The Linux Questions Thread: a bunch of pitfalls, but technically it's possible

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > The Linux Questions Thread: a bunch of pitfalls, but technically it's possible

«‹›30 »

RFC2324: Jun 7, 2012; http 418

evol262 posted:

Transparent proxying squid is exactly the right answer. But it's going to take significant setup. Alternatively, you could harass someone in another thread about writing a browser extension which forwards through a local proxy that goes through the VPN if and only if you get a 302. But because the problem presented is somewhat difficult and there's no tool to install that handles it all for you, I'd recommend just doing everything across the VPN.

Just set the metrics on the adapter properly. It will use the primary connection unless it can't route to the address, them it will try to route through the VPN adapter.

# ¿ Nov 5, 2013 06:50

Adbot: ADBOT LOVES YOU

# ¿ May 14, 2024 16:55

RFC2324: Jun 7, 2012; http 418

my stepdads beer posted:

Is it possible to run a Windows VM in the background (or remotely?) and have windows applications open in their own windows on my Gnome 3 desktop?

Empathy + SIPE just doesn't work well enough to replace Lync yet

Virtualbox has a seamless mode that I used to do a thing like this with. You still need the windows taskbar to show, tho.

# ¿ Jan 5, 2014 21:47

RFC2324: Jun 7, 2012; http 418

hackedaccount posted:

sudo's kinda out of the picture too because there would be a huge shitstorm about giving non-admins root and blah blah blah.

You can configure sudo to only allow certain commands, like, say, the one you want to have run.

# ¿ Jan 13, 2014 22:45

RFC2324: Jun 7, 2012; http 418

So, I am working on a script, and having some issues with an eval function. This is in a solaris 8/9/10 environment, so things are a little different than I am used to seeing.

code:

for i in $*; do
	if [[ "ssh -t \"$i\" find /opt/bmc/scripts/ -name check-var-crash-files.sh | wc -l" = "0" ]]
		then 
			ssh -t $i sudo scp x0210454@utility:/opt/bmc/scripts/check-var-crash-files.sh /opt/bmc/scripts
			echo "--------------- Missing script on $i placed. ----------------"
			exit
	fi
done

This is appearing to fail every time, and I cannot figure out why. Anyone have any suggestions?

# ¿ Jan 28, 2014 22:11

RFC2324: Jun 7, 2012; http 418

All arguments being passed to the script. The idea is for it to take a list of servers, and correct all of the issues that are causing that particular alarm to go off(this section checks for the script that triggers the alarm, and places it if it is missing).

Everything else in the script works, just not this one segment, and it looks like the eval is failing.

# ¿ Jan 28, 2014 22:35

RFC2324: Jun 7, 2012; http 418

telcoM posted:

I see no eval there, just a string comparision that is obviously always going to be false.

You'll want something like this instead:
code:
        if [[ $(ssh -t "$i" find /opt/bmc/scripts/ -name check-var-crash-files.sh | wc -l) = "0" ]]
Edit: Or if the Solaris shell does not accept the "$( ... )" syntax, then the backticks:
code:
        if [[ `ssh -t "$i" find /opt/bmc/scripts/ -name check-var-crash-files.sh | wc -l` = "0" ]]
The "if [[ ... ]]" construct is essentially similar to "if [ ... ]", but it uses shell's internal conditional expression evaluator instead of the POSIX standard /usr/bin/[, also known as /usr/bin/test. There may be some differences between the POSIX standard conditional expression evaluator and the shell's built-in one, but the built-in one is definitely not going to evaluate a double-quoted string as a command just like that.

Thanks. I wasn't able to get it working with the wc -l bit, but by changing it to

code:

if [[ $(ssh -q -t "$i" find /opt/bmc/scripts/ -name check-var-crash-files.sh) = "" ]]

I was able to make it work with the correct container for the function.

# ¿ Jan 29, 2014 15:07

RFC2324: Jun 7, 2012; http 418

If you are in a gui session, I'm very fond of terminator for split screen behaviour.

