| gnarface | the thing about leaving unpartitioned space for "wear leveling" is something i do remember hearing when SSDs were new too, but i haven't heard anyone seriously talking about it since 1st generation tech... once the longevity of SSDs started approaching HDDs after the 2nd or 3rd generation of the tech, i stopped hearing about that | 05:46 |
|---|---|---|
| gnarface | i'm not sure what the real situation is, but i've been lead to believe that the concern is not significant anymore, and modern built-in wear leveling should obviate the problem as long as you're using the TRIM feature | 05:49 |
| gnarface | so whether leaving unpartitioned space even can help now is even in question for me | 05:50 |
| gnarface | but it wouldn't be any sort of surprise to me if a company like kingston were to be purposefully leaving outdated advice floating around when it encourages people to buy more product | 05:51 |
| gnarface | for example, most companies even still tell you to "burn in" your rechargeable batteries with 3 overcharges followed by complete discharges in a row - despite that the provable rationale for this advice expired with ni-cad tech, and with modern tech it actually significantly degrades the longevity of your batteries | 05:53 |
| gnarface | (openbsd still encourages you to leave unpartitioned space, but their distro only takes like 80MB fully installed and their rationale is not related to wear leveling) | 05:55 |
| gnarface | se7en: people do use SSDs for their swap these days as though it's no problem, and even i'm still leery about that, if you only have one harddrive slot you might consider checking out the zramswap package | 06:01 |
| gnarface | *sorry, zram-tools package | 06:01 |
| gnarface | to use zram for swap | 06:02 |
| gnarface | it's becoming a popular solution for mobile devices | 06:02 |
| rustyaxe | i dont understand why people still use swap -- besides zram. | 06:02 |
| rustyaxe | If your machine is swapping, theres a process causing trouble.. Just kill that process. | 06:02 |
| rustyaxe | we're not in the days of 1 or 2gb of ram anymore | 06:03 |
| gnarface | sometimes it's still nice to have a net in case you fall | 06:03 |
| rustyaxe | I'd rather OOM killer dispatch with the broken process than trash some write cycles on my SSD in the process before eventually killing the offender | 06:03 |
| gnarface | yea, but then it OOM kills your world of warcraft instance in the middle of a rated match | 06:04 |
| rustyaxe | thats why the have vidja game machines that plug in the tv | 06:04 |
| gnarface | hehe | 06:04 |
| rustyaxe | but either way | 06:04 |
| rustyaxe | if it ate all the ram like that.. swap aint saving the day | 06:04 |
| rustyaxe | its just delaying the inevtiable OOM killer and restart of hte offending process | 06:04 |
| rustyaxe | 20 years ago? I'd tell you swap was quite important.. Today? Swapping to any form of persistant media is a waste | 06:05 |
| gnarface | eh, i think most people will be better off just setting /proc/sys/vm/swappiness to 0 and using a small swap partition rather than none | 06:22 |
| gnarface | but using zram for that swap partition can be a real speed boost, especially if your only other choice is like a microSD or something (for which, aside from being super slow, i think the wear-leveling concerns may still apply for many models in the wild) | 06:24 |
| onefang | I have 256 GB of ram and use the swapspace program to create swap files when needed. It was actually needed once. Gave me time to figure out what I had done to need it. | 10:18 |
| clemens3 | I have enough swap so oomk never hits me.. of course, i should disable oomk from source, too lazy so far.. oom killer is imho by design criminal.. | 12:08 |
| clemens3 | software should be designed to request memory, and if it is designed, print a message and exit e.g... then you can stop some other program and restart.. | 12:10 |
| clemens3 | if it is all on server fully automated, why you want your OS to kill some random process and think that you have a stable well defined system.. | 12:10 |
| clemens3 | designed=well designed | 12:12 |
| FatPhil | having some swap space has very little overhead, so there's never a reason not to just have a little bit of headroom available. | 12:19 |
| gnarface | it's not just the extra headroom, some stuff just literally doesn't work right when it doesn't have the secondary bank of space to actually swap to | 12:35 |
| gnarface | so yea, it's important to have even if you're only giving it 128MB | 12:36 |
| gnarface | these days we usually have enough physical ram that you can avoid most those cases | 12:37 |
| gnarface | but it's possible you might still occasionally be causing yourself unnecessary drama if you have literally 0 | 12:38 |
| gnarface | you wouldn't for example imagine apt as something memory intensive, but i remember it would at one point crap out without swap if your update was more than the amount of physical ram you had | 12:39 |
