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

«‹›1041 »

ShoulderDaemon: Oct 9, 2003; support goon fund; Taco Defender

ToxicFrog posted:

Set the directory +r-x - read permission will let them list the directory contents, but without execute permission they can't access anything in the directory.

Without execute permission on the directory they won't be able to create files in it, either.

It's probably simplest to leave the directory permissions alone, and set the umask in the FTP daemon such that new files are created with permissions 000 or something. That way they can create files, but they can't read them.

# ? Feb 12, 2013 20:16

Adbot: ADBOT LOVES YOU

# ? Jun 12, 2024 18:25

JHVH-1: Jun 28, 2002

fletcher posted:

With pure-ftpd, is there a way to make it so you can only upload, you can't download? I can kinda accomplish this using upload/download ratio & bandwidth throttling, but I was wondering if there is a better way to do it?

I see this in the readme:
- '-s': The "waReZ protection". Don't allow anonymous users to download
files owned by "ftp" (generally, files uploaded by other anonymous users) .
So that uploads have to be validated by a system administrator (chown to
another user) before being available for download.

Unless you aren't using anonymous then ratio or bandwidth limit maybe. I think vsftpd has a way to just disable all downloads (and you can have per user config files if you wanted to just limit a certain user).

# ? Feb 13, 2013 03:29

Lawen: Aug 7, 2000

Anyone use and know a bunch about Red Hat Satellite server? RHN Portal and KB, man pages, and Google aren't helping with the following:
1) Is there a way from the CLI to determine if a client is already registered with the Satellite? rhn_register doesn't seem to have a flag that will tell you the current registration status. I assume there's a way to query the Satellite's Oracle DB but don't know how the schema is setup and DBs aren't really my thing anyway.
2) Is there a way to unregister/deregister a client from the Satellite via the CLI? I know I can go into the Satellite webapp and remove a client there, but I'd like a way to do it from the command line so I can roll it into a script.

# ? Feb 13, 2013 21:53

Adult Sword Owner: Jun 19, 2011; u deserve diploma for sublime comedy expertise

From what I can tell, if it's registered, it will have the file /etc/sysconfig/rhn/systemid

If the file does not exist, it's not registered.

I think you HAVE to use the GUI to fully unregister a system.

Disclaimer: I've only been fooling around with it for a week.

Adult Sword Owner fucked around with this message at 22:00 on Feb 13, 2013

# ? Feb 13, 2013 21:58

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

Lawen posted:

Anyone use and know a bunch about Red Hat Satellite server? RHN Portal and KB, man pages, and Google aren't helping with the following:
1) Is there a way from the CLI to determine if a client is already registered with the Satellite? rhn_register doesn't seem to have a flag that will tell you the current registration status. I assume there's a way to query the Satellite's Oracle DB but don't know how the schema is setup and DBs aren't really my thing anyway.
2) Is there a way to unregister/deregister a client from the Satellite via the CLI? I know I can go into the Satellite webapp and remove a client there, but I'd like a way to do it from the command line so I can roll it into a script.

Tried looking at the API? What you want to know is readily available (and you can remove systems). Something like:

code:

#!/usr/bin/python

import os
import urllib
import xmlrpclib

server   = "your.satellite.server"
user     = "admin"
passwd   = "secret"

#
# setup session
#
client = xmlrpclib.Server("https://"+server+"/rpc/api", verbose=0)
session = client.auth.login(user, passwd)

#
# find latest rpms
#
systemlist = client.channel.listSystems(session)
for system in systemlist:
   for k,v in system.items():
       print "%s : %s" % (k,v)
       #Do whatever
       client.system.delete(systemid) #From the dict you got earlier
client.auth.logout(session)

You should also look at spacecmd, which works with Satellite.

# ? Feb 13, 2013 22:14

Bob Morales: Aug 18, 2006; ~~Just wear the fucking mask, Bob~~

I don't care how many people I probably infected with COVID-19 while refusing to wear a mask, my comfort is far more important than the health and safety of everyone around me!

Does anyone use Fedora?

Ask.FedoraProject.org is running some terrible Walmart imitation StackOverflow software. Ugh.

# ? Feb 13, 2013 22:21

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

Bob Morales posted:

Does anyone use Fedora?

Ask.FedoraProject.org is running some terrible Walmart imitation StackOverflow software. Ugh.

Yes.

And Askbot isn't much worse than Stack Exchange. It's not .NET, either.

# ? Feb 13, 2013 22:36

Qtotonibudinibudet: Nov 7, 2011; Omich poluyobok, skazhi ty narkoman? ya prosto tozhe gde to tam zhivu, mogli by vmeste uyobyvat' narkotiki

Bob Morales posted:

Does anyone use Fedora?

Ask.FedoraProject.org is running some terrible Walmart imitation StackOverflow software. Ugh.

How did Ubuntu get StackExchange to do their site not on the SE domain anyway.

