View previous topic :: View next topic |
How many Gentoo machines do you maintain? |
1 |
|
24% |
[ 28 ] |
2 |
|
14% |
[ 17 ] |
3 |
|
16% |
[ 19 ] |
4 |
|
13% |
[ 16 ] |
5 |
|
4% |
[ 5 ] |
more than 5 |
|
24% |
[ 28 ] |
what the hell, you run Gentoo? You masochist... |
|
1% |
[ 2 ] |
|
Total Votes : 115 |
|
Author |
Message |
eccerr0r Watchman
Joined: 01 Jul 2004 Posts: 9824 Location: almost Mile High in the USA
|
Posted: Sat Feb 06, 2021 3:23 pm Post subject: How many Gentoo boxes do you maintain? |
|
|
This was starting to bother me a bit due to the apparent difficulty in upgrading my systems. If I only had one machine this wouldn't be so bad, but...
So the poll here for curiosity. How many Gentoo boxes do you maintain?
If you replicate boxes where @world and USE (include CPU_FLAGS) are the same, count that as just one machine.
Count virtual machines separately if they have their own @world and USE flags. Also do not count non-Gentoo (but you may add in Funtoo and any other distributions that use portage as long as you maintain them and not fire and forget). _________________ Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching? |
|
Back to top |
|
|
alamahant Advocate
Joined: 23 Mar 2019 Posts: 3918
|
Posted: Sat Feb 06, 2021 3:45 pm Post subject: |
|
|
One openrc,one systemd,one hybrid,some VMs and some lxc containers..
_________________
|
|
Back to top |
|
|
saellaven l33t
Joined: 23 Jul 2006 Posts: 654
|
Posted: Sat Feb 06, 2021 3:59 pm Post subject: |
|
|
I'm currently at 6, but I update regularly and mostly use stable. _________________ Ryzen 3700X, Asus Prime X570-Pro, 64 GB DDR4 3200, GeForce GTX 1660 Super
openrc-0.17, ~vanilla-sources, ~nvidia-drivers, ~gcc |
|
Back to top |
|
|
eccerr0r Watchman
Joined: 01 Jul 2004 Posts: 9824 Location: almost Mile High in the USA
|
Posted: Sat Feb 06, 2021 4:46 pm Post subject: |
|
|
How do you find time to maintain all of them (again, I excluded replicated machines as that's an obvious way though it's not always appropriate)? The per-machine time is taking longer and longer to figure out all the conflicts is making me think again, else I'd be adding more machines.
I think I'm up to 7 machines I must keep up to date, and 3 more I'd like to keep up to date. Every single one has a different setup, some openrc, some systemd, it's been chaos and becoming a huge time sink. Them also being a mixture of stable and individually unmasked package makes it harder. I suspect running unstable would be even worse.
Currently I'm in the 32-bit phase of updating to 17.1 on my main workstation, as well as updating my VMM server and PVR (at same time). Thank goodness for distcc, but what I really need is distrust - cbindgen, firefox, and libsvg are starting to get annoying as they can't be distributed.
high priority (7)
- main workstation (64 bit) -- this one is usually hardest as it has the largest number of software packages installed
- VMM server (64)
- outward facing webserver on VM (32)
- 32 bit laptop
- 64 bit laptop
- 2xFriend's machines (remote, 32)
Medium priority (3)
- HTPC/PVR machine (64)
- disk backup server (32)
- old 64 bit workstation
Low priority
- old 32 bit laptop
- few other scratch machines
Not updated
- few others like my ia64 box because it got dropped from support for a lot of software... and a USB stick that I use as a rescue image that could be used as a makeshift workstation. _________________ Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching? |
|
Back to top |
|
|
Tony0945 Watchman
Joined: 25 Jul 2006 Posts: 5127 Location: Illinois, USA
|
Posted: Sat Feb 06, 2021 4:51 pm Post subject: |
|
|
saellaven wrote: | I'm currently at 6, but I update regularly and mostly use stable. |
ditto. About to build a seventh box. All PC's except one Raspberry Pi. I've been using Sakaki's builds but when I complete the seventh box (3900x), I'm going to try cross-compilation again.
eccerr0r,
Only problems are marathon updates of python3. Occasionally perl has fits, but I just unmerge all perl then do an update.
The 32bit install has special problems and I don't always update it. The latest machine is a recycled k10 board I'm using as an appliance. It has no access to the internet (I sync one central server then sync the others to that), I'm considering never syncing. It's purpose is to run an out of tree program and I'll just update that until it needs a later library or something.