| gnarface | i don't know if this has since been addressed, and even at the time there was a non-default option you could put in the config to mitigate the issue | 12:40 |
| clemens3 | well, you can have 256gb ram, i have 2 on an old thinkpad that just works perfect for irc and more | 12:40 |
| gnarface | but the point was, with any swap space at all, it was completely capable of finishing the update, even if the update was larger in size than your memory and swap total | 12:40 |
| gnarface | i'm not really explaining it well, but i think that some people who disable swap are genuinely never in such a situation, but many of them just don't realize what they're actually disabling | 12:41 |
| gnarface | they associate it with "slow, therefore bad" but fail to realize that it has a fundamental functionality aside from just being cheap overflow for your ram | 12:42 |
| clemens3 | to each their own.. let's say I have 32gb ram and 32gb swap, the other dude has 64gb ram... well, he is better off.. should i toss out the swap? well, why? to save my ssd... well, if i never use more than 32gb ram it won't be trashed much either.. if i use, good too, not? | 12:44 |
| gnarface | and many of those people, like me, didn't realize you can just set /proc/sys/vm/swappiness to 0 and it's just like not having swap except when you really do need it | 12:44 |
| clemens3 | ... i keep my swaps.. | 12:44 |
| gnarface | heh, i keep my spinning rust just for swaps | 12:45 |
| clemens3 | well, if swap got filled, then you needed it, not? | 12:45 |
| gnarface | when i replace a metal drive with a ssd i keep the old drive for swap alone | 12:45 |
| clemens3 | i have yet to replace a single ssd.. i must say | 12:46 |
| gnarface | (and sometimes backups, you can never have too may backups) | 12:46 |
| clemens3 | a few hdds went over the jordan river | 12:46 |
| rustyaxe | clemens3: No if swap got filled you need more RAM or fix your leaking process :o | 12:50 |
| gnarface | yea, admittedly if you're not using hibernation you probably don't really need to double your ram these days on a modern machine | 12:52 |
| gnarface | and it's worth noting that the last time anyone with anything resembling any sort of official authority was willing to say how much swap you actually need it was "double your ram up to 128MB" | 12:53 |
| gnarface | (from the redhat installation guide sometime in the 90's when they used to actually print you a physical copy an include it with the cd) | 12:53 |
| djph | ... I think my first PC I boult I splurged and bouth 512M | 12:54 |
| gnarface | aside from hibernation, the only time i've ever actually used up all of 8GB swap was indeed world of warcraft, and it was indeed a memory leak | 12:54 |
| djph | *bought, even | 12:54 |
| clemens3 | rustyaxe: true, i meant used instead of filled.. | 16:21 |
| tom-1 | Hello . I tried to install the system without Dbus and was very surprised when, after installing several programs, I found that Dbus automatically works)) Correctly, I understand that in order to use the system without Dbus, I need to check the dependencies every time and if I see that the program has a dependency, then do not install it . | 17:25 |
| tom-1 | https://www.devuan.org/os/documentation/dev1fanboy/en/devuan-without-dbus.html | 17:26 |
| tom-1 | maybe it makes sense to use for without Dbus dependencies? https://sourceforge.net/projects/slim-fork/ | 18:07 |
| tom-1 | s/for/fork/ | 18:09 |
| gnarface | hey tom-1 | 18:10 |
| gnarface | i think fsmithred had some experimental dbus-less builds of some things | 18:11 |
| gnarface | or a dbus-less install or something like that | 18:11 |
| tom-1 | gnarface: If I'm not mistaken, you already mentioned this. You don't mind giving a link ? | 18:15 |
| gnarface | i'll try | 18:16 |
| gnarface | here, tom-1: https://get.refracta.org/files/experimental/?2022 | 18:17 |
| gnarface | this was also linked: https://dev1galaxy.org/viewtopic.php?id=2158 | 18:17 |
| gnarface | your assessment about having to "watch" it sounds correct | 18:18 |
| gnarface | i don't know how much of that issue carries over to the live images, i haven't tested them myself | 18:18 |
| gnarface | if none of that does what you want, i don't have anything better other than the possible suggestion to just disable dbus and leave it installed and see what breaks | 18:19 |
| tom-1 | gnarface: in my case it was like this, I set up a minimal system without Dbus as it was written on the site and out of habit I installed some program, I don’t remember which one)) then I checked `rc-status` and saw that Dbus was working. I deleted again and half of the system again had to be installed again)) Therefore, I wanted to understand the scheme of how best to be in order not to constantly reinstall the system from scratch | 18:25 |
| gnarface | tom-1: oh, you can just use apt pinning to block dbus from being installed, then when you request a program that would install it you just get an error instead | 18:26 |
| fsmithred | pin dbus to -1 so it can't be installed | 18:27 |
| gnarface | that won't fix the fundamental error but at least it'll keep you from having to reinstall everything over and over | 18:27 |
| fsmithred | oh, I should have read the last post | 18:27 |
| fsmithred | put the following in a file in /etc/apt/preferences.d/ | 18:29 |
| fsmithred | https://dev1galaxy.org/viewtopic.php?id=5668 | 18:30 |
| fsmithred | no, not that | 18:30 |
| fsmithred | not even close | 18:30 |
| fsmithred | Package: dbus* | 18:30 |
| fsmithred | Pin: version * | 18:30 |
| fsmithred | Pin-Priority: -1 | 18:30 |
| fsmithred | ^^^ that | 18:30 |
| fsmithred | if you want to be really radical, use *dbus* instead of just dbus* but lots more packages won't install without the libraries. | 18:32 |
| FatPhil | Planning on pulling down my old desktop some time soon, and shoving the newly-installed OS (via QEMU) drive into it. Hoping it will be pwer down, couple of screws, cable cable, screws, power up, all running. However, I'm worried about the UEFI/grub aspect. the OS isn't on the primary drive, which is an on-board ssd. Will I need to do anything in the installer to get it to look for an OS on the new 2nd | 19:37 |
| FatPhil | disk? | 19:37 |
| n4dir | wouldn't any "update-grub" after everything is in place take care of that? | 19:41 |
| rwp | n4dir, There are at least three different partition layouts needed depending upon the boot type of the system. Legacy, UEFI, UEFI-CSM, and more. | 20:02 |
| rwp | FatPhil, Keep a rescue boot image available just in case! You can probably boot rescue media and then use it to bootstrap your image. Probably. It's sometimes tricky. | 20:03 |
| rwp | Unfortunately I have had very unpleasant experience with UEFI. It feels like the creators of that entire way of booting have conspired to make it fail to boot for me. | 20:04 |
| n4dir | rwp: as far i understand he wants grub to look for another OS. | 20:04 |
| rwp | For one all of the boot configuration is not on disk. It is coupled to boot information managed in non-volatile flash memory that must also match. | 20:04 |
| n4dir | i wouldn't know if you had to do that during the installation process, but can do it afterwards. But then: i hardly know UEFI | 20:05 |
| rwp | There is the "efibootmgr" tool which is the management interface to the non-volatile memory holding the boot configuration. | 20:05 |
| rwp | But most of my several UEFI systems all have various different behavior of buggy UEFI booting! | 20:06 |
| n4dir | from this one and only UEFI system i can't even boot from USB, have to use CD. | 20:06 |
| n4dir | what a crap | 20:06 |
| rwp | On one I can only boot the fallback image for example. | 20:06 |
| rwp | On that one I found "refind" to be useful. It compensates for buggy UEFI booting problems. | 20:07 |
| rwp | So to share multiple boot on one system all systems must have boot files installed in the ESP EFI System Partition *AND* the non-volatile memory needs to have proper entries to boot them. | 20:09 |
| rwp | Which means that in the old days one could simply move a disk from hardware to hardware. Those days are gone. | 20:09 |
| rwp | I find I must boot rescue media and then reset the non-volatile memory to the newly installed storage. And then _sometimes_ it boots. | 20:10 |
| * n4dir has foam at mouth | 20:10 | |
| rwp | One of my buggy systems that I boot the fallback will boot the assigned image *once* and once only, resetting it to a non-working default after every boot. | 20:11 |
| rwp | That one was very painful for me to figure out. Eventually found the Debian wiki page listing it as a bug. Tested specifically for that bug. Yes! I have that specific bug! | 20:12 |
| rwp | This is a useful reference https://wiki.debian.org/UEFI | 20:13 |
| n4dir | rwp: all this is so damn confusing, i guess whenever i will need it, i will ask others. | 20:13 |
| n4dir | but still thanks for the link | 20:14 |
| rwp | I have Problem2 as described here: https://wiki.debian.org/GrubEFIReinstall | 20:15 |
| rwp | That page is also another useful reference page for UEFI problems. | 20:15 |
| FatPhil | n4dir: if I'm able to run update-grub, I've already booted the system. | 20:57 |
| FatPhil | I have the installer media still, so I could try to boot off that and then just do the "install bootloader" bit | 20:57 |
| FatPhil | the uefi partition is on the on-board SSD, and the prior devuan on the old disk was an older different one from what I put on the new disk, so I don't expect it to just work first time. | 20:59 |
| rwp | FatPhil, The installer has a Rescue Mode under Expert actions. Rescue Mode is one of the things I use it for the most. Or the refracta live-boot also. | 21:01 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!