0 users browsing Discussion. | 1 guest | 10 bots  
    Main » Discussion » (Mis)adventures on Debian ((old)stable|testing|aghmyballs)
    Pages: First Previous 2 3 4 5 6 7 8 9 10 11 12 Next Last
    Posted on 20-11-08, 23:26

    Post: #102 of 105
    Since: 11-13-19

    Last post: 1224 days
    Last view: 1224 days
    The best part of client side physics is when they're being fed by another player on a laggy connection, then you start to see horribly jittery client side physics.
    Posted on 20-12-10, 16:13 (revision 1)
    Dinosaur

    Post: #837 of 1282
    Since: 10-30-18

    Last post: 4 days
    Last view: 5 hours
    Not Debian-related news, except that they should get ready for a sudden influx of CentOS/Red Hat refugees these days (some of them even former Debian defectors coming back to home!), because someone at IBM lost their fucking reason:
    https://lists.centos.org/pipermail/centos-announce/2020-December/048208.html
    https://www.phoronix.com/scan.php?page=news_item&px=CentOS-8-Ending-For-Stream
    https://news.ycombinator.com/item?id=25345428

    CentOS 8 (the current stable) is being suddenly EOL'd in late 2021 instead of having the usual long-lived 10 year LTS, as Red Hat (which now operates CentOS instead of being upstream, something I wasn't aware) decided to switch to a weird rolling release model, CentOS Stream (which resembles more like an outdated Debian Testing - where Fedora is more like proper Testing, and Rawhide would be Sid) to please the hipster container and Javascript frat boys crowd that can't get their abominations running on CentOS stable without recompiling the universe (which clearly defeats the purpose of running a Linux distro in first place). This doesn't affect CentOS 7, whose support ends in 2024.

    Needless to say, graybeard sysadmins of the world (who either run stuff at home or work for non-profits or whose employers are too cheap to pay for legit RH) are rising in arms, loudly claiming that CentOS committed suicide, that RH/IBM is dead for them, and threatening to defect to any of the following:

    - Orrible® Linux (it's yet another RH clone, albeit with a custom kernel, and of course those guys are being irrational to suggest migrating to a product controlled by One Raging Asshole Called Larry Ellison)
    - Debian, which is ironic because a lot of (now former) CentOS users had left Debian years/decades ago!
    - Ubuntu, because the irrational knee-jerk reactions have no bounds! (you would be pretty lunatic to run Ubuntu on a server - just run Debian, people!)
    - Surprisingly I haven't heard votes of confidence towards Linux defectors' camp (the *BSDs) during this crisis.
    - Vanilla RH (!!!), because it's now free as long as you don't want support.
    - Since this is FOSS, the natural way out is a fork (and there is already a project for that, Rocky Linux). Good Luck With That™ (yay more fragmentation~)

    2020 just keeps giving on and on and on... and it's still not over yet!

    Licensed Pirate® since 2006, 100% Buttcoin™-free, enemy of All Things JavaScript™
    Posted on 20-12-14, 00:58
    Full mod

    Post: #427 of 443
    Since: 10-30-18

    Last post: 864 days
    Last view: 60 days
    Bad news if you liked the Chrome UI but wanted to use it on Debian without all the Google integrations: Chromium has been removed from Debian Testing, meaning it won't be in the next official Debian release (unless something changes between now and then). This sometimes happens when dependencies don't update in sync (say, an app gets removed because it depends on a library with a security issue, but as soon as the library is fixed, the app is rebuilt and re-uploaded), but one of the issues listed is "Unsupported version, many security bugs unfixed", and, uh, yeah, there's a lot of them.

    I guess your options are to install a branded Google Chrome build, or switch to Firefox or, uh... GNOME Web?

    The ending of the words is ALMSIVI.
    Posted on 21-01-27, 20:26 (revision 2)
    Dinosaur

    Post: #896 of 1282
    Since: 10-30-18

    Last post: 4 days
    Last view: 5 hours
    How to enable SSH X11 forwrding, 2021 edition.

    You need to do this:
    #/etc/ssh/sshd_config
    # ...blah blah blah...
    X11Forwading yes
    X11DisplayOffset 10
    X11UseLocalhost no # the commented default is "yes"!
    # ...blah blah blah...

    Restart sshd on the host (restarting sshd over SSH is quite the chicken-and-egg solution, so the easiest way is to reboot the remote host if you don't have physical access), then "ssh -X user@host" will work, despite the insistence of those newfangled nerds to exorcise X11 because it's OOOOOOOOOOLD, DANGER MINES UNSAFE!!!, and Wayland is going to save us tomorrow, for real this time!.