Once the python mess is squared away on one machine, I know what to do for the rest. Unfortunately, unlike perl, I can't just blow it all away and reinstall. But I could if portage was written in C or C++.
All my machines are openrc-0.17. None are gaming machines. The Raspberry Pi is an appliance too. I use it to send xine output to my big TV from a samba mount on the basement server. Three of the machines are in the basement as is the cable modem, router and two access points (One for each end of the house).
My wife has a Win7 laptop that I might convert to Gentoo if we stay locked up until Fall. That will only take a SATA SSD which are getting really cheap. yes, I can boot the last Gentoo based sysrescuecd by usbstick on the laptop. Considering using it for an appliance to to provide wireless TV, but I can do that on Win7.
Last edited by Tony0945 on Sat Feb 06, 2021 5:33 pm; edited 1 time in total |
|
Back to top |
|
|
NeddySeagoon Administrator
Joined: 05 Jul 2003 Posts: 54578 Location: 56N 3W
|
Posted: Sat Feb 06, 2021 5:10 pm Post subject: |
|
|
My chroot on the Gentoo ARM 64 build box.
X-Gene 1 which builds for the Pi4
Raspberry Pi 4, which uses a binhost.
Acer R13 Chromebook, also arm64, currently suffering from neglect.
HP Microserver G7, bare metal install for hosting KVMs.
Four KVMs in the above.
E35M-1 Silent System. Media Player. (I should upgrade this to a Pi4)
My main PC. (Phenom II)
Bare metal server install at Hetzner
Three KVMs in the above.
That's 15 but KVMs come and go. _________________ Regards,
NeddySeagoon
Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail. |
|
Back to top |
|
|
saellaven l33t
Joined: 23 Jul 2006 Posts: 654
|
Posted: Sat Feb 06, 2021 5:36 pm Post subject: |
|
|
eccerr0r wrote: | How do you find time to maintain all of them (again, I excluded replicated machines as that's an obvious way though it's not always appropriate)? The per-machine time is taking longer and longer to figure out all the conflicts is making me think again, else I'd be adding more machines.
I think I'm up to 7 machines I must keep up to date, and 3 more I'd like to keep up to date. Every single one has a different setup, some openrc, some systemd, it's been chaos and becoming a huge time sink. Them also being a mixture of stable and individually unmasked package makes it harder. I suspect running unstable would be even worse.
Currently I'm in the 32-bit phase of updating to 17.1 on my main workstation, as well as updating my VMM server and PVR (at same time). Thank goodness for distcc, but what I really need is distrust - cbindgen, firefox, and libsvg are starting to get annoying as they can't be distributed. |
All of mine are AMD64 with a small handful of unstable packages*, sharing an overlay of my modified packages between them, all of which run openrc-0.17.
alpha is where I first test out and deal with updates (and it contains all of the software every other machine has on it too, in case I need to use it as a hot swap for any of them), which lets me figure out how to deal with potential breakage before I upgrade the other machines. Once alpha is taken care of, I'll immediately do the other machines while everything is fresh in my mind. During the occasional breakage, I'll either wait a day or two for the fix to propagate through portage or apply it to my overlay if it's a pressing need.
Maybe once every other month I have to deal with some kind of minor breakage, most of which is inflicted by devs doing silly things without fully thinking through what they're doing. I suspect the upcoming removal of python-2.7 breaking packages that need it for build time (palemoon and webkit) and potentially the display-manager-init bikeshedding to be the next things I might have to deal with.