# ¿ Feb 2, 2014 17:12

RFC2324: Jun 7, 2012; http 418

Xik posted:

Yeah, already tried that. sftp doesn't appear to interpret those escape strings. The frustrating thing is that in a local and ssh shell I can actually enter in the character directly, but not in sftp.

A further complication, and one I just noticed, is that the target server (which I'm not an admin of) has messed up locale settings. So even though ssh would normally be able to handle the character, it can't when connected to this server. In the end I just settled on juggling around the files into folders with names I could pull down in sftp.

The problem will probably pop up in the future and I would really like to know the "correct" way to handle it. How can I enter unicode characters into sftp assuming both servers in question have UTF8 locale settings? I tested between two machines that have correct settings, ssh can accept the characters but sftp can't. I feel like I'm missing something super obvious, how do admins that work primarily in languages other than English deal with this problem?

Why not use something other than sftp, like scp or rsync then?

# ¿ Feb 4, 2014 01:24

RFC2324: Jun 7, 2012; http 418

Ninja Rope posted:

When kickstarting, there's an option (sshpw setting) that lets users ssh in to the currently installing host. However, the actually install process can only be seen by VNCing in because the install runs in GUI mode. If I disable GUI mode install (text setting), can I watch the install when I ssh in? Or will I be dropped to a prompt as usual?

I'd like to be able to ssh in and debug an install without having to connect via VNC.

Pipe all the textmode output to a log file, tail -f the log file.

# ¿ Feb 6, 2014 03:50

RFC2324: Jun 7, 2012; http 418

evol262 posted:

I'll never understand this Stockholm Syndrome.

"I just installed this distro and it's immediately so badly configured that networking doesn't work at all. Slick. I'm lovin' it."

:archlinux:

At least with Gentoo its not a default config issue, its a 'You are dumb and compiled your system like a retard. Try again.'

# ¿ Feb 20, 2014 19:13

RFC2324: Jun 7, 2012; http 418

spankmeister posted:

With Gentoo it's "You are dumb and installed the wrong distro like a retard. Try again."

That was actually my reaction. I got halfway through and said 'gently caress this, too much effort'.

Now I use Kubuntu like a retard :downs:

# ¿ Feb 20, 2014 19:27

RFC2324: Jun 7, 2012; http 418

Lysidas posted:

High five, Kubuntu buddy

KDE 4.11 (Kubuntu 13.10) and 4.12 (14.04, still in development) are really good.

I still ruin my pants slightly using the desktop cube with a touch screen every time. I'm just glad this isn't my work setup.

# ¿ Feb 21, 2014 08:02

RFC2324: Jun 7, 2012; http 418

Xik posted:

They both look black to me, does that help?

same here.

# ¿ Mar 1, 2014 07:24

RFC2324: Jun 7, 2012; http 418

Combat Pretzel posted:

So yeah, I figured, I set up a Linux VM to run The Thing System for all my fancy gadgets, and for ease downloaded Ubuntu. I guess that was a big mistake, because there doesn't seem an easy way to start a shell. None of the default shortcuts offers a shell, nor does the desktop context menu, nor that universal search thing. Really?!

do a search in the launcher menu for 'terminal'

# ¿ Mar 15, 2014 22:17

RFC2324: Jun 7, 2012; http 418

the posted:

Um... 12.04? I think? And it's a Lenovo Z400 touch

It's not that I'm having sensitivity issues, there's just literally no response anywhere. The touchscreen is totally not active.

If you do a lspci -v, or a lsusb -v, do you see the touchscreen listed? I am using a Lenovo Flex15 wiht Kubuntu 13.04, iirc, and my touchscreen was detected, and mostly worked, right out of the box(worked like a left click only mouse, no nifty touchscreen features like scrolling or multitouch).

I believe you should see an eGalax device listed, and if you do a -vv it will give you a features list.

# ¿ Apr 3, 2014 08:15

RFC2324: Jun 7, 2012; http 418

Suspicious Dish posted:

There's a few more terminal emulators based on VTE out there that retained transparency. I believe people like Terminator now?