    For extra scariness level: say you need to run some remote X11 app with superuser privileges (before you scream "DON'T DO IIIIIIT!!!!", for a legit case I've got this):

    sudo xauth add $(xauth -f ~/.Xauthority list|tail -1)


    Older distros didn't required to do any of this (except maybe for the xauth bit?), but we're living in horrible times so I've already went through that pain for you. You're welcome~

    Licensed Pirate® since 2006, 100% Buttcoin™-free, enemy of All Things JavaScript™
    Posted on 21-01-28, 02:51 (revision 1)
    Dinosaur

    Post: #897 of 1282
    Since: 10-30-18

    Last post: 4 days
    Last view: 5 hours
    Access Denied
    You don't have permission to access "http://download.virtualbox.org/errors/otnembargoed.html" on this server.

    Reference #18.4f7b0660.whatever.meh


    The time to get my shit off Orrible® VirtualBox has come.

    WTF, even the "Your shithole is embargoed" error is embargoed to me!

    [GAH!Politics]The Angry Cheeto is gone, but his legacy persists. And I guess that the new Democ-rat at charge will change nothing because all USAian presidents are the same garbage who can do nothing but token acts that only screw up everthing. And due to such fucking useless token efforts, Maduro is still a murderer dictator that will not go away anytime soon, yet I -the poor lowlife scum- end paying the price. Can't the planet just implode already!?[/GAH!Politics]

    Any libvirt/QEMU-for-dummies guide I can follow to migrate my stuff, pretty please?

    Licensed Pirate® since 2006, 100% Buttcoin™-free, enemy of All Things JavaScript™
    Posted on 21-01-31, 03:01 (revision 5)
    Dinosaur

    Post: #898 of 1282
    Since: 10-30-18

    Last post: 4 days
    Last view: 5 hours
    pgAdmin III got booted off Debian sometime last year

    This is because it has been abandoned upstream years ago, and it has already been unusable with PostgreSQL versions later than 9.6 (Buster ships with PostgreSQL 11, current version is 13). Their intended replacement? You've guessed it: pgAdmin 4, which is some unusable webshit abomination for Silly Valley's most beloved OS: Google Chrome web browsers! It's slow, bloated, SLOW, broken, hides essential info (like pg_dump/pg_restore logs which are often eaten by a grue) behind its flashy dashboard that it's SLOW AND BROKEN, and it even suffers of the Electron pandemic on its "desktop" version (as it is nothing but yet another Chrome-in-a-can™ webshit, ready to assrape your CPU cores and eat all of your RAM before breakfast). Even worse: when they initially released pgAdmin 4, they never came up with a plan to sunset pgAdmin III (something sensible would have been develop both in parallel until the newest shiny reached feature parity with the old stale, while maintaining the latter to add basic compatibility with new server versions until a date set in stone). Instead... they quickly killed pgAdmin III after releasing the first 4.x versions (which were even less usable than the current ones!), following the latest UX trends (does that remind you of someone? Like a certain desktop environment whose UI toolkit was formerly developed by a now former cellphone maker...)

    Yes, I've tried to embrace pgAdmin 4. As expected, it didn't worked for me. When you have your tools actively getting in the way, you know it's the time to look somewhere else. But with regards to PostgreSQL GUI-based management tools, your options are... not great:

    0) pgAdmin is the official management tool, endorsed and developed upstream. But while pgAdmin III was a lightweight TRUE desktop application that rarely got into your way, pgAdmin 4 seems to be designed to be as obnoxiously unusable as possible.
    1) There are other bigger, badder and better GUI tools, but none of them are FOSS, can cost some good money (which you're expected to pay if you are a professional developer making a living out of PostgreSQL, it's the fair thing to do... but not an option for inhabitants of communist countries, hobbyists, or FOSS purists)
    2) Use Access/LibreOffice... as long as you stick only to tables and nothing else. Forget about messing with stored procedures, triggers, views, or any of the cool features for big boys. Like being able to backup your database, or maintain it. Been there, quickly ran away as soon as Office ate my first table relationship, never got the T-shirt.
    3) Be a Troo UNIX® Way caveman, forget about GUIs, and do everything through the console. While having some CLI skills can save your life every now and then, expecting to become a 24x7 CLI freak is my idea of "oh fucking hell NOPE!".
    4) Resist. There is still hope for having sane, free tools. You've got the code for that, amirite?