* alpha - home desktop
Packages installed: 1558
Packages in world: 287
package.accept_keywords
Code: |
sys-kernel/vanilla-sources
sys-kernel/linux-headers
x11-drivers/nvidia-drivers
#not available on AMD64
mail-client/gbuffy ~x86
#all versions masked
games-board/pysolfc
net-im/zoom
app-emulation/vkd3d
games-misc/legendary
games-misc/minigalaxy
sys-apps/x86info
#flightgear and dependencies all need keyword
games-simulation/flightgear
dev-games/simgear
games-simulation/flightgear-data
#for freecad
media-gfx/freecad::waebbl
=dev-libs/OpenNI2-2.2_beta2
=sci-libs/med-4.1.0
=sci-libs/orocos_kdl-1.4.0
~sci-libs/opencascade-7.4.0
=sci-libs/vtk-8.2.0
=dev-python/pivy-0.6.5
~dev-python/pyside2-5.15.1
=dev-python/shiboken2-5.15.1
=media-libs/quarter-1.1.0
=app-eselect/eselect-opencascade-1
#for calibre
~app-text/calibre-5.6.0
#like to stay up to date
net-print/foomatic-db
net-print/foomatic-db-ppds
app-emulation/wine-mono
app-emulation/wine-vanilla
app-emulation/wine-gecko
games-util/lutris
#fix bugs
=games-strategy/freeciv-2.6.2
=games-strategy/wesnoth-1.14.14
~media-video/handbrake-1.3.3-r1
|
_________________ Ryzen 3700X, Asus Prime X570-Pro, 64 GB DDR4 3200, GeForce GTX 1660 Super
openrc-0.17, ~vanilla-sources, ~nvidia-drivers, ~gcc |
|
Back to top |
|
|
eccerr0r Watchman
Joined: 01 Jul 2004 Posts: 9824 Location: almost Mile High in the USA
|
Posted: Sat Feb 06, 2021 5:37 pm Post subject: |
|
|
Tony0945 wrote: | Only problems are marathon updates of python3. Occasionally perl has fits, but I just unmerge all perl then do an update.
|
That's the thing, if you unmerge everything, this is time burned as the machine tends to be unusable during the downtime, plus you're spending the time to figure out.
Just wondering how to minimize this so I can deal with more machines. Or is it that portage needs to get smarter to figure out more of these problems on its own?
People end up saying that time between upgrades is what causes these strange merge issues, but this doesn't quite work either:
* * * * * emerge --sync && emerge -uUD @world
(hint: what UN*X file has 5 fields before a command line?)
Well, it doesn't work not specifically because it'd get you banned from the sync servers or the virtual fork bomb it creates, but rather it still requires manual labor to make sure the update completes successfully. And packages still will get removed from portage causing pain.
Another thing too... yes a lot of my machines get updated less and less frequently as more and more conflicts come in and require significant effort to clean up, I have some other Gentoo VMs that I simply don't update because it's taking too much time to figure out the critical ones. I mean, these last few updates I did I'm sure I spent many hours before finally figuring out a solution to get emerge to find a solution. Plus if I had to unmerge something, I couldn't use that software in the meantime.
(Yes I also should ask people to not include throwaway virtual machines, because one is not actively maintaining these.) _________________ Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching?
Last edited by eccerr0r on Sat Feb 06, 2021 5:43 pm; edited 1 time in total |
|
Back to top |
|
|
NeddySeagoon Administrator
Joined: 05 Jul 2003 Posts: 54578 Location: 56N 3W
|
Posted: Sat Feb 06, 2021 5:42 pm Post subject: |
|
|
eccerr0r,
every now and again, you need -N in place of -U and --keep-going is a good idea too. _________________ Regards,
NeddySeagoon
Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail. |
|
Back to top |
|
|
eccerr0r Watchman
Joined: 01 Jul 2004 Posts: 9824 Location: almost Mile High in the USA
|
Posted: Sat Feb 06, 2021 6:07 pm Post subject: |
|
|
I figure a lot of VMs are throwaway VMs that aren't maintained - and thus shouldn't be counted in this poll.
For instance I have my VMM server running 4 VMs. The VMM server I count because it's Gentoo and I need to update it.
Then the VMs: one is a Gentoo VM that's outward facing. I count that. One is a pfSense VM... I don't count. Then there are two more VMs that are Gentoo but I don't update them. These I don't count these because I don't maintain them -- so total count of 2 for this one computer.
So basically I need to maintain one computer twice (not 4 times despite it having 4 copies of Gentoo installed) - and this is what I'm counting here.
saellaven - I'm surprised you're not running into the merge messes when updating. I really would like to keep more machines up to date, even the VMs mentioned above, but it just costs too much time especially when things get kicked out of the tree. How have your updates been going? _________________ Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching? |
|
Back to top |
|
|
Anon-E-moose Watchman
Joined: 23 May 2008 Posts: 6148 Location: Dallas area
|
Posted: Sat Feb 06, 2021 6:15 pm Post subject: |
|
|
Updating daily doesn't keep problems from popping up.
Spent an hour the other day, figuring out why my old python2 ebuilds, in my local repo, were now complaining when I tried to do a changed-use run, and the complaints were about python and python-exec. Nothing on my end. Turned out to be a couple of python* eclass changes Now, I'm adding eclasses to my local repo. Thanks gentoo devs
Edit to add: One thing that has helped, is emerging got a little better when I went from multilib to straight 64bit.