I really liked Terminator myself, but lately I have found just basic Konsole is really doing well. I like the way it handles both tabs and splits screens(tabs are replicated on each split so you can switch each portion of the screen).

Terminator is really nice, but it has a habit of crashing when I am moving subwindows inside it that annoys me when I am working.

# ¿ Apr 5, 2014 03:29

RFC2324: Jun 7, 2012; http 418

What is the problem with ext4?

# ¿ Apr 5, 2014 10:25

RFC2324: Jun 7, 2012; http 418

Ashex posted:

So I wrote this cool sed expression:

sed -e 's/$^.*tracker_$$.*$$:.*$$:.*$/\2/'

For this string:

tracker_ip-10-85-57-55.ec2.internal:localhost/127.0.0.1:49191

To get this:

ip-10-85-57-55.ec2.internal

But it doesn't work for this:

tracker_ip-10-21-199-91.ec2.internal:45454

I'm assuming it's because I have it looking for two occurrences of : as if I remove the second one then it comes out right. In my head it's supposed to give me everything between tracker_ and the first occurrence of : but my head is bad at regex.

Would this work?

code:

awk -F'[_:]' '{ print $2 }'

# ¿ Apr 14, 2014 16:23

RFC2324: Jun 7, 2012; http 418

Time to distro hop again. Went to Gentoo for about a week(Gentoo never seems to last more than a week before I get annoyed because something basic just won't work) and decided to try the base Ubuntu. I have decided I hate Ubuntu, so switched to Fedora with KDE. Sadly, Fedora seems to be annoying the hell out of me fighting the various license issues just to get flash working, and the pipelight repo doesn't seem to want to work right(getting 404 errors), tho that might be the firewall here at work.

Any suggestions for making Fedora more painless in dealing with basic things that happen to lack a non-free license?

# ¿ May 25, 2014 07:35

RFC2324: Jun 7, 2012; http 418

Suspicious Dish posted:

Install rpmfusion.

I saw that mentioned before, but didn't follow up. Obviously I should have, thanks

# ¿ May 25, 2014 11:31

RFC2324: Jun 7, 2012; http 418

Suspicious Dish posted:

Really, a lot of us on the desktop team want Fedora you to be able to install MP3 codecs and stuff like that so your brand new OS everybody has been talking about is actually useful out of the box, but of course the Fedora Engineering Steering Committee wants to "keep your OS free". I still have no idea what that means.

Oh, I can see the idealism behind it, and why it might be stuck with, but it does make it annoying. At least in Gentoo you can add a single like to /etc/portage/make.conf and tell it to autoaccept licenses if you don't want to deal with them(it is just a restricted until you do that).

And the fact that I haven't touched anything Red Hat since they split Fedora out from RHEL doesn't help, all I have thats applicable is CentOS experience. With no GUI.

# ¿ May 25, 2014 11:57

RFC2324: Jun 7, 2012; http 418

Dilbert As gently caress posted:

Can I ask a solaris 10 question here or am I better looking else where, I am more rusty than I like to admit on my solaris.

What do you need to know?

# ¿ Jun 8, 2014 02:47

RFC2324: Jun 7, 2012; http 418

Suspicious Dish posted:

What do you think booting is?

Totally organized and well designed? :ohdear:

# ¿ Jun 13, 2014 14:15

RFC2324: Jun 7, 2012; http 418

looks like a broken tooltip, I see those in windows occasionally too.

Usually need to just restart the session.

# ¿ Jun 17, 2014 16:54

RFC2324: Jun 7, 2012; http 418

I have gotten the 'operation not permitted' coming from VMs as well, particularly from VirtualBox.

Its listed as a known issue that VBox guests do not send pings properly.

# ¿ Jun 19, 2014 06:16

RFC2324: Jun 7, 2012; http 418

Don't forget the multifactor authentication!