    You guys knew what option I'm picking on this one. Luckily for dinosaurs like me, I wasn't alone back then: while some randos out there figured out what patches were required to bring pgAdmin III in compliance with PG10+, one of the PostgreSQL "partners" releasing commercially-supported versions, OpenSCG (now part of the Jeff Bezos' empire of doom) realized that while the fate was already set on stone for III, the way upstream handled it was a dick move, so instead they moved to implement their own gradual phaseout, and their custom PGSQL distro (under the "BigSQL" brand) was born, complete with their own fork of pgAdmin III. They never made standalone Linux versions of the latter, so for years I was using a hackjob of Debian's final pgAdmin III packages with some shoddy patches that somehow managed to not break things that much (my database only have 50 stored procs, not over 1000!)

    Fast forward to yesterday, where I finally got fed up of navigating into a list of hundreds of sprocs that are not mine and decided to take a look of what happened with good ol' pgAdmin III. While BigSQL is no more (although their GiggityHub is still active, but as expected their pgAdmin3-lts repo has been archived since 2017), the legacy lives on, and other dinosaurs are still keeping the flame alive. I found this fork of pgAdmin3-lts which brings things in sync up to PG13.1, and seems to work fine in my end. No more system sprocs littering my schema listings! No more mysterious error messages about some pg_internalvomit aborting all and any attempts to view a table! And best of all, NO FUCKING WEB BROWSER REQUIRED!

    But... I don't want it living as a mess of files stashed at either /usr/local or /opt. I need it integrated with my desktop. I need a Debian package. Long short story: here is how you can bake your very own pgAdmin3-lts Debian packages at home (WARNING: it's not the Debian Way®, but pretty close, and more patchy than poor ol' Patchouli Knowledge, but it Works For Me™):

    1) Clone AbdulYadi's repo. Make a tarball of it, just in case you screw up something.

    2) Get Debian's pgAdmin3 source package. You really only need pgadmin3_1.22.2-5.debian.tar.xz, but if unsure, just get all three files, put them on a new directory somewhere, and dpkg-source -x pgadmin3_1.22.2-5.dsc - we only want the /debian directory, which contains the recipe to build the .DEBs.

    3) On the pgAdmin3-lts sources dir, run ./bootstrap, to invoke some mandatory arcane autotools dance which usually already has been done on release tarballs, but not on code freshly cloned from trunk/master. Neglect this, and the build will die with mysterious errors at first, followed by "can't find branding/Makefile.in" on subsequent attempts.

    4) Copy the /debian directory from Debian's pgAdmim3 sources into pgAdmin3-lts sources (you can throw away the rest of the source tree). We can't use it as-is - we need to fix some stuff before attempting a build. In order:
    - 4.1) Go into /debian/patches and delete pg10 and pgversion patches (they aren't needed anymore, and will break the build!). Don't forget to edit series to remove those from the patch sequence too!
    - 4.2) There is a README that it's now README.md - that one will go into the pgadmin3-data package, so we must fix that or our build will die somewhere at 98%. Edit /debian/pgadmin3-data.docs and add the fancy .md extension to README there.
    - 4.3) Since this technically is the unreleased pgAdmin III 1.23 version, we need to set that too on the package to not conflict with the Debian version. This is (kinda counterintuitively) defined on /debian/changelog, which has a very specific format for that. Here is how mine looks now:

    pgadmin3 (1.23.0-1) UNRELEASED; urgency=high

    * Non-maintainer upload.
    * Switched to patched pgAdmin3 LTS by BigSQL.
    * Added compatibility with PostgreSQL 10-13.
    * Post-BigSQL compatibility patches from AbdulYadi/pgadmin3.

    -- Tom Maneiro <tomman@himawari.tomman.net.ve> Sat, 30 Jan 2021 13:44:00 -0400

    pgadmin3 (1.22.2-5) unstable; urgency=medium

    * Mark pgadmin3-data Multi-Arch: foreign.
    * Remove obsolete Build-Depends on devscripts.
    * Move maintainer address to team+postgresql@tracker.debian.org.

    -- Christoph Berg <christoph.berg@credativ.de> Wed, 06 Feb 2019 15:48:33 +0100
    ...snip...

    I'm sure Debian packaging toolchain has a tool or two to automate that. I don't care. I just obey the format and don't get into troubles, so you do too!

