Jun 11, 2019 23:34
|
|
|
Something is borked regarding my DNS and I can't figure it out. And I dont know if all of the things I've messed with have made it worse. Ubuntu 18.10 headless in virtualbox on Windows 10 host. Before it stopped working, everything was collected from the DHCP server, with the router handing out my Pi-Hole IP address as the DNS. But as part of my fiddling I've tried cutting out PI-Hole. But still nothing. code:code:code:MeKeV fucked around with this message at 11:35 on Jun 14, 2019 |
|
#
?
Jun 14, 2019 09:56
|
|
|
MeKeV posted:Something is borked regarding my DNS and I can't figure it out. And I dont know if all of the things I've messed with have made it worse.
|
|
#
?
Jun 14, 2019 12:24
|
|
|
I have been through various contents, currently the file doesnt exist and isn't getting generated. Edit: just created one and added just nameserver 1.1.1.1 and I seem to be up and running again. Not sure if editing the file directly is a temporary fix. But it will do for now. MeKeV fucked around with this message at 13:19 on Jun 14, 2019 |
|
#
?
Jun 14, 2019 13:04
|
|
|
I just ran into something similar yesterday and it was because my router was setting the ipv6 dns servers. And that was getting added first in resolv.conf. It took me a while to find the cause and then figure out how to setup the pihole correctly.
|
|
#
?
Jun 14, 2019 14:25
|
|
|
|
| # ? Apr 23, 2024 10:51 |
|
|
Our baseline config at work disables ipv6 on install because the networking group is slowly enabling ipv6 on all their equipment and so much poo poo breaks if the host sets up an ipv6 address on an interface. So when we get the notice that a site/vlan now has ipv6 available we pick a test box so users of that segment can reconfigure all their poo poo to work like they need, and then we turn on ipv6.
|
|
#
?
Jun 14, 2019 14:32
|
|