# ¿ Jun 28, 2014 12:48

RFC2324: Jun 7, 2012; http 418

I am preparing to take the RHCSA test, would just installing CentOS work, or is an eval of RHEL going to be required? Haven't picked up the books yet, but I was going to set up a VM to play with.

# ¿ Aug 15, 2014 02:48

RFC2324: Jun 7, 2012; http 418

My US based company is converting from Solaris to SUSE. I have no idea why, tbh, I would have expected them to go RHEL, since we already have some RHEL from prior to the decision to move to a Linux platform.

# ¿ Aug 30, 2014 15:47

RFC2324: Jun 7, 2012; http 418

spankmeister posted:

You could defend choosing between SEL or RHEL easily, but defending supporting two different linuces? Ehhhh...

I actually officially support Solaris 5, 6, 7, 8, 9, and 10, SEL, RHEL 4, 5 and 6, and I think we have a couple HPUX boxes still active.

My company does not update anything if its not forced on us, but we do install new stuff when we do new builds. Some of those solaris 6 boxes have uptimes of over a thousand days, and upper management will not let us even reboot them for fear they will not come back(its happened once with a production critical server since I have been here)

# ¿ Aug 30, 2014 16:03

RFC2324: Jun 7, 2012; http 418

evol262 posted:

I know this is super pedantic, but there never was a Solaris 5 or 6. There was Solaris 2.5 and 2.6

and even 10 is actually 5.10. Their versioning doesn't match up to what anyone actually calls it.

# ¿ Aug 30, 2014 17:16

RFC2324: Jun 7, 2012; http 418

evol262 posted:

Sure, in some cases. Nobody calls 7,8,9,10,11 5.anything, but I've never heard anyone talk about Solaris 5/6, either. Maybe it's because the last time I saw 2.6 was in 2003, but there was a very clear distinction between "Solaris 2.x" and "Solaris X" at the time.

Also, 5.10 isn't... It's SunOS 5.X, not Solaris 5.X, if we want to keep being pedantic.

Holy crap, even the guy I work with who spent 25 years at Sun doesn't call it SunOS.

# ¿ Aug 30, 2014 19:09

RFC2324: Jun 7, 2012; http 418

evol262 posted:

Because SunOS effectively died with Solaris 2.x, which effectively died with Solaris 7, but "uname" -> 5.11 is SunOS 5.11, not Solaris 5.11. It's Solaris 11. But 5.6 is SunOS 5.6, Solaris 2.6, not Solaris 6.

SunOS was a very different thing, but they kept continuity in naming (plus retroactively naming SunOS 4.x Solaris 1.x) to not scare people, despite SunOS and Solaris having different lineages and being basically incompatible.

Naming for marketing is hard.

SunOS until Solaris
Solaris -> SunOS 5
Solaris 2.6 until 2.7
Solaris 2.7 -> Solaris 7
Solaris 11 -> SunOS 5.11

Doesn't matter anyway. Just saying that I've never seen anyone talk about Solaris 5 or 6, and Sun didn't either. They differentiated minor -> major at 7, though there was some talk of making Sol11 Solaris 3.x/SunOS 6.x

Weird. The handful we still have on 6(which all report SunOS 5.6) all all referred to as Solaris. I don't remember what the one or two that still ran 5 were referred to as other that 'that piece of poo poo again?'

You would think a fortune 500 company would want to spend a little on keeping a consistent environment, but noooooo...

# ¿ Aug 31, 2014 00:37

RFC2324: Jun 7, 2012; http 418

babies havin rabies posted:

I've been playing around with Linux directory solutions the past week or so in my spare time. All I'm really interested in producing is central user account authentication with a secure file-system. I'm purely curious because this is an area of Linux I'm not versed in at all. Here's the things I've tried:

OpenLDAP with Debian server/clients. Worked fine, set up some users and then NFS. Put NFS share in client fstab and the users basically get roaming profiles when you map a home directory. Client setup was a breeze. Then I discover NFS basically has zero security, you can root a machine and change your UID and get access to anything. No go.