# ? Feb 13, 2013 22:37

Lawen: Aug 7, 2000

Saint Darwin posted:

From what I can tell, if it's registered, it will have the file /etc/sysconfig/rhn/systemid

If the file does not exist, it's not registered.

I had the same thought but I'm converting a bunch of servers that used to talk to the RHN Network for updates and it looks like the systemid shows up if it's been registered with either the RHN Network or the Satellite. I could just grep out my Satellite server hostname from the /etc/sysconfig/rhn/up2date file but that doesn't really tell me if it's been registered with the Satellite or just configured to register with the Satellite.

quote:

I think you HAVE to use the GUI to fully unregister a system.

I was starting to think the same thing but...

evol262 posted:

Tried looking at the API? What you want to know is readily available (and you can remove systems). Something like:

...

You should also look at spacecmd, which works with Satellite.

Skimming through the API it looks like the listSystem and deleteSystem methods under the System handler might do the trick. I'll do some testing and see what I can figure out. I also didn't realize that RHN Satellite was just re-branded spacewalk (or maybe that spacewalk is GPL'ed RHN Satellite?), I'll read through those docs too.

# ? Feb 13, 2013 22:41

Varkk: Apr 17, 2004

Bob Morales posted:

Does anyone use Fedora?

Ask.FedoraProject.org is running some terrible Walmart imitation StackOverflow software. Ugh.

I use Fedora, but have never used Ask.FedoraProject.org, after briefly browsing I don't think that is going to change.

# ? Feb 13, 2013 22:50

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

fivre posted:

How did Ubuntu get StackExchange to do their site not on the SE domain anyway.

They didn't. It's a CNAME:

code:

[root@filer ~]# nslookup ubuntu.stackexchange.com
Server:		192.168.0.3
Address:	192.168.0.3#53

Non-authoritative answer:
Name:	ubuntu.stackexchange.com
Address: 198.252.206.16

[root@filer ~]# nslookup askubuntu.com
Server:		192.168.0.3
Address:	192.168.0.3#53

Non-authoritative answer:
Name:	askubuntu.com
Address: 198.252.206.16

Lawen posted:

Skimming through the API it looks like the listSystem and deleteSystem methods under the System handler might do the trick. I'll do some testing and see what I can figure out. I also didn't realize that RHN Satellite was just re-branded spacewalk (or maybe that spacewalk is GPL'ed RHN Satellite?), I'll read through those docs too.

listSystems and deleteSystem will definitely work, if you can handle Python, java, or anything else that talks XML-RPC.

Spacewalk is upstream, like oVirt is upstream for RHEV, Katello is upstream for Cloudforms, Gluster is upstream for Red Hat Storage Server, etc. Any given Red Hat project has upstream (generally with RH engineers working on it).

Really, look into spacecmd (which is nothing more than a pretty-ish Python wrapper around the Satellite/Spacewalk API), and you can remove 'em with bash.

# ? Feb 13, 2013 23:59

Adult Sword Owner: Jun 19, 2011; u deserve diploma for sublime comedy expertise

A user with Gnome on Red Hat 5 is getting miffed that his trash can icon (and the trash itself) never shows files he deletes. We've checked the .Trash folder and the files are going there, but he can't empty them without going CLI and deleting them by hand, which he does not want to do.

We've tried deleting the .Trash and recreating it, moving .Nautilus away so it's recreated on login, and tried to empty it in KDE. Nothing is fixing this issue.

I keep seeing references to the bug on Google but their fixes don't work or it says it's fixed in versions 7 years old.

# ? Feb 15, 2013 20:41

Suspicious Dish: Sep 24, 2011; 2020 is the year of linux on the desktop, bro; Fun Shoe

Do you see anything in ~/.xsession-errors that looks like it could relate? What about $ gvfs-ls trash://. Does that give any output?

# ? Feb 15, 2013 21:45

Adult Sword Owner: Jun 19, 2011; u deserve diploma for sublime comedy expertise

.xsession-errors is pretty clean. It doesn't look like we have anything with gvfs installed (again I'm pretty new in this position so I'm still getting around our setups)

# ? Feb 15, 2013 21:53

Suspicious Dish: Sep 24, 2011; 2020 is the year of linux on the desktop, bro; Fun Shoe

Oh, so you have an old gnome-vfs-using system, then. Try gnomevfs-ls trash:///.

# ? Feb 15, 2013 22:09

other people: Jun 27, 2004; Associate Christ

Saint Darwin posted:

A user with Gnome on Red Hat 5 is getting miffed that his trash can icon (and the trash itself) never shows files he deletes. We've checked the .Trash folder and the files are going there, but he can't empty them without going CLI and deleting them by hand, which he does not want to do.

We've tried deleting the .Trash and recreating it, moving .Nautilus away so it's recreated on login, and tried to empty it in KDE. Nothing is fixