Other Distros

[Billtoo]

Installed Linux Mint 17.2 Rafaela to the hard drive of a hp desktop pc.

Operating System
Version
Kernel Linux 3.16.0-38-generic (x86_64)
Compiled #52~14.04.1-Ubuntu SMP Fri May 8 09:43:57 UTC 2015
C Library Unknown
Default C Compiler GNU C Compiler version 4.8.4 (Ubuntu 4.8.4-2ubuntu1~14.04)
Distribution Linux Mint 17.2 Rafaela
Desktop Environment MATE (mate)

Monitors
Monitor 0 1920x1080 pixels
Monitor 1 1920x1080 pixels

OpenGL
Vendor Advanced Micro Devices, Inc.
Renderer AMD Radeon R7 200 Series
Version 4.4.13374 Compatibility Profile Context 15.20.1013
Direct Rendering Yes

It's working well.

EDIT:I installed Linux Mint 17.2 Rafaela Mate to the hard drive of another
desktop pc.
Also installed XFCE4 and KDE window managers with synaptic.
This release gets updates until 2019.
It's working well on this pc too.

[prehistoric]

I've been distro-hopping with a purpose, though not one others here may have considered. (I'd post screenshots if I could get both installation and networking up at the same time. I'm sure you've seen screenshots of Quirky 7 April 64.)

Recently, I built a machine with 8-cores I had no excuse for. ("But the used parts were so cheap!")
It has a Gigabyte GA-970A-DS3P rev. 1.0 motherboard with an AMD 8320 processor. There is a rev. 2.0 board, so I'm guessing this one is not suitable for overclocking. It seems OK if I stick to specs. Some distros are more tolerant than others. I'm not worried by problems which show up with high-performance video cards; I don't play games. I'm even willing to downgrade to an old GeForce card which is very stable.

I checked basic reliability by using it to compute Pi to 5,000,000,000 decimal places.

I'm using this as a testbed for problems we have with installing on newer machines with a UEFI bios and GPT disks.

At the moment I'm booting from either an SSD small enough to use a conventional MBR or a flash drive. I have terabytes of rotating storage. As much as possible, I would like to dedicate resources on this machine to computational tasks, not eye candy or compatibility. For this reason I've been running BarryK's Quirky 7.

An unexpected problem cropped up when the gigabit Ethernet using a RT8169 chip repeatedly failed to connect, on a line my other machine uses all the time. I've even connected two 50' CAT5 cables end-to-end to go directly from my router to the machine without using the powerline adapter I typically use.

After I traced the problem to an error reported in the dmessage, we decided it must be a kernel issue, because this did not show up with BarryK's April 64 Quirky 7.0.4.1, which has a 3.19.2 kernel. I've now found the same problem in a distro with a 4.0x kernel.

Even stranger, when I stuck in an old Netgear PCI card with a 100 Mb/s Ethernet interface it had a problem in networking with the same distros that failed on the RT8169. I'm now looking for a problem in initialization and DNS handling.
(One discovery I've made is that some distros ignore having the built-in interface disabled in the BIOS. This is a problem for people who get a buggy chip, and need to work around with a separate card.) I may need to tweak PCI bus latency.

I've concluded that BarryK's networking is simply more robust than some very popular distributions, and I'm trying to track down the offending code, and the reason it is in so many distros. Most distros with automatic network setup have no GUI for problematic machines, and for debugging you go directly to CLI tests which vary between distros. Simply telling me which interfaces are available, and asking which I want to configure, without demanding CLI skills specific to the distro, is a big help. Quirky 7 is also the fastest and least resource hogging.

My standard vanilla heavyweight Linux is Linux Mint 17.2 "Rebecca". This installed on this machine, but had the Internet fail, even when it claimed this was connected. Ubuntu 15.04 ran into some problem which prevented easy installation, and I'm not enough of a guru to get around it easily. The Fatdog 700/701 I'm using at the moment on an older machine is the one which had the dmesg showing network problems during boot. (On a recent attempt it failed to boot from DVD to a live desktop, probably because of video. I can work around this, but that still leaves networking.)

The distro with the latest kernel which installed properly was Sparky Linux 4.0, a rolling-release based on Debian unstable. It also had the network problem. Similar problem with Sabayon 15.06, another rolling release.

The 64-bit version of Vector Linux, VLocity 7.1, had multiple problems which stopped me from installing. After experimenting with it I had a boot failure in which it appeared the processor clock was set to 4.0 GHz. That processor is only rated for 3.5 GHz. It might be tweaked carefully to get to 4.0 GHz, but I am not trying. Reset the BIOS and it booted without problems.

I've used the distros which installed themselves properly to get a working bootloader, while waiting for BarryK's version that handles that problem itself.