FreeIPA with Fedora server/clients. Works fine for authentication, easiest setup on client/server. Could not get NFS to run kerberized following any guide whatsoever. I'd get as far as everything working and wouldn't be able to remove auth=sys from the NFS export, so it wasn't any better than OpenLDAP. Disappointed that there is no ipa-client for Debian distros either.

Samba, Debian server/Deb+Win7 clients. Successfully added a Win 7 machine to the domain, user home directories map to H:. Working on getting a Debian workstation on the domain, but I'm finding it's actually more complex to do a client than setting up the goddamn server was.

Anyway, am I missing anything? Am I in over my head or is every directory solution on Linux really this obtuse, complex, and decentralized? The sanest Linux solution for this seems to be just mounting poo poo with sshfs or smb piecemeal. The whole experiment has really made me appreciate that I administer Active Directory for my living. Anything else out there that I should try?

Not sure how its implemented, but I know in my work environment root does not actually have access to the contents of an NFS share, and I am unable to remount certain NFS filesystems as root, and this is using an LDAP implementation, so I think you may be missing something there.

# ¿ Sep 1, 2014 06:12

RFC2324: Jun 7, 2012; http 418

spankmeister posted:

root_squash means your client's root will map to nobody on the nfs server but you can change your uid and access other people's files, yes.

This really works in an LDAP environment? I have never tried, outside of su(which I assume is the main flaw in all N*X security, become root and you are EVERYONE) but that seems awful easy, even without gaining root.

# ¿ Sep 1, 2014 08:09

RFC2324: Jun 7, 2012; http 418

If you want the RHCE, currently, you have to already have the RHCSA as well.

I'm currently working on both, if I can ever stop drinking long enough to study.

# ¿ Sep 27, 2014 03:13

RFC2324: Jun 7, 2012; http 418

eXXon posted:

... and I guess I just now realized that F20 doesn't install or enable sshd by default, so I know what I'm doing tomorrow.

Can't speak for anything else, but since Fedora is a desktop distro, it would not make sense for sshd to be there by default, since having it open is a security risk that most users don't need to have available. If you want a server, Fedora isn't the best route anyway.

# ¿ Sep 27, 2014 04:24

RFC2324: Jun 7, 2012; http 418

caberham posted:

Anyways, I should stop asking questions and hogging the thread, sorry.

Nah, some of the rest of us are taking notes too

# ¿ Sep 28, 2014 04:16

RFC2324: Jun 7, 2012; http 418

In the process of installing CentOS 7 on a new(old) server and I discovered that RHEL no longer packages the forcedeth driver(for nVidia network cards) in the default kernel, and therefore neither does CentOS.

Apparently not the e1000 driver either, and a few more older, but very common, net drivers

What the gently caress? Every single server I work on uses one of these 2 drivers.

# ¿ Oct 3, 2014 07:09

RFC2324: Jun 7, 2012; http 418

spankmeister posted:

that sucks but your hardware is really old dude.

anyway elrepo has kmod-forcedeth and ~~kmod-e1000~~ for EL7

e: e1000 not for el7

Yeah, I found it, but even my gigabit network card on a 3 year old motherboard uses forcedeth, so its not like its only on decade old hardware(I found out setting up a refurb sun x4140 I picked up real cheap as a quick and dirty home lab virtualization host).

I was just shocked to see that it wasn't even included, I would assume the module would be there but not loaded in any way. Its only 47k.

# ¿ Oct 3, 2014 09:51

Adbot: ADBOT LOVES YOU

# ¿ May 14, 2024 16:55

RFC2324: Jun 7, 2012; http 418

RFC2324 posted:

(I found out setting up a refurb sun x4140 I picked up real cheap as a quick and dirty home lab virtualization host).

On this note, is there a port of the Solaris diagnostic tools for linux? specifically prtdiag and cestat/cediag. Those are ever so much more useful for keeping tabs on sun hardware than anything I have ever found for linux.

# ¿ Oct 3, 2014 10:21

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > The Linux Questions Thread: a bunch of pitfalls, but technically it's possible

«‹›30 »