As far as machines, I have one main desktop/server/binhost/etc, one laptop (updated weekly) and one small webserver that I don't count, as it's gentoo, but I never update it, when the hardware dies is when it gets updated. Well, I'm just about to start playing with 32bit chroots (as an alternative to multilib), gentoo and funtoo.
ETA2: I just checked (out of curiosity) and after hiding the python* eclasses in my local repo, it was working today. I just love it when they start changing things, that are out of sight, it works ... it doesn't work ... it works ~le sigh~
But it does look like the kind of straightened out the mess they made re python2 "removal" _________________ UM780, 6.1 zen kernel, gcc 13, profile 17.0 (custom bare multilib), openrc, wayland
Last edited by Anon-E-moose on Sat Feb 06, 2021 7:43 pm; edited 1 time in total |
|
Back to top |
|
|
saellaven l33t
Joined: 23 Jul 2006 Posts: 654
|
Posted: Sat Feb 06, 2021 6:33 pm Post subject: |
|
|
I update daily with
emerge -a --update --changed-use --with-bdeps=y --deep --keep-going --autounmask @world
When things I need get kicked out of the tree, I move them into my overlay.
It's been a few months since the last time I had much of an issue (or maybe I'm so used to dealing with certain things, I'm mentally ignoring them at this point). Like i said, a lot of the things I run into are because of devs that should know better not thinking about what they're doing before committing.
With regard to python, I had a few issues maybe 6 months ago, but nothing much as of late (though, like I said, I'm expecting 2.7 to break stuff soon). elogind has been my most recent headache that I can think of, but it's more of an annoyance (I really, really, really hate that it decides to consume my tty rather than outputting errors to the tty while launching X on vt7, but I haven't had time to investigate that yet to see if I can unbreak that bit of stupidity... I mean, who needs to see errors that don't get logged to X.org.log anyway, right?)
I don't remember the last time I had a perl issue.
I'm not sure what I might be doing differently than you... I know I've kept a lot of my saellavenisms from my pre-gentoo days, when I put together my own distro for my computers, so that might unknowingly factor in somewhere (over the years, I've seen (and avoided) a whole lot of gentoo devs do or recommend things that really make me scratch my head since they're more likely to break things and make systems unstable). I can state with certainty that a lot of my personal scripts, utils, and .config type files have migrated with me for 26+ years of using linux, though not all of them are going to keep gentoo from breaking. Maybe it's more of a mindset I have in how I manage my boxes, I don't know (which is a big reason why you've seen me opposing certain ideas from certain devs over the years). _________________ Ryzen 3700X, Asus Prime X570-Pro, 64 GB DDR4 3200, GeForce GTX 1660 Super
openrc-0.17, ~vanilla-sources, ~nvidia-drivers, ~gcc |
|
Back to top |
|
|
Zucca Moderator
Joined: 14 Jun 2007 Posts: 3708 Location: Rasi, Finland
|
Posted: Sat Feb 06, 2021 6:50 pm Post subject: |
|
|
3.
Desktop, laptop and home server. _________________ ..: Zucca :..
My gentoo installs: | init=/sbin/openrc-init
-systemd -logind -elogind seatd |
Quote: | I am NaN! I am a man! |
|
|
Back to top |
|
|
eccerr0r Watchman
Joined: 01 Jul 2004 Posts: 9824 Location: almost Mile High in the USA
|
Posted: Sat Feb 06, 2021 7:42 pm Post subject: |
|
|
I sure hope this latest update saga is an exception versus a rule for the future with all these use and slot conflicts. It's not like these just showed up either, they showed up months ago and I put it off until now.
Yes going only 64-bit mode probably simplifies things a bit, just to reduce number of packages installed. Then again they're the same package for 32/64...
As an aside, I've only updated one amd64 box to 17.1 - it's one of the mid priority machines as that one I can sacrifice during this learning curve. I'm in progress of migrating my main workstation to 17.1 - it's currently recompiling the 32-bit libraries. I'm _very_ worried about killing my server when doing the migration, at least it's up to date now as the first step. I'm currently updating my PVR so I can do the migration, and I need to clean up and update my 64-bit laptop before I can do the migration.
I still have a lot of work is pending *sigh* and having lots of boxes is an even bigger *sigh* ... _________________ Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching? |
|
Back to top |
|
|
Anon-E-moose Watchman
Joined: 23 May 2008 Posts: 6148 Location: Dallas area
|
Posted: Sat Feb 06, 2021 7:51 pm Post subject: |
|
|
There were a few reasons for going straight 64, the emerge time wasn't particularly one of them, the abi 32 stuff I got tired of, so that was one reason.