    5) We're ready to go! Ensure you've got installed all of the required build dependencies (libwxgtk3.0-dev being the main one, and yes, pgAdmin III still relies on the good ol' GTK+2 :), and invoke the build: dpkg-buildpackage -b -us -uc -tc -j4. Those switches are for the following:
    -b: Build binary packages only. We need that one because we don't have a true Debian source package anymore, and omitting it would cause dpkg-buildpackage to bitch and refuse to build from a source tarball that doesn't exist!
    -us, -uc: Do not sign sources/changelogs/whatever. Omit those and your build may abort at the end for some stupid reason regarding digital signatures you don't have (and don't want)
    -tc: Cleans the source tree after build, to not leave hundreds of megabyte of temporary object code junk lying around your filesystem!
    -j4: Adjust this for the number of cores/threads your computer has.

    6) Go eat something, water the plants, walk the dog, play something on your other PC - surprisingly, building pgAdmin III is a very taxing process: on my i5-2450M it takes over 30 minutes!

    7) You should now have 3 .DEBs outside the sources tree dir (alongside some other bunch of noise): pgadmin3, pgadmin3-data, and pgadmin3-dbgsym (this one can be safely ignored, unless you want to be able to debug crashes). Install them and enjoy one of the last bastions of sanity on desktop database management GUI tools!

    As for Debian and pgAdmin 4... that's just not happening anytime soon, due to the culture clash between old-school conservative distros like Debian (where every dependency has to be a package, and bundled libraries are a serious no-no) and the new kids on the block, the JavaScript Frat Boys (where everything is a moving target and they would put their mothers inside a Docker container if they could do so!). Yes, you've heard it from me first: current and next Debian releases have absolutely NO usable PostgreSQL GUI management tools! (Buster has a very broken pgAdmin 3 which doesn't really work with its companion PostgreSQL release, while Bullseye has NOTHING AT ALL) Progress!

    Licensed Pirate® since 2006, 100% Buttcoin™-free, enemy of All Things JavaScript™
    Posted on 21-01-31, 03:02
    Custom title here

    Post: #980 of 1150
    Since: 10-30-18

    Last post: 6 days
    Last view: 2 hours
    Posted by tomman

    WTF, even the "Your shithole is embargoed" error is embargoed to me!
    That's actually hilarious, in a terrible sort of way.


    Posted by tomman
    Been there, quickly ran away as soon as Office ate my first table relationship, never got the T-shirt.

    Was the shirt stolen in the mail?


    As for Debian and pgAdmin 4... that's just not happening anytime soon, due to the culture clash between old-school conservative distros like Debian (where every dependency has to be a package, and bundled libraries are a serious no-no) and the new kids on the block, the JavaScript Frat Boys (where everything is a moving target and they would put their mothers inside a Docker container if they could do so!). Yes, you've heard it from me first: current and next Debian releases have absolutely NO usable PostgreSQL GUI management tools! (Buster has a very broken pgAdmin 3 which doesn't really work with its companion PostgreSQL release, while Bullseye has NOTHING AT ALL) Progress!

    Speaking of hilarious in a terrible sort of way...

    --- In UTF-16, where available. ---
    Posted on 21-03-13, 17:49
    Dinosaur

    Post: #916 of 1282
    Since: 10-30-18

    Last post: 4 days
    Last view: 5 hours
    Kernel 5.10 (which is now available for us Debian Stable plebs via the magic of buster-backports) breaks TRIM on NTFS partitions mounted via ntfs-3g (what else?):
    https://bugzilla.kernel.org/show_bug.cgi?id=211167

    Workarounds:

    - Boot to kernel 5.9 when you need to TRIM your NTFS partitions.
    - Revert the offending patch as seen on the bugreport, then build your own kernel.
    - Consider migrating away from NTFS, if possible.
    - Wait for the next kernel release.

    Licensed Pirate® since 2006, 100% Buttcoin™-free, enemy of All Things JavaScript™
    Posted on 21-03-23, 22:52
    Post: #156 of 203
    Since: 11-24-18

    Last post: 9 days
    Last view: 6 days
    Posted by tomman
    Kernel 5.10 (which is now available for us Debian Stable plebs via the magic of buster-backports) breaks TRIM on NTFS partitions mounted via ntfs-3g (what else?):
    https://bugzilla.kernel.org/show_bug.cgi?id=211167

    Workarounds:

    - Boot to kernel 5.9 when you need to TRIM your NTFS partitions.
    - Revert the offending patch as seen on the bugreport, then build your own kernel.
    - Consider migrating away from NTFS, if possible.
    - Wait for the next kernel release.


    Wait, people still use NTFS on anything other than their Windows OS partitions? Didn't Windows get native file system support for ext4 a while back?
    Posted on 21-03-23, 23:40 (revision 2)
    Dinosaur

    Post: #919 of 1282
    Since: 10-30-18

    Last post: 4 days
    Last view: 5 hours
    Posted by wertigon
    Posted by tomman
    Kernel 5.10 (which is now available for us Debian Stable plebs via the magic of buster-backports) breaks TRIM on NTFS partitions mounted via ntfs-3g (what else?):
    https://bugzilla.kernel.org/show_bug.cgi?id=211167

    Workarounds:

    - Boot to kernel 5.9 when you need to TRIM your NTFS partitions.
    - Revert the offending patch as seen on the bugreport, then build your own kernel.
    - Consider migrating away from NTFS, if possible.
    - Wait for the next kernel release.


    Wait, people still use NTFS on anything other than their Windows OS partitions? Didn't Windows get native file system support for ext4 a while back?

    Many of us still dual-boot.

    MS never released EXT2/3/4 support, instead there are many 3rd-party drivers for it (my choice is Ext2fsd). But those are kinda useless if you want to store native Windows executables there, as UAC will not let run any executable not stored on any of the "blessed" filesystems: FAT/exFAT/UDF/ISO9660, and of course, NTFS. The list is hardcoded, so no chance to expand it, and your only other option would be to be fool enough to shoot yourself at your foot turn off UAC. I tried with UDF years ago, and sadly no OS really fully supports UDF for anything other than read-only optical media (I had written a long thread on the topic years ago on the long defunct bBoards)

    At least I can get POSIX file permissions on NTFS under Linux, but it's a minor hell on itself (suddenly you also have to deal with executable permissions under Windows too!), which is another hard requirement for me. But then, exFAT is supposedly a first class citizen under Linux these days (mostly thanks to Samsung), so I'll have to revisit the whole "filesystem interoperability" topic someday. Nevermind, it sounds like exFAT can't do file permissions at all, so I guess I'll be stuck with NTFS(-3G) until the day democracy comes back to Soviet Venezuela :/

    Licensed Pirate® since 2006, 100% Buttcoin™-free, enemy of All Things JavaScript™
    Posted on 21-03-24, 02:47
    Dinosaur

    Post: #920 of 1282
    Since: 10-30-18

    Last post: 4 days
    Last view: 5 hours
    Hmmm... I hadn't learned that Paragon is opensourcing their kernelmode NTFS driver for Linux. Even better: they're trying to get it included in the mainline kernel!

    https://www.theregister.com/2020/09/08/paragon_ntfs_linux/

    If this dream comes true, I can pretty much forget about the crippled mess of exFAT/UDF/whatever, also get rid of the massive performance hit of FUSE junk, and this will cement NTFS as the most useful cross-platform filesystem (at least for dual-booters), leaving ideological concerns aside (I don't care about those - I just want something that WORKS).

    Here are the things I need for a cross-platform filesystem, in recap:

    - POSIX permissions (the "rwx" model): NTFS-3G can emulate those through native NTFS ACLs, but it requires a mapping file (.NTFS-3G/UserMapping) to be created beforehand. They will screw up things a bit under Windows (you need to manually set your .EXEs as executable under either Windows or Linux, which for me is a nice security bonus)
    - Symbolic links: NTFS supports those natively, but for some reason, NTFS-3G instead relies on the long-gone Interix symlink format, which is alien to NTFS.
    - Large files (>4GiB): basically anything but FAT can do that nowadays.
    - Reliable filesystem checking tools: Sadly NTFS-3G is lacking those, but at least I get the option to reboot to Windows and use good ol' CHKDSK. Much better than my miserable experience with UDF, where Windows XP had no write support at all, while Windows 7 would turn the whole drive read-only if it found something weird, and their CHKDSK UDF support would completely REFUSE to even recognize UDF partitions on a 1TB drive! And Linux got nothing at all after all those years, because the "U" in UDF is for "Useless on anything that is not an ROM optical disc".
    - Kernelmode drivers: not an option for NTFS under Linux... but not for long?
    - Play nice with UAC: This one is on Microsoft - why in the hell they had to come up with a HARDCODED "approved filesystem list" to decide if you could run executables from any media?! This is the kicker that keeps me locked in to NTFS :/

    Licensed Pirate® since 2006, 100% Buttcoin™-free, enemy of All Things JavaScript™
    Posted on 21-05-11, 02:50
    Dinosaur

    Post: #943 of 1282
    Since: 10-30-18

    Last post: 4 days
    Last view: 5 hours
    Posted by tomman
    Kernel 5.10 (which is now available for us Debian Stable plebs via the magic of buster-backports) breaks TRIM on NTFS partitions mounted via ntfs-3g (what else?):
    https://bugzilla.kernel.org/show_bug.cgi?id=211167


    Aaaaand this bug got squashed upstream. For 5.10, you want kernel 5.10.24 at the very least, which thankfully is now shipping on Debian (linux-image-5.10.0-0.bpo.5).