This machine may have buggy I/O, depending on how you use it, but there is an enormous range of behavior between different distros. On most, tracking the problem down looks like an ordeal.

Added: now posting from April 64 Quirky 7.0.4.1 booted from SSD on that machine. I used Sparky linux 4.0 to install a bootloader (grub2) then added a custom menuentry using clues from the alternate Grub4Dos entry suggested by Barry to get the uuid. This was originally 40_custom in /etc/grub.d; moved to 7_custom after checking that it worked to put Quirky at the top of the list.

Code: Select all

#!/bin/sh
exec tail -n +3 $0
# This file provides an easy way to add custom menu entries.  Simply type the
# menu entries you want to add after this comment.  Be careful not to change
# the 'exec tail' line above.
menuentry "Quirky April64 7.0.4.1 frugal in sda2 dir april64-7.0.4.1"{
set root='(hd0,2)'
search --no-floppy --fs-uuid --set c02f47ec-bc89-4742-af5e-97f54efa5769
linux /april64-7.0.4.1/vmlinuz
initrd /april64-7.0.4.1/initrd.q
}

Naturally, I had to run update-grub from a root console within Sparky linux to execute the update scripts. This strikes me as a problem if you are trying to recover a damaged installation.

Still not a clue as to why this machine seems solid while running Quirky, and problematic while running other distros.

[wyzguy]

Prehistoric

On newer Gigabit motherboards with rtl8169 problems, try changing
IOMMU in the bios to enabled. If that does not work, or has some undesirable side effect, try adding the iommu=soft kernel boot parameter.

[prehistoric]

Prehistoric

On newer Gigabit motherboards with rtl8169 problems, try changing
IOMMU in the bios to enabled. If that does not work, or has some undesirable side effect, try adding the iommu=soft kernel boot parameter.

Thanks for the suggestion. I'm on a different machine at the moment, but will try that right away.

In the past, I've been able to salvage some machines declared defective by adjusting things like PCI latency in BIOS settings. I now see that option no longer available, though I'm getting evidence of device timeouts, and not just on that one chip.

I started this exercise in an attempt to prepare for the day when most machines will not have MBR disks, and vendors will make another attempt to freeze out those who want to boot systems they have compiled themselves. Last year's disclosures about widespread fundamental vulnerabilities in USB devices cause me to suspect we will have problems in booting from external USB devices after the next generation of security measures.

(This morning's news doesn't make the security situation look better, certainly not for individuals.)

The W10 preview seems to be aiming toward a white-list approach to security, with that list of certificates for approved programs coming from centralized authorities who have strong economic reasons for anti-competitive behavior.

At present I am unable to boot on a machine with all GPT disks without some workarounds that may well be disabled by later security updates. We haven't yet been frozen out by innovations, but I can easily see the "security through obscurity" approach gaining ground. This benefits large organizations over individuals and small organizations.

Reliance on identical code from a single source has made attacks on current security measures highly profitable. The only long-term solution is to get rid of the idea of a world-wide monoculture of code, and encourage diversity of implementations, making the benefits of any particular hack less attractive, and raising the costs of finding vulnerabilities common to many systems.

Added: Now posting from the problem machine. The iommu is definitely part of the solution. I knew something had changed when it tried to boot from a non-existent PXE server on that network.

(I don't recall ever telling the BIOS to boot from PXE, but that is apparently an option inserted automatically if other boot options fail. Other BIOS strangeness involves an entry concerning a network stack on a different tab from the entry for the IOMMU. What is more it is off the first visible page on that tab, making it easy to overlook.)

This is the first I've heard about the kernel option iommu=soft.

I'm getting too old to deal with steadily increasing confusion about the minimum requirements to get a computer to do anything useful. I have no idea how people without years of IT experience are supposed to cope.

Part of my collection of IT junk is a router which had been remotely modified to redirect all searches to an engine which paid the perpetrators. I have yet to find one which directs Internet banking to a site run by a perp, but I know these are out there.

Most of the computers people give me are not actually broken. I've returned several machines to people, after finding why they couldn't use them. One case with true hardware problems ran fine after I removed several pounds of dust, replaced the thermal paste on the heatsink for the chipset and the one on the video card, and reinstalled the OS.

Part of the problem we have is that IT people can make money by selling new equipment to replace equipment the owner simply does not understand. Obfuscation pays.

I am no longer clear on the difference between legitimate businesses which sell software which limits your access to data you create, and ransomware.

[prehistoric]

Just to make up for anything bad I said about Sparky Linux 4.0, here is a screenshot of it running on that problem machine. The problem really was the IOMMU setting in the BIOS.