Another I moved away from wine (it only worked on some of the things I wanted to run), so I opted to put together an old machine strictly for that reason.
Although I've been wanting to play with wine again (just to see what improvements have been made), I think I will do it by way of a i686 chroot instead of multilib or vm.
By going straight 64bit I also had an easy time with 17.0 -> 17.1 by making a bastardized profile, keeping */lib and */lib64 as symlinks, and not recompiling even one package. Though I could have kept the regular 17.1 and just removed one particular file that checks for the symlink and refuses to run if it sees it. But I would have to remove it after every sync. _________________ UM780, 6.1 zen kernel, gcc 13, profile 17.0 (custom bare multilib), openrc, wayland |
|
Back to top |
|
|
pjp Administrator
Joined: 16 Apr 2002 Posts: 20485
|
Posted: Sat Feb 06, 2021 11:27 pm Post subject: |
|
|
eccerr0r wrote: | I think I'm up to 7 machines I must keep up to date, and 3 more I'd like to keep up to date. Every single one has a different setup, some openrc, some systemd, it's been chaos and becoming a huge time sink. | eccerr0r wrote: | That's the thing, if you unmerge everything, this is time burned as the machine tends to be unusable during the downtime, plus you're spending the time to figure out.
Just wondering how to minimize this so I can deal with more machines. Or is it that portage needs to get smarter to figure out more of these problems on its own? | Whether or not portage could or should be smarter, that every machine is unique is the underlying problem. This is why "automation" solutions such as ansible / chef / puppet aren't an easily viable solution. They primarily benefit environments that deploy a lot of systems with the same configuration.
eccerr0r wrote: | People end up saying that time between upgrades is what causes these strange merge issues, but this doesn't quite work either:
* * * * * emerge --sync && emerge -uUD @world | Update Roulette offers almost no gain with an extremely high downside risk. I'm always amazed when I see people mention that's how they perform updates.
What I hope to have one day is a system that shows me what is available to update, will then go create the binaries and notify me when it is complete. Then I'll begin deploying the binaries. In ZFSland, I'd even perform some automatic binary updates due to the availability of boot environments (just reboot back to the previous BE in case of emergency).
To abuse a quote, $#!% ain't like that!
And to respond to your original question, my process:
workstation
chroot (laptop build)
laptop
NFS server (limbo)
stable with very few ~ programs.
Because I monitor forums for spam, etc., I also often notice potential update problems. Fortunately (for me), this has helped me avoid being completely blindsided on most major issues. Special thanks to frequent updaters / early adopters.
Updating:
start with workstation. No GUI, fewer problems.
Helps identify potentially critical problems with toolchain / openssh / PAM / etc., changes to installed packages where the ebuild doesn't receive a -r bump. Use --exclude or mask for short term and long term avoidance of some upgrades.
This system also generates squashfs images of ::gentoo. I have not yet completely resolved smooth / automatic management of that process, including distribution.
Chroot:
lessons learned from the workstation hopefully make this easier. Otherwise this is where I first encounter issues with X or anything not installed on the workstation.
Previously mentioned monitoring of forums help with timing when I perform updates. If one browser receives a security update, Firefox is probably not far behind.
semi-automated scripts help with squashfs images for ::gentoo and the kernel and making binary packages available to the laptop. /etc/portage is currently made available by compressed tar.
Laptop:
Hopefully this goes smoothly with the previous work. And except for the initial and long transition period to using the chroot for creating binaries, it has mostly been fairly smooth. This is why I still have hope that I might eventually reduce how time consuming (for me) Gentoo maintenance has become. Many years ago I didn't have as much wariness about what I was going to see with the next sync.
semi-automated scripts retrieve squashfs images and the /etc/portage compressed tar, but extracting the tar is unfortunately still manual (I hope to replace it, but haven't decided how).
NFS server (limbo):
All in all, the above process is time consuming enough that I've neglected finding a way to resolve this.
End goal (hopefully):
Generic binaries for everything (AMD/Intel). The chroot currently has "-mtune=generic"
Common build - binaries used everywhere.
Common + NFS
Common + Workstation
Common + Laptop
My hope is that each layer of binaries can be reused as much as possible, but I have no idea how to make that happen. And my process for chroot / client updating isn't automated enough to extend the use of chroots to the workstation, NFS server, and VMs (one area I'd prefer to spend free time rather than on updates).
