Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us $3,400 per month for bandwidth bills alone, and since we don't believe in shoving popup ads to our registered users, we try to make the money back through forum registrations.
«733 »
  • Post
  • Reply
Odette
Mar 19, 2011



Thermopyle posted:

lol, am I on Slashdot in 2002

No, I'm just sick of Microsoft's lack of respect for it's users preferences.

I turned off ALL remote desktop permissions and sometime in the last ~4 months, an update turned these remote desktop permissions back on as well as all privacy settings back to full collection mode.

Not to mention that after every update, Windows turns Fast Startup back on, which breaks my access from Linux.

Adbot
ADBOT LOVES YOU

evol262
Nov 30, 2010
#!/usr/bin/perl

Windows expects to be the only OS.

I'm not gonna defend the choices they make, but it hasn't been unreliable since Vista (which was a new kernel and driver model, so that was expected). For whatever else you wanna say about Microsoft, BSODs are rare now.

And hey, at least they're better than Apple (or at least the iTunes team)

Varkk
Apr 17, 2004




evol262 posted:

Windows expects to be the only OS.

I'm not gonna defend the choices they make, but it hasn't been unreliable since Vista (which was a new kernel and driver model, so that was expected). For whatever else you wanna say about Microsoft, BSODs are rare now.

And hey, at least they're better than Apple (or at least the iTunes team)

Windows 10 seems to be bringing them back. I feel Windows reached a peak with 7 on the desktop and has been downhill since then. For the servers the R2 releases are still good. E.g 2008r2 and 2012r2 are great.

ToxicFrog
Apr 26, 2008



Harik posted:

I didn't see anyone call it "Micro$haft windurrrs" here so I think you're giving /. too much credit.

I'm running across ridiculous IO latency that I can't track down. 2+ second stalls on SSD. Very visible in things like tab completion - it just hangs for a while when it should be basically instant. BTRFS on 4.8.9. It's not consistent, even if I drop the caches and try to reproduce.

Are you sure btrfs isn't the problem? When I used it I was plagued with random slowdowns and things blocking on IO that really shouldn't be, although if it's only ~2 seconds for you it's been much improved. Switched to XFS and ZFS on the same disks and the problems went away.

quote:

The write load seems high (17TB), which is why I'm thinking something's rogue.

Try iotop or atop to see what's chewing on the disk? With atop you can also change the collect interval, so you could, e.g., tell it to collect stats over a five second window for real-time, iotop/htop style stats, or over a 24 hour window to see stats for the previous day.

evol262 posted:

Windows expects to be the only OS.

I'm not gonna defend the choices they make, but it hasn't been unreliable since Vista (which was a new kernel and driver model, so that was expected). For whatever else you wanna say about Microsoft, BSODs are rare now.

And hey, at least they're better than Apple (or at least the iTunes team)

You seem to be using a very narrow definition of "unreliable" that reduces to "doesn't kernel panic". And, to be fair, I haven't seen any BSODs in 10 yet, and didn't see any in years of running 7 that weren't directly attributable to hardware failures.

On the other hand, there's plenty of ways an OS can be "unreliable" that don't require a kernel panic, like not remembering settings or credentials you told it to (7), randomly resetting settings to the default (10), failing to install updates or requiring multiple days of continuous uptime to do so (7), loving with your boot configuration (both)...

ToxicFrog fucked around with this message at Dec 9, 2016 around 00:24

xzzy
Mar 5, 2009

wakey wakey to
this bowl of tasty


Yams Fan

Try using Bluetooth anything on windows. It's still a crapshoot and the process always ends up digging through control panels or the registry to try and convince it that your device is there.

I'm partially sympathetic because they have to run on so much different hardware and macOS only has a couple configurations to deal with, but not by much because Linux actually does a lot better on the exact same hardware that windows barfs on.

Hibernation is pretty sketchy too.

gourdcaptain
Nov 16, 2012



xzzy posted:

Try using Bluetooth anything on windows. It's still a crapshoot and the process always ends up digging through control panels or the registry to try and convince it that your device is there.

I'm partially sympathetic because they have to run on so much different hardware and macOS only has a couple configurations to deal with, but not by much because Linux actually does a lot better on the exact same hardware that windows barfs on.

Hibernation is pretty sketchy too.

To be fair, Bluetooth is sketchy on everything. See my Android phone trying to connect to a speaker hookup turned off in my car a mile away instead of the headphones on my person turned on for a minute straight. And then not having working audio until I disconnect and reconnect it. It's so bad Dolphin Emulator added a mode where it uses libusb to directly control the Bluetooth adapter just so it can get the OS out of the way to get all the functions of the Wiimote to properly work.

evol262
Nov 30, 2010
#!/usr/bin/perl

Varkk posted:

Windows 10 seems to be bringing them back. I feel Windows reached a peak with 7 on the desktop and has been downhill since then. For the servers the R2 releases are still good. E.g 2008r2 and 2012r2 are great.

This is always an odd argument, since Windows server and consumer Windows aren't strongly differentiated from an architectural/kernel perspective, and they haven't been since XP, give or take Windows 2000 (which not a lot of consumers used).

AFAIK, Win10 is/was (before reversioning) NT 6.4. 7 was NT 6.1. The kernel space isn't that different.

ToxicFrog posted:

You seem to be using a very narrow definition of "unreliable" that reduces to "doesn't kernel panic". And, to be fair, I haven't seen any BSODs in 10 yet, and didn't see any in years of running 7 that weren't directly attributable to hardware failures.

On the other hand, there's plenty of ways an OS can be "unreliable" that don't require a kernel panic, like not remembering settings or credentials you told it to (7), randomly resetting settings to the default (10), failing to install updates or requiring multiple days of continuous uptime to do so (7), loving with your boot configuration (both)...

"Not remembering remembering settings" is unreliable. This also happens on Linux and MacOS. As does resetting resetting to the defaults. And failing to install updates. And loving with your boot configuration. We just expect that Linux will do this sometimes when you enable some random repo and try "dnf update". We expect that OSX will eat your EFI partition on major upgrades and break your dual booting setup. There's a long standing bug with Office on OSX which results in crashes every 2-3 minutes. The only fix is "reinstall OSX".

Operating systems break. Windows used to be a steaming pile. Now it's just as bad as everything else. There are reasons to dislike it, but cherry picking problems other OSes have isn't a great one.

VikingofRock
Aug 24, 2008


Slippery Tilde

Phosphine posted:

Check if you have a "tabs" command. On my ubuntu machine i can do 'tabs -N' to set tab stops to every N characters. if it exists, man tabs was fairly readable.

Welp this worked when I typed it in manually, but when I added it to my .zshrc it didn't do anything. Some more googling turned up setterm --regtabs 4, but I got an error about xterm not recognizing regtabs. So the super janky solution I ended up with was adding TERM=linux setterm --regtabs 4 to my .zshrc. Somehow that works.

Morter
Jul 1, 2006


Time after time,
I tried to walk your way yeah


uhn, uhn uhn

And you don't break my heart!





So i'm running Xubuntu (specifically, Ubuntu 17.04, 32-bit) and I'm trying to get Dwarf Fortress running.

I've spent a good chunk of today trying to get past this error that reads:

"/df_linux/libs/libstdc++.so.6: version 'GLIBCXX_3.4.20' not found (required by /usr/lib/i386-linux-gnu/libproxy.so.1)
Failed to load module: /usr/lib/i386-linux-gnu/gio/modules/libgiolibproxy.so"

And I have not succeeded. I've downloaded the many different ppa's and libraries and whatnots that so many have suggested, updated and upgraded and dist-upgraded, all that jazz, nothing seems to work.

When I gave up a few hours ago, the last thing I did mentioned something about using export, for example here. Now, rather than following advice blindly, which I did and it brought me nothing but frustration, I'm asking because I want to know what my distro's problem is and what it needs to be fixed, since this seems more like a Linux issue than a DF issue.

Hypnobeard
Sep 15, 2004

Obey the Beard


Slippery Tilde

I'm trying to play around with Fedora 25 on a Lenovo laptop, and I'm having difficulties with the wireless adapter not being recognized.

It's a Broadcom BCM4352. I can't seem to find anything more current than Fedora 23 for getting it installed, and using the packages from rpmfusion.org hasn't resolved the issue.

Am I just SOL because lol Broadcom? Or are there steps I can look into to get in working? I'm a reasonable newbie to Linux, so be gentle.

There Will Be Penalty
May 18, 2002

Makes a great pet!

VikingofRock posted:

This is probably a stupid question, but for whatever reason I'm having trouble finding the answer: how do I change the default tab width in the terminal? I'm using zsh as my shell and konsole as my terminal emulator--I'm not sure which one controls the tab width. Googling "zsh tab width" turns up a bunch of info about zsh tab completion, and googling "konsole tab width" turns up a bunch of information about using multiple tabs in konsole.

I'm curious as to why you want to do this.

Any good pager (less, etc.) or editor (emacs, vim, etc.) will happily support any tab width you specify without you changing your terminal's tab width.

I mean, unless you want to just `cat` a file or something I guess.

apropos man
Sep 5, 2016

You get a hundred and forty one thousand years and you're out in eight!

Kassad posted:


I'm leaning towards setting the Windows partition not to mount at all (won't be anything on it I'd want to access while in Ubuntu, anyway) but it probably wouldn't hurt to add this option as well.

I'd also employ a kind of reverse situation of what you describe: use full disk encryption on your Linux system.

Then you can be sure that whatever Windows is doing can't have any effect on your Linux install. It would need the luks/dm-crypt password to physically change anything in your Linux environment.

Not that there's any viruses currently around that would try to do any cross platform tricks, or even access ext file systems, probably.

Full disk encryption with luks is easy to set up in Ubuntu and it actually uses your chosen password to unlock the real symmetric encryption/decryption key that scrambles your data. So you could be extremely lazy and set your password as a simple "x" but the underlying encryption is as strong as if you had used a complicated password. I have my dual boot set like this.

VikingofRock
Aug 24, 2008


Slippery Tilde

There Will Be Penalty posted:

I'm curious as to why you want to do this.

Any good pager (less, etc.) or editor (emacs, vim, etc.) will happily support any tab width you specify without you changing your terminal's tab width.

I mean, unless you want to just `cat` a file or something I guess.

It's that, and also I'll occasionally run command line programs which use tabs for formatting. It's not really a big deal, and I could always just pipe them into vim / less, or just not care. I was just surprised it wasn't easier to accomplish.

Kassad
Nov 11, 2005

It's about time.


apropos man posted:

I'd also employ a kind of reverse situation of what you describe: use full disk encryption on your Linux system.

Then you can be sure that whatever Windows is doing can't have any effect on your Linux install. It would need the luks/dm-crypt password to physically change anything in your Linux environment.

Not that there's any viruses currently around that would try to do any cross platform tricks, or even access ext file systems, probably.

Full disk encryption with luks is easy to set up in Ubuntu and it actually uses your chosen password to unlock the real symmetric encryption/decryption key that scrambles your data. So you could be extremely lazy and set your password as a simple "x" but the underlying encryption is as strong as if you had used a complicated password. I have my dual boot set like this.

Oh yeah, I was planning to do that as well just as a generic safety measure in case the computer gets stolen or something.

evol262
Nov 30, 2010
#!/usr/bin/perl

Morter posted:

So i'm running Xubuntu (specifically, Ubuntu 17.04, 32-bit) and I'm trying to get Dwarf Fortress running.

I've spent a good chunk of today trying to get past this error that reads:

"/df_linux/libs/libstdc++.so.6: version 'GLIBCXX_3.4.20' not found (required by /usr/lib/i386-linux-gnu/libproxy.so.1)
Failed to load module: /usr/lib/i386-linux-gnu/gio/modules/libgiolibproxy.so"

And I have not succeeded. I've downloaded the many different ppa's and libraries and whatnots that so many have suggested, updated and upgraded and dist-upgraded, all that jazz, nothing seems to work.