    We now return to our scheduled programming of using filesystems on environments they weren't meant for.

    Licensed Pirate® since 2006, 100% Buttcoin™-free, enemy of All Things JavaScript™
    Posted on 21-06-05, 20:06 (revision 1)
    Dinosaur

    Post: #951 of 1282
    Since: 10-30-18

    Last post: 4 days
    Last view: 5 hours
    So today I went to update code from some random Giggityhub repo, just to met this:
    fatal: unable to access 'https://github.com/SOMEDEV/somecode/':Failed sending HTTP2 data


    A quick search unveiled the perils of sourcing your updates from Backports:
    https://superuser.com/questions/1642858/git-on-debian-10-backports-throws-fatal-unable-to-access-https-github-com-us
    Unfortunately the relevant bug report has been sitting dormant for nearly 2 months:
    https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=987187

    If you're still on Buster, the solution here is to downgrade the version from buster-backports to buster:
    sudo apt-get install libcurl3-gnutls=7.64.0-4+deb10u2


    This is the very same curl version that as of now will ship on the next Debian stable. "Can't clone Git repos" looks like a big fat release blocker issue to me, IMO. Someone suggested that this is a upstream bug, and a fix is now available on the next minor version, 7.76. But as we're in the middle of the ice age, this unfortunately looks like a no-go there :/

    And libcurl3-gnutls is a hard dependency for like half of my desktop setup, so you can't just get rid of it (there are alternate implementations of SSL support, aside of curl). WTF Debian. And WTF curl!?

    Licensed Pirate® since 2006, 100% Buttcoin™-free, enemy of All Things JavaScript™
    Posted on 21-06-05, 22:19
    Dinosaur

    Post: #953 of 1282
    Since: 10-30-18

    Last post: 4 days
    Last view: 5 hours
    Oh, if pulling stuff from backports, watch out for this too:
    https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=989231

    FFS Debian, it's not even your first rodeo!

    (hold onto libglvnd and friends for a while)

    Licensed Pirate® since 2006, 100% Buttcoin™-free, enemy of All Things JavaScript™
    Posted on 21-07-29, 00:29
    Dinosaur

    Post: #976 of 1282
    Since: 10-30-18

    Last post: 4 days
    Last view: 5 hours
    Debian 11 "Bullseye" will ship in August 14th, 2021 - that is, in two weeks!

    The libglvnd reverse dependencies bug on buster-backports remains open (yay deep freeze~), same as on the libcurl3-gnutls bug. Hope those two get fixed after release, but I'm not holding my breath... At this stage, I'll be moving to Bullseye Soon™, so these should become irrelevant for me in due time.

    Licensed Pirate® since 2006, 100% Buttcoin™-free, enemy of All Things JavaScript™
    Posted on 21-08-08, 22:48
    Dinosaur

    Post: #981 of 1282
    Since: 10-30-18

    Last post: 4 days
    Last view: 5 hours
    How to enable full-duplex on an good ol' Realtek RTL8029AS (NE2000 PCI clone)

    Short version: DON'T! Just buy/steal any 100/1000 PCI NIC, people. (For retroboxes, stick to NICs made at the turn of the millennium, like the RTL8139A, and NOT the RTL8139D, as it requires a 3.3V PCI slot despite what the Chinglish datasheet says!) But if you're stuck with ancient junk like me in SOCIALISM™, read on...

    Long version:
    Since this shit is barely documented (if at all), in the Troo UNIX® Way™ fashion, let me enlighten your path, for those that want a little bit of S&M in their lives:

    - Many of those NE2000 PCI clones support Full Duplex, unlike its ISA counterparts. But they're usually not enabled by default, you can't use ethtool/miitool for that stuff, etc, etc, etc.

    - Your module is ne2k-pci, and the magic enchantment for forcing full duplex is "options=32,32 full_duplex=1,1". Why duplicated? Dunno, as I only have one NE2000 clone here, and I had to make sense of the driver sources for understanding that detected card index starts at ONE. Yay. Record that magic string somewhere in /etc/modprobe.d/.

    - Regenerate your initramfs, otherwise your settings will be ignored at the next boot!

    - Reboot (or simply rmmod and modprobe the ne2k-pci kernel module)

    - Verify with ne2k-pci-diag (Debian package: nictools-pci, but it got dropped on Buster):
    ne2k-pci-diag.c:v2.06 2/28/2005 Donald Becker (becker@scyld.com)
    http://www.scyld.com/diag/index.html
    Index #1: Found a Realtek 8029 adapter at 0x6300.
    The 8390 core appears to be active, so some registers will not be read.
    To see all register values use the '-f' flag.
    Initial window 3, registers values by window:
    Window 0: 22 04 64 63 03 00 68 00 6a 63 50 43 01 00 00 00.
    Window 1: 62 00 06 4f 16 9b fe 64 00 00 00 80 00 00 00 40.
    Window 2: a2 00 80 ff 40 ff ff ff ff ff ff ff cc e0 c9 bf.
    Window 3: e2 30 ff 00 ff 00 70 ff ff ff ff ff ff ff 29 80.
    RTL8029 transceiver: 10baseT/coax (autoselected on 10baseT link beat) forced full duplex.
    0K boot ROM.

    No interrupt sources are pending (00).
    Parsing the EEPROM of a RTL8029:
    Station Address XX:XX:XX:XX:XX:XX (used as the ethernet address).
    Configuration is 30 00:
    0K boot ROM, half duplex
    Transceiver: 10baseT/coax (autoselected on 10baseT link beat).
    PCI Vendor ID 0xffff Device ID 0xffff
    Subsystem ID: vendor 0xffff device 0xffff
    Use '-a' or '-aa' to show device registers,
    '-e' to show EEPROM contents, -ee for parsed contents,
    or '-m' or '-mm' to show MII management registers.

    Yes, "forced full duplex" is what you want to see on the transceiver settings. If you're feeling hax, get the RTL8029AS datasheet and use ne2k-pci-diag to brick your card tamper with your EEPROM to force full duplex there too, change your MAC address, etc.

    - You'll notice that your network connection is now dead after using ne2k-pci-diag, so reset your interface (ifdown/ifup) to restore connectivity.

    Or simply forget about NE2000 clones - these were very shitty cards (both in ISA and PCI) even when they were new, and find some better stuff, like a 3Com, Intel, AMD, or whatever!

    Licensed Pirate® since 2006, 100% Buttcoin™-free, enemy of All Things JavaScript™
    Posted on 21-08-15, 18:15
    Dinosaur

    Post: #985 of 1282
    Since: 10-30-18

    Last post: 4 days
    Last view: 5 hours
    Mission: Aim for the Gold Bullseye: Up To 11!

    The new Stable is here! And it's time to image the HDDs and welcome the latest pile of slightly obsolete suck from upstream.

    My Asus will have to wait, but I'm now in the process of deploying the update on my Dell and IBM boxes. Of course, there are already some gotchas, judging from the apt-get failures and removal lists:

    - The URL for the security updates repository has changed, so be aware when editing your sources.list.

    - For graphical network managers, you've got two options: NetworkManager and wicd. Oops, wicd is no more on Debian, because Python 2 is dead, and so are its respective PyGTK bindings which haven't been updated since GNOME 3 launched! Supposedly there is work upstream to modernize wicd with Python 3, but so far all we've got is a buggybuggybuggy version on Experimental. I have nothing against NM - in fact, it's my choice on beefy systems with more than just "one RJ45 jack", but for resource-starved vintage boxes that will never ever go wireless, NM is bloaty and overkill. So yeah, my IBM box just got fucked raw there. Embrace NM's bloat, listen to modern CLI freaks telling me to move to systemd-networkd (which will never have a GUI because it's aimed at servers, and because Red Suck doesn't want competition for NM), or embrace my inner Troo UNIX® Way™ caveman and ifconfig like a boss. Yay.

    - The mupen64plus version from Experimental finally landed at Stable. Your PJ64_v1.x ROMhacks are now dead. Yay.

    - Default artwork now is Homeworld, so don't forget to reconfigure Plymouth before your next reboot.

    - BACKUP YO' SHIT, PEOPLE! Seriously, DO IT. NOW!

    - xsane is on the chopping block because upstream has vanished, and it barely survived for Bullseye. I guess kids these days don't use scanners anymore, and instead use their cellphones for everything :/ On the flip side, it seems Debian 11 finally supports driverless printing/scanning, which means it's time to buy new printers and scanners. Yay.


    Expect more ranting after my Dell and IBM finish installing nearly 5GB of updates, combined. This is the Largest Debian Update, EVER.

    Licensed Pirate® since 2006, 100% Buttcoin™-free, enemy of All Things JavaScript™
    Posted on 21-08-15, 20:07 (revision 1)
    Dinosaur

    Post: #986 of 1282
    Since: 10-30-18

    Last post: 4 days
    Last view: 5 hours
    1/3: the IBM TV box is done:

    - PulseAudio took over my SB Live!. AGAIN. POEEEEEETTERIIIIING!!!!!. Thankfully the same solution for Buster still applies on Bullseye: purge "pulseaudio" package, reboot, done.

    - I had no option but to use NetworkManager, which conveniently got shoehorned for my pleasure, including its GNOME-based tray applet. Suck! The tray applet can be enabled on MATE's "Settings->Personal->Startup", then tick the box to show hidden items (if nm-applet isn't there, run it once from a console).

    - I got a cute blowfish tray applet I didn't really requested for. Turns out it's a clipboard manager, "diodon". You can also disable it on Startup Applications.

    - The Great Conspiracy Against Cleanlooks still goes on: my window border of choice for MATE (Clearlooks-Phenix) now exhibits corruption just at the bottom of the title bar (a issue that I used to get only with apps using CJK chars on window titles, but not as bad as this time), so for the sanity of my eyesight, I had no option but to switch to Dopple (it looks VERY similar, but with a XPesque colored border, and a slightly thinner titlebar. SUUUUCK!)

    - Time to clear a bunch of Python 2 remnants...

    - Need to redo my LIRC configuration. AGAIN. Meeeeeeeeeeeeeh.....

    Licensed Pirate® since 2006, 100% Buttcoin™-free, enemy of All Things JavaScript™
    Posted on 21-08-15, 21:19 (revision 4)
    Dinosaur

    Post: #987 of 1282
    Since: 10-30-18

    Last post: 4 days
    Last view: 5 hours
    2/3: the Dell is done:

    - Got an IBus applet on my tray, including the dreaded "emoji imput" garbage. WTF IS THIS SHIT!? Oh, and it fucked with my keyboard layour (this Dell was configured and built for LatAm, not Spain!!!!). Plus my accented letters now are subject to an IME I don't need/want. Solution: run im-config, say OK a couple times, select REMOVE (to reset your setup to defaults), confirm, reboot, run im-config again, select "none", logoff, logon again, done.

    - GDM3 now succumbed to the moronic trend of rounded avatars, and it now looks more Win10 as hell. WTF IS THIS SHIT!??!?!

    - Clearlooks-Phenix is dead. The Too Poor To Buy A Mac™ design cartel is on the lam. I guess I'll have to get used to Dopple + TraditionalOK which provide a kinda convincing replacement... until the UXtards kill that, again. On the flip side, I don't need qt5gtk2 (which never got into the Debian repos) anymore for extending my GTK styles to Qt 5 applications - instead, just install qt5-gtk2-platformtheme (for whatever reason, qt5-gtk-platformtheme is there and uses your GTK3 theme, but it doesn't seem to work?), and use the Qt 5 settings control panel for select your style.

    - Surprisingly there is no need to reconfigure Plymouth's default theme - the new Homeworld theme will take over as soon as the very first reboot! Nice~

    - Longest. Update. EVER! Almost 4 hours to install 3.5GB of junk! What's this, Windows 10!?

    - Wammu is gone, which means no way to talk to my half-broken Motorola V9x remains on any Linux computer I use. Lovely.

    - Got some interesting dependency hell with Python 2 stuff (which is not totally gone yet!), Wammu, and VirtualBox - the latter two lost it. There are still no official VBox packages for Bullseye yet (not that I care that much as The Embargo™ forbids me from ever using VirtualBox ever again, according to Orrible®)

    - Samba no longer enables SMBv1 support by default, which is great for security... not so much for your vintage XP hosts at home. If you're feeling the need for crazy, here are the magical strings for opening the doors to hell making your vintage XP guests welcome:
    # /etc/samba/smb.conf
    server min protocol = NT1
    client min protocol = NT1

    Don't forget to restart smbd/nmbd!

    Licensed Pirate® since 2006, 100% Buttcoin™-free, enemy of All Things JavaScript™
    Posted on 21-08-19, 21:38
    Post: #77 of 77
    Since: 10-31-18

    Last post: 952 days
    Last view: 879 days
    Arch has a wammu AUR package, though it has only 1 vote and some commenters say they're running into build errors. Shame Arch is such a pain upfront to install "properly" (to find a workable local optimum out of the many choices for bootloaders, desktop environments, NTP daemons, networking, etc.).
    Pages: First Previous 2 3 4 5 6 7 8 9 10 11 12 Next Last
      Main » Discussion » (Mis)adventures on Debian ((old)stable|testing|aghmyballs)
      Yes, it's an ad.