That doesn't address ebuilds evicted from ::gentoo or chroot to client world file changes.
"Busy work" comes to mind, but I haven't figured out how to reduce it.
I'm probably adding some unnecessary complexity, but haven't thought of a "better" approach. Keeping a chroot and client using the same repos, kernel and /etc/portage isn't trivial. NFS isn't ideal, and not something I'd want to use of wireless. Maye it would be "good enough" for fetch only (which is what I use before updates).
If migrating to a binary distro didn't seem to be at least as much effort, I'd probably have switched. But I'm not interested in something even more fringe than Gentoo, or anything with systemd. I never liked apt-get, so I don't know it at all. Does anything rpm/yum not use systemd? I assosciate dnf with Did Not Finish (haven't used it yet). _________________ Quis separabit? Quo animo? |
|
Back to top |
|
|
The Doctor Moderator
Joined: 27 Jul 2010 Posts: 2678
|
Posted: Sun Feb 07, 2021 1:54 am Post subject: |
|
|
I'm kind of surprised that there isn't a lot of "2" votes. One desktop and one laptop. I'm at one because I basically consider my laptop to be dead and is therefore unmaintained. _________________ First things first, but not necessarily in that order.
Apologies if I take a while to respond. I'm currently working on the dematerialization circuit for my blue box. |
|
Back to top |
|
|
eccerr0r Watchman
Joined: 01 Jul 2004 Posts: 9824 Location: almost Mile High in the USA
|
Posted: Sun Feb 07, 2021 2:01 am Post subject: |
|
|
Honestly I'm both appalled and comforted there are a large number of people with more than five machines.
I suspect some of them are just reporting number of machines and not number of distinct setups being actively maintained which was the intent of the poll. I would have thought that once you get this many machines the aggregate time spent on all machines fixing conflicts as the portage tree moves on will become prohibitively time intensive... Or perhaps none of the machines have GUIs. But perhaps there are just slackers like me that if I don't see any serious GLSAs I let them slide for a while. _________________ Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching? |
|
Back to top |
|
|
Tony0945 Watchman
Joined: 25 Jul 2006 Posts: 5127 Location: Illinois, USA
|
Posted: Sun Feb 07, 2021 2:38 am Post subject: |
|
|
Actually I reported the number of machines, but several have chroots also. And one machine doubles as a build box for big builds on another (ryzen building for k10). Also k10 building in a chroot for k6. |
|
Back to top |
|
|
Ionen Developer
Joined: 06 Dec 2018 Posts: 2853
|
Posted: Sun Feb 07, 2021 3:23 am Post subject: |
|
|
Only 3 for real boxes, but few other installs as well.
Updated daily:
- desktop, and server (server uses binpkgs and doesn't build anything), both ~amd64. These are updated together with a single script. Never gives me major problems, but it's ~testing so may get a build failure now and then. Not using stable for the server is questionable but I can do basic checks before pushing the binpkgs.
Only updated when I need it:
- 2nd desktop used for testing normality running amd64 stable. Uses systemd, a full gnome desktop, wayland, and other things I normally don't use. Never had problems with it, it just works. I could use a VM instead (like I do for win10 with GPU passthrough) but might as well put the hardware to use, that old i7-2600 still compiles fine.
- a prefix installation used for testing, still no real problems but prefix is more out there.
- a x86, and musl chroot for testing (may add more, like arm64 with qemu), along my own system I tend to use overlayfs over those for testing (while keeping most binpkgs for re-use). musl support is kind of up in the air so I don't expect much there. Still developing the whole testing env stuff though, the overlayfs deletes native-symlinks by default and gives the bugzilla tinderbox nagging experience.
I have no laptop anymore But I'd likely treat it like the server with binpkgs if I did.
I might add that I've never, not once, had problems dealing with python -- or anything really, my experience with gentoo been mostly smooth sailing. I also don't use overlays except my own.
Most installations share the same /etc/portage tracked in git that loads different files based on the machine along with common-for-all settings. Server also mirror the gentoo repo and everything syncs with it only when needed, also nginx distfiles cache given propagating mounts everywhere can be pain and a cache mirror does the job. Server also does some CI for my local overlay like merging changes from gentoo using that mirror'ed repo, and spams me with filtered commit mail. |
|
Back to top |
|
|
superjaded l33t
Joined: 05 Jul 2002 Posts: 802
|
Posted: Sun Feb 07, 2021 3:26 am Post subject: |
|
|
My main desktop is probably the only system that qualifies as unique by the criteria outlined by OP, but that desktop also has at least 4 different configurations between the main system, the VMs and binhost that I manage on it.