When I gave up a few hours ago, the last thing I did mentioned something about using export, for example here. Now, rather than following advice blindly, which I did and it brought me nothing but frustration, I'm asking because I want to know what my distro's problem is and what it needs to be fixed, since this seems more like a Linux issue than a DF issue.

This is a DF issue. Or more specifically, whomever packaged it didn't put the correct dependencies in as requirements. So maybe an issue with Ubuntu's repos or whatever PPA you used. Did they really repackage libstdc++?

How did you install Dwarf Fortress? What version of libstdc++ do you have installed?

Hypnobeard posted:

I'm trying to play around with Fedora 25 on a Lenovo laptop, and I'm having difficulties with the wireless adapter not being recognized.

It's a Broadcom BCM4352. I can't seem to find anything more current than Fedora 23 for getting it installed, and using the packages from rpmfusion.org hasn't resolved the issue.

Am I just SOL because lol Broadcom? Or are there steps I can look into to get in working? I'm a reasonable newbie to Linux, so be gentle.
Is wl loaded? Using akmod-wl or kmod-wl?

Hypnobeard
Sep 15, 2004

Obey the Beard


Slippery Tilde

evol262 posted:

Is wl loaded? Using akmod-wl or kmod-wl?

I did install kmod-wl, but I'm not sure how to check if wl is loaded. Is that modprobe wl or some other command?

evol262
Nov 30, 2010
#!/usr/bin/perl

lsmod

Hypnobeard
Sep 15, 2004

Obey the Beard


Slippery Tilde


No, does not appear in that listing.

So the next question would be how to get it installed/loaded?

Docjowles
Apr 9, 2009



modprobe <module name> and then lsmod again to see if it worked.

Hypnobeard
Sep 15, 2004

Obey the Beard


Slippery Tilde

modprobe's coming back with:

Module wl not found in directory /lib..

which I guess tells me there's something wrong with the installation.

Morter
Jul 1, 2006


Time after time,
I tried to walk your way yeah


uhn, uhn uhn

And you don't break my heart!





evol262 posted:

This is a DF issue. Or more specifically, whomever packaged it didn't put the correct dependencies in as requirements. So maybe an issue with Ubuntu's repos or whatever PPA you used. Did they really repackage libstdc++?

How did you install Dwarf Fortress? What version of libstdc++ do you have installed?

I installed df by going to this Bay12 download page and getting the 32-bit linux download of the latest version. I just extracted it, and attempted to run "./df" in the main folder. Alongside that error, I get a "Not found: data/art/curses_640x300.png" error, which is actually there. I've tried the issues for that too, but with no dice. Something about adding some sort of command to the main df file (which is supposed to launch the game) that prioritizes something. I didn't catch it.

Supposedly, I have 'libstdc++.so.6 (libc6)', which I got from the first answer here.

RFC2324
Jun 7, 2012

Http 418


Morter posted:

I installed df by going to this Bay12 download page and getting the 32-bit linux download of the latest version. I just extracted it, and attempted to run "./df" in the main folder. Alongside that error, I get a "Not found: data/art/curses_640x300.png" error, which is actually there. I've tried the issues for that too, but with no dice. Something about adding some sort of command to the main df file (which is supposed to launch the game) that prioritizes something. I didn't catch it.

Supposedly, I have 'libstdc++.so.6 (libc6)', which I got from the first answer here.

You need to install the 32-bit version of the library if you are using 32-bit DF. Most linux distros don't install 32-bit libs by default, I don't think.

A quick google says to add :i386 to the end of the packagename when you install, but I don't know if this is still true.

E: Or just use the new 64bit 43.05 of df, it runs better anyway

RFC2324 fucked around with this message at Dec 9, 2016 around 20:21

evol262
Nov 30, 2010
#!/usr/bin/perl

Morter posted:

I installed df by going to this Bay12 download page and getting the 32-bit linux download of the latest version. I just extracted it, and attempted to run "./df" in the main folder. Alongside that error, I get a "Not found: data/art/curses_640x300.png" error, which is actually there. I've tried the issues for that too, but with no dice. Something about adding some sort of command to the main df file (which is supposed to launch the game) that prioritizes something. I didn't catch it.

Supposedly, I have 'libstdc++.so.6 (libc6)', which I got from the first answer here.

DF is also packaged incredibly badly. I can't be bothered going through the tarball, but it looks like they're shipping every dependency. Try setting LD_LIBRARY_PATH to whatever's inside their tarball so you use that instead.

Wicaeed
Feb 8, 2005


Hi all,

Quick scripting question (I didn't see a Linux specific scripting thread in the COBOL forum, point me there if there is one ):

I've got a directory tree that looks somewhat like this: https://gist.github.com/Wicaeed/f64...f72e7501090f6a7

What's the best way I could parse that directory tree and get the newest %date% folder for each hostname?

Ideally I would want to end up with something like this:

code:
Dirlist:
<path>/hostname1/dumps/20161207
<path>/hostname2/dumps/20161208
<path>/hostname3/dumps/20161209

evol262
Nov 30, 2010
#!/usr/bin/perl

Quick and dirty
code:
for dir in `ls -d /path/to/hostnames`; do latest=`ls -ltr $dir | tail -n 1 | awk '{print $9}'`; ...

xzzy
Mar 5, 2009

wakey wakey to
this bowl of tasty


Yams Fan

Is this going in a script or does it need to be a shell one liner? Because a one liner is kind of messy.

If it's a one shot I'd do something like "ls -t hostname*/dumps/*" and chop the wanted lines from the top of the output.

If it's a script that's going to be around for a while I'd probably do a for loop that runs 'ls -t | head -1' on each "hostname/dumps" directory because that way it you can easily put the newest directory into a variable without a bunch of extra logic.


But the fun part is there's a billion ways to do it. If you were working in python or something you could play with os.walk() and go really bonkers.

xzzy
Mar 5, 2009

wakey wakey to
this bowl of tasty


Yams Fan

evol262 posted:

Quick and dirty
code:
for dir in `ls -d /path/to/hostnames`; do latest=`ls -ltr $dir | tail -n 1 | awk '{print $9}'`; ...

ls -t | head -1 is faster.

Which probably doesn't matter unless there's a billion directories being read.

Morter
Jul 1, 2006


Time after time,
I tried to walk your way yeah


uhn, uhn uhn

And you don't break my heart!





RFC2324 posted:

E: Or just use the new 64bit 43.05 of df, it runs better anyway

I can't because i'm using a 32 bit OS (I tried the 64-bit, and it didn't work)

evol262 posted:

DF is also packaged incredibly badly. I can't be bothered going through the tarball, but it looks like they're shipping every dependency. Try setting LD_LIBRARY_PATH to whatever's inside their tarball so you use that instead.

Explain this like i'm a big dumb babby who doesn't know more than ls

Wicaeed
Feb 8, 2005


xzzy posted:

Is this going in a script or does it need to be a shell one liner? Because a one liner is kind of messy.

If it's a one shot I'd do something like "ls -t hostname*/dumps/*" and chop the wanted lines from the top of the output.

If it's a script that's going to be around for a while I'd probably do a for loop that runs 'ls -t | head -1' on each "hostname/dumps" directory because that way it you can easily put the newest directory into a variable without a bunch of extra logic.


But the fun part is there's a billion ways to do it. If you were working in python or something you could play with os.walk() and go really bonkers.

Yeah it's going to be a script. The goal here is to be able to focus on the newest created directory for an individual hostname and upload that directory to S3.

I think I found a solution I like:

code:
for i in `find <path>/* -maxdepth 0 -type d` ; do find $i -type d | sort -r | head -1 ; done
The folder names are in a YYYYmmdd format so sorting as such works nicely

Wicaeed fucked around with this message at Dec 9, 2016 around 22:03

Powered Descent
Jul 13, 2008

We haven't had that spirit here since 1969.

Yam Slacker

evol262 posted:

Quick and dirty
code:
for dir in `ls -d /path/to/hostnames`; do latest=`ls -ltr $dir | tail -n 1 | awk '{print $9}'`; ...

Parsing the output of ls or find in a bash loop is fraught with peril. Most obviously, it won't work right if there are spaces in any of the names.

My brain tries to do everything in awk, so here's how I'd attack the problem:

code:
find path/to/hostnames -type d | awk -F '/' '{if (NF > 4) {if ($0 > highest[$5]) {highest[$5] = $0} } } END {for (name in highest){print highest[name];}}'
Note that this will only work as written if the path to the hostnames is exactly three layers deep. If it's literally /path/to/hostnames it'll work fine, but if it's /the/path/to/the/hostnames you'll need to change the numbers in awk. It's treating the slashes as the field separators and explicitly looking at number five for the hostname. (Remember that the first field is empty, since it starts with a slash. Field five of /opt/data/foobar/hostnamegoeshere/dumps/20161201 is hostnamegoeshere.) So adjust all those fours and fives to match your actual setup.

evol262
Nov 30, 2010
#!/usr/bin/perl

Morter posted:

Explain this like i'm a big dumb babby who doesn't know more than ls
When applications are built, there's no need to reinvent the wheel. So you use something you can expect to find on the system. Wanna do hostname resolution? Use libc.

Ok, so Windows shoves shared libraries in \Windows\System[32]

When you install stuff (like a game or whatever), it often repeatedly reinstalls the C++ runtime, so you have a zillion versions.

MacOS shoves everything under the package root (.app files are directories trees)

Linux assumes libraries will be under /lib[64], /usr/lib[64], /usr/local/lib[64], etc. When applications (like DF) ship their own libraries (like their version of libstdc++ or whatever), your system doesn't know to look under "/path/to/df_linux". LD_LIBRARY_PATH says "hey, look here for libraries" first. Libraries are versioned. "libstdc++.so.6" is not always the same as "libstdc++.so.6" (they could be different versions). The linker path (from ldconfig) assumes that something looking for "libstdc++.so.6" will be happy with any version that it finds. Major versions (libstdc++.so.5) are broadly incompatible, but any "so.6" should behave basically the same. That's not always true, though. Sometimes you need newer versions. I don't know what DF needs.

I don't know if DF's startup script sets this or not. It may, but they're calling a function which should be found in libgioproxy (or libproxy), which you may or may not have. Or it may not set it at all. If it's set, and they don't ship libproxy (or libgioproxy), but do ship libstdc++, then they're doing it wrong. You may need to install that library yourself. Or maybe they don't set it at all. I don't know. Can you provide "tree df_linux"?

xzzy
Mar 5, 2009

wakey wakey to
this bowl of tasty


Yams Fan

Wicaeed posted:

code:
for i in `find <path>/* -maxdepth 0 -type d` ; do find $i -mindepth 2 -type d | sort -r | head -1 ; done
The folder names are in a YYYYmmdd format so sorting as such works nicely

Don't want to get into "tell you how to do your job" territory but this is vulnerable to unexpected directories. If someone creates hostname1/butts/farts/ your command will return that instead of one of the timestamped directories.

If you can guarantee that no one will ever do that you're fine, but if I was publishing data to some cloud service I'd want sanity checks in there.

Wicaeed
Feb 8, 2005


I added a regex filter to at least ensure that the directories we are searching meet the yyyymmdd format.

This line of the script isn't actually uploading anything to S3 yet, it's just there so I can make a list o directories that don't have open file handles so that I don't upload anything when it's in the process of being created.

In the actual upload job I can add some more logic to make sure that I'm only uploading files that meet a certain naming convention.

Hypnobeard
Sep 15, 2004

Obey the Beard


Slippery Tilde

Hypnobeard posted:

modprobe's coming back with:

Module wl not found in directory /lib..

which I guess tells me there's something wrong with the installation.

Can anyone tell me what I should be installing for this? I feel like I just missed the right package, and my google-fu is weak.

RFC2324
Jun 7, 2012

Http 418


Hypnobeard posted:

Can anyone tell me what I should be installing for this? I feel like I just missed the right package, and my google-fu is weak.