[prehistoric]

I can now add Ubuntu 15,04 and Linux Mint 17.2 to the list of systems which work now that I've fixed the BIOS setting.

Getting all these to play nicely together with a multiboot is still a work in progress. Every distribution wants to control the boot manager.

[Billtoo]

I installed Mint 17.2 on my Acer Aspire laptop:

Computer
Processor 4x Intel(R) Core(TM) i5 CPU M 430 @ 2.27GHz
Memory 3907MB (567MB used)
Operating System Linux Mint 17.2 Rafaela
User Name bill (Bill)
Date/Time Fri 10 Jul 2015 12:00:55 PM EDT
Display
Resolution 1600x900 pixels
OpenGL Renderer AMD Mobility Radeon HD 5000 Series
X11 Vendor The X.Org Foundation

Current Session
Computer Name bill-Aspire-7740
User Name bill (Bill)
Home Directory /home/bill
Desktop Environment XFCE 4

OpenGL
Vendor Advanced Micro Devices, Inc.
Renderer AMD Mobility Radeon HD 5000 Series
Version 4.4.13374 Compatibility Profile Context 15.20.1013
Direct Rendering Yes

Dual booting LMDE2 Betsy on this laptop.

[James C]

AntiX 15-V-x64 base version tracking unstable Sid.

Code: Select all

james@antix1:~
$ inxi -F
System:    Host: antix1 Kernel: 4.0.5-antix.2-amd64-smp x86_64 (64 bit)
           Desktop: Fluxbox 1.3.5
           Distro: antiX-15-V_x64-base Killah P 30 June 2015
Machine:   Mobo: ASRock model: N68-S UCC
           Bios: American Megatrends v: P1.70 date: 03/03/2011
CPU:       Dual core AMD Athlon 64 X2 5200+ (-MCP-) cache: 1024 KB 
           clock speeds: max: 2712 MHz 1: 2712 MHz 2: 2712 MHz
Graphics:  Card: NVIDIA GT218 [GeForce 210]
           Display Server: X.Org 1.16.4 drivers: nouveau (unloaded: fbdev,vesa)
           Resolution: 1440x900@59.89hz
           GLX Renderer: Gallium 0.4 on NVA8 GLX Version: 3.0 Mesa 10.3.2
Audio:     Card-1 NVIDIA MCP61 High Definition Audio driver: snd_hda_intel
           Card-2 NVIDIA High Definition Audio Controller
           driver: snd_hda_intel
           Sound: ALSA v: k4.0.5-antix.2-amd64-smp
Network:   Card: NVIDIA MCP61 Ethernet driver: forcedeth
           IF: eth0 state: up speed: 100 Mbps duplex: full
           mac: 00:25:22:61:55:fc
Drives:    HDD Total Size: 1250.3GB (1.3% used)
           ID-1: /dev/sda model: WDC_WD10EZEX size: 1000.2GB
           ID-2: /dev/sdb model: ST3250318AS size: 250.1GB
Partition: ID-1: / size: 49G used: 2.2G (5%) fs: ext4 dev: /dev/sda9
           ID-2: swap-1 size: 8.49GB used: 0.00GB (0%) fs: swap dev: /dev/sda5
           ID-3: swap-2 size: 6.40GB used: 0.00GB (0%) fs: swap dev: /dev/sdb5
Sensors:   System Temperatures: cpu: 36.0C mobo: N/A gpu: 36.0
           Fan Speeds (in rpm): cpu: N/A
Info:      Processes: 136 Uptime: 1:08 Memory: 603.6/3955.2MB
           Client: Shell (bash) inxi: 2.2.25 

[Bindee]

Need a slim-line work environment? We recommend the best lightweight Linux desktop worlds for extreme open-sourcers.

http://www.theregister.co.uk/2015/07/09 ... t_linuxes/

LOL

[rokytnji]

Need a slim-line work environment? We recommend the best lightweight Linux desktop worlds for extreme open-sourcers.

http://www.theregister.co.uk/2015/07/09 ... t_linuxes/

LOL

No ram usage there. Kinda sparse reporting I guess. Before I would Crunhbang++. I would probably (and just might anyways) give this baby a spin.
http://sourceforge.net/projects/crunchbangmonara/

The dude is from Poland and nice enough to make a go of improving Crunchbang on his own to Jessie and systemd compliant and used.

I am a AntiX die hard though. Surprised the register gave AntiX no love.
But this border rat rolls how he pleases any hows.

[darry1966]

New version of Astrumni 3.2.0 - released: ftp://austrumi.ru.lv/

Anyone tried it?

[Billtoo]

I installed antiX 15 to a USB-3.0 SSD with the installer on the live
cd,it's running on my macmini.

EDIT:

My macmini install began to have problems so I did a new install to the
hard drive of an older hp desktop pc.

Computer
Processor 2x Intel(R) Core(TM)2 Duo CPU E4500 @ 2.20GHz
Memory 4040MB (319MB used)
Operating System antiX 15
User Name bill (Unknown)
Date/Time Mon 13 Jul 2015 12:29:52 PM EDT
Display
Resolution 1920x1080 pixels
OpenGL Renderer Mesa DRI Intel(R) G33
X11 Vendor The X.Org Foundation

Current Session
Computer Name antix1
User Name bill (Unknown)
Home Directory /home/bill
Desktop Environment XFCE 4

It's working great on this pc.

[Colonel Panic]

I'd had a couple of problems wirh Scientific Linux 7.1 , but I've now installed one of its predecessors (SL 6.5) and it's working fine with no problems. The CentOS 6.x series looks rock solid (Stella 6.x, which is also based on CentOS, is fine too).

EDIT; it still worked fine after I ran su -c 'yum update' on it to update the distro from the command line.

[musher0]

Hello, all.

FYI: According to DW, a new slackware version is just out.
(See date at the top left, under the title.)

BFN.

musher0

[prehistoric]

Now running Sabayon Linux 64 15.06 live on that 8-core box that gave me problems before.

I'm still trying to set up a multiboot system where I can easily switch between Linux distros, and having problems with automatic installers and the ESP partition needed with GPT disks.

I found a comment on the Sabayon forum which finally explains a good bit of the trouble with various Linux distros and that special partition needed with UEFI systems and GPT disks:

The ESP partition: How many are needed, where, and how big should it be?
Different sources on the internet says different things. Microsft only support one such partition – and that partition needs to be on your boot disk. MS also recommends it to be the first partition on your disk. As for size, according to Rod Smith (see link above) Windows installers create 100MB ESP partitions, Macs create 200MB, while Linux utilities gets confused with FAT16/32 setups if partitions are smaller than 520MiB, so he recommends creating 550MiB for the ESP. My system came with a 300MB ESP partition.

I had no idea I needed a 550 MB partition to avoid getting this confused with ordinary FAT32. There are work-arounds for the problem, even on some systems where automatic installers fail, but different Linux systems may not recognize this special partition the same way. If you have a Windows installation, these installers will likely get the information they need from that, but they are much less competent at figuring out what is going on on machines with other Linux systems and without Windows.

I've already noticed that Ubuntu 15.04 creates a boot partition with an EXT4 filesystem, if you don't have Windows, and give it the whole disk.

Linux Mint 14.2 creates a 512 MB FAT32 partition with the ESP flag. This conflicts with Ubuntu, and falls below the size some Linux utilities need on other distros. It is hard enough to multiboot with Linux distros that have dedicated partitions. Puppy versions that do not require a partition are simply not going to be picked up by any installer that searches for existing installations during automatic configuration of Grub2. You need to install them last. I've posted some code for one such installation here.

[James C]

Fresh install of Debian Jessie 8.1 Mate x86-64.
https://www.debian.org/

Code: Select all

james@debian:~$ uname -a
Linux debian 3.16.0-4-amd64 #1 SMP Debian 3.16.7-ckt11-1 (2015-05-24) x86_64 GNU/Linux

Code: Select all

james@debian:~$ free
             total       used       free     shared    buffers     cached
Mem:       4058672     999476    3059196      48016      23268     352016
-/+ buffers/cache:     624192    3434480
Swap:      8294396          0    8294396

[8Geee]

Hello, all.

FYI: According to DW, a new slackware version is just out.
(See date at the top left, under the title.)

BFN.

musher0

Thanx for this, but I went to slackware org and found nothing, not even a pre-announcement. What gives?

[musher0]

Hi, BGeee.

I have no idea. I just chanced on that info while checking out the distro
list at DW. I was a bit surprised myself, because usually when a major
distro puts out a new version, DW has an article on its main page.

Maybe it's a typo by the DW people?

As an aside, now would be a good time for a new Slackware version
(after two years).

BFN.

musher0

[8Geee]

Agreed, time for 14.2 or 15.0 or whatever, much has changed. I'm sure a LOT of effort and cleanup went into it.

Chers

[prehistoric]

I don't think the Slackware change is really a new release. They seem to have just included some fixes in their distribution disk for recent security vulnerabilities. If your secure networking was OK before, you could just update from an earlier version, but this is exactly the problem with several recently announced vulnerabilities.

I think Slackware just did this to avoid continuing problems with people downloading a vulnerable version, and reporting difficulties.

If anyone has inside information, I'd like them to post that. What I'm doing is speculating based on what I've noticed in that ISO image.