- haibane: The actual system running amd64 with a smattering of stuff from ~amd64. This machine more or less does everything I need from a server: my wireguard server, docker and my fileserver. It's also my daily driver desktop and virtualization station. I have the hardware setup for PCI passthrough so the below VMs are used with GPU acceleration typically.
- gentoo-plasma-unstable: VM I use to mess around with "unstable" desktop stuff, ACCPET_KEYWORDS="~amd64", Currently running plasma 5.21 beta
- gentoo-plasma-git: VM running ACCEPT_KEYWORDS="~amd64" but also plasma from git.
- binhost chroot ("intel-stable"): A chroot I manage to act as a binhost for my laptop and intel NUC.
The world file is going to be largely the same between all four of those systems, but they're all managed a little differently. haibane has quite a bit of server-type stuff on it while the VMs are just desktop specific stuff for the most part. I also manage USE flags independently, but since they're largely the same use case, the USE flags are not going to vary particularly wildly between each other.
Honestly, I'd argue that the binhost that I manage for the two machines that consume binaries from it was actually a little more difficult to get right than managing the other systems independently. The binary package system is a little more finicky BECAUSE it requires your configuration between the various machines involved to be more or less the same in order to be able to take advantage of those binaries. It probably would have been easier just to manage those systems independently (even if more time consuming in the long run), but I wanted those systems not to have to do much compiling. |
|
Back to top |
|
|
eccerr0r Watchman
Joined: 01 Jul 2004 Posts: 9824 Location: almost Mile High in the USA
|
Posted: Sun Feb 07, 2021 3:54 am Post subject: |
|
|
Yeah I think my topic wording could have been better chosen though I think my explanation should be close. I think I did not clearly indicate the reason for this poll: see how many unique setups any one particular person maintains.
Since a "Gentoo Machine" sort of implies physical hardware, I might be requesting an undercount as I'd like virtual machines and chroots that are maintained separately should be counted separately - if one
# chroot /this_new_chroot
# emerge --sync
# emerge --update @world
... I'd like something like this counted as another "machine" (unless if you also use --usepkgonly !).
It also could be an overcount if someone counts a machine they don't maintain - either a physical machine or a stage3 tarball one dumps out and never updates it, whether it's used in an airgap or not. Perhaps I'd like honeypots not be counted - though they are "used" but generally honeypots were meant to be exploited and upgrading software kind of defeats the purpose. So in this respect, I specifically mean "maintain" as one again "emerge --update @world" (and goes through cleaning up the fallout) in some respect on any root filesystem - virtual or not, chroot or not.
Perhaps a poll of "How many hours do you spend updating any Gentoo system every month" is a better poll of update effort, but that too has a lot of polling holes... _________________ Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching? |
|
Back to top |
|
|
pjp Administrator
Joined: 16 Apr 2002 Posts: 20485
|
Posted: Sun Feb 07, 2021 5:24 am Post subject: |
|
|
eccerr0r wrote: | ... I'd like something like this counted as another "machine" (unless if you also use --usepkgonly !). | Interesting. The machine on which I use --usepkgonly is the machine I worry most about because it is the system I would be most disrupted by having any problems. At least for me, there is more involved to the upgrade process than only running --usepkgonly. *shrug*
eccerr0r wrote: | It also could be an overcount if someone counts a machine they don't maintain | I guess that depends on what you wanted from the information. The machine I'm not maintaining is the result of the time spent maintaining the others leaves me without enough time to address the unmaintained system.
I have no idea how much time I spend on the process of an update, because I do not watch emerge output. But my other activities are distracted while periodically checking on build progress. Oftentimes at least one overnight is involved. Maybe it just feels like more time than it is, which may be due to my perception of less stability with more frequent updates / changes. _________________ Quis separabit? Quo animo? |
|
Back to top |
|
|
figueroa Advocate
Joined: 14 Aug 2005 Posts: 3005 Location: Edge of marsh USA
|
Posted: Sun Feb 07, 2021 5:35 am Post subject: |
|
|
I'm only just recently down to 6. A year ago I was also keeping up all the desktops plus two servers at a small private school, but now that's just down to one server + a VM and the other machines at school are either Mint or MX. The school is now 600 miles remote to me. There is someone there to take care of the hardware.