Quick google turned up this:

https://bbs.archlinux.org/viewtopic.php?id=149516

Try searching 'modprobe wl' if that doesn't help.

evol262
Nov 30, 2010
#!/usr/bin/perl

Hypnobeard posted:

Can anyone tell me what I should be installing for this? I feel like I just missed the right package, and my google-fu is weak.

akmods --force? Along with akmod-wl.

You can check whether or not "kmod-wl" matches "uname -r", though. RPMFusion's kmods aren't always up to date.

Docjowles
Apr 9, 2009



xzzy posted:

ls -t | head -1 is faster.

Which probably doesn't matter unless there's a billion directories being read.

This is absolutely the kind of poo poo that gets asked in interviews, btw, so never stop

Not literally specific flags or commands, unless your interviewer is an rear end in a top hat. But "what would happen to your previous answer if I told you it was suddenly being run against a 100TB file or 5 billion tiny files on slow storage? Anything you'd want to note or change about your approach?"

Eletriarnation
Apr 6, 2005

People don't appreciate the substance of things...
objects in space.



Oven Wrangler

This feels like a super basic question but I don't have any idea how to make progress on it. I'm setting up TigerVNC on my home server running CentOS 7. By default, my .vnc/xstartup file looks like this:

code:
#!/bin/sh

unset SESSION_MANAGER
unset DBUS_SESSION_BUS_ADDRESS
exec /etc/X11/xinit/xinitrc
All of the tutorials I read act like this default xstartup should work for the default GNOME desktop, but when I connect using the TightVNC viewer I see this:


If I hit the Log Out button, I get this:


I've searched around using various terms but the vast majority of pages I see asking about this issue are using Ubuntu, and those about RHEL-likes are mostly really old. I've tried a bunch of different suggested xstartups nonetheless but can't get anywhere really - the biggest change I've seen is some icons from a normal GNOME desktop, which after a few seconds are replaced with the error screen above.

Because all of the information on this kind of thing is "try this xstartup instead" without explaining how anything works, I don't really know how to troubleshoot it constructively. I did a "tail -f" on the VNC log file but it seems pretty happy so I guess the issue is with a different layer of what's going on. Any suggestions?

e: Oh, and the desktop that I get by connecting with physical KVM works flawlessly. I have no idea how much that has to do with what this xstartup is invoking here though.

Eletriarnation fucked around with this message at Dec 10, 2016 around 04:41

xzzy
Mar 5, 2009

wakey wakey to
this bowl of tasty


Yams Fan

Try an xstartup that only runs an xterm, will tell you if it's the wm causing problems. You'll just get a command prompt and nothing else. Alternatively look for the Xorg.0.log in var on the vnc server and see if it has any hints.

Adbot
ADBOT LOVES YOU

Eletriarnation
Apr 6, 2005

People don't appreciate the substance of things...
objects in space.



Oven Wrangler

Running only an xterm works fine.

The Xorg log is super unexciting unfortunately, I stopped/started the service and dis/reconnected while this was going on:
code:
[admin@server ~]$ sudo tail -f /var/log/Xorg.0.log
[  1856.686] AUDIT: Fri Dec  9 23:45:00 2016: 1681: client 6 disconnected
[  1856.705] AUDIT: Fri Dec  9 23:45:00 2016: 1681: client 6 connected from local host ( uid=1000 gid=1000 pid=8858 )
  Auth name: MIT-MAGIC-COOKIE-1 ID: 259
[  1856.710] AUDIT: Fri Dec  9 23:45:00 2016: 1681: client 6 disconnected
[  1862.092] AUDIT: Fri Dec  9 23:45:06 2016: 1681: client 6 connected from local host ( uid=1000 gid=1000 pid=8858 )
  Auth name: MIT-MAGIC-COOKIE-1 ID: 259
[  1862.094] AUDIT: Fri Dec  9 23:45:06 2016: 1681: client 6 disconnected
[  1862.094] AUDIT: Fri Dec  9 23:45:06 2016: 1681: client 14 connected from local host ( uid=1000 gid=1000 pid=8805 )
  Auth name: MIT-MAGIC-COOKIE-1 ID: 259
[  1862.098] AUDIT: Fri Dec  9 23:45:06 2016: 1681: client 14 disconnected

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply
«733 »