Two get updated daily as a practical matter: an x86 machine (my old AMD Phenom desktop) that just does server functions, and my main amd64 i7 desktop.
Two other get updated more-or-less every few weeks: testing box here and server at school, both older AMD Athlon x2
My ancient Celeron laptop which gets updated about monthly, usually having to run at least overnight.
The VM (no X11) at school, which is mission critical, but now never updated. It ran the school admin system for 10 years and is needed to generate older student transcripts, which it does on-the-fly. (I keep this VM turned off and keep several clones of the VM running -- but off -- on multiple machines.)
That's six, but maybe I should just count the 5 that I update. I'm becoming a slacker.
I did once (2006) set up and run about 10 Gentoo PCs to support a two-week deployed (civilian) situation, but they were all clones of one machine and just given unique hostnames. That was fun. Afterwards the machines were all given to non-profits -- some went into the school and I continued to support them for many years. _________________ Andy Figueroa
hp pavilion hpe h8-1260t/2AB5; spinning rust x3
i7-2600 @ 3.40GHz; 16 gb; Radeon HD 7570
amd64/23.0/split-usr/desktop (stable), OpenRC, -systemd -pulseaudio -uefi |
|
Back to top |
|
|
superjaded l33t
Joined: 05 Jul 2002 Posts: 802
|
Posted: Sun Feb 07, 2021 9:46 am Post subject: |
|
|
eccerr0r wrote: | ... I'd like something like this counted as another "machine" (unless if you also use --usepkgonly !). |
Did you mean --usepkg or specifically --usepkgonly? Because if you meant specifically usepkgonly, I guess that would put me at six systems rather than four.
I exclude several things from having binaries created such as virtual/* and sys-kernel/gentoo-sources, so I need to be able to do usepkg on the binary clients to be able to perform a full update. The NUC and laptop have custom kernel .configs so I do still compile kernels on those systems. If I were to redo those systems for some reason, I might end up using sys-kernel/gentoo-kernel-bin instead of a custom config, but I haven't really felt the need or desire to switch.
Quote: | Perhaps a poll of "How many hours do you spend updating any Gentoo system every month" is a better poll of update effort, but that too has a lot of polling holes... |
I think even here, we'd have to parse exactly what you mean by this question. If we're talking about CPU time, it might get up there in certain weeks. But if we're talking about having to go through /etc/portage or troubleshoot compiling issues, it's not much.
The "haibane" system is the only one where I'll typically sync and update at least once a day. haibane also acts as an rsync server for the other systems. Updates happen more sporadically on the other system, but updates on the VMs probably at least once every few days. The binhost is usually at least once a week. Binary clients may occasionally go a couple weeks without updates, but I'm typically pretty good about keeping most of my systems that I care about decently up to date.
I would say that one of the first things I do when I've booted up, ssh'd into or chrooted into a system that I don't use every day, it's do updates. So the limiting factor for me is how often I go into those systems.
But my actual upgrade process is very rarely more complicated than eix-sync && emerge -auDNv @world anyway. It's slightly more complicated when there's a kernel update, but even then, not by much. I also, honestly speaking, don't really install anything that takes too terribly long to compile. I'm pretty liberal about using -bin equivalents unless there's a specific reason for me to compile it instead. I use www-client/firefox-bin and dev-lang/rust-bin instead of their source based equivalents. Even on my testing VMs, updates don't typically take much longer than 20-30 minutes to complete on a bad day.. and that's typically just CPU time spent compiling, so I'm more or less going back to whatever I was doing before I started updates.. which is typically something productive like watching youtube or reading reddit.
The one outlier that does cause some pain is dev-qt/qtwebengine for my binhost since my NUC needs it for media-tv/plex-media-player. I might be a little more methodical about when I do an update when that's in queue. Or at least I'll be likely to do something else when that compile is going.
If the underlying premise of your questions is whether or not Gentoo puts unreasonable strain on system admins, I suppose that's ultimately a subjective question as to whether or not the flexibility that Gentoo provides is worth the theoretical attention you may have to pay to the individual systems. I don't feel managing the same number of systems under a binary based distribution would be noticeably easier. Easier on the CPU, maybe, but I don't really look at CPU time inherent in compiling as the same thing as something being time consuming to manage. If you do believe that Gentoo is putting an unreasonable burden on you from an a maintenance aspect, then it does seem irrational to continue using it. |
|
Back to top |
|
|
|
|
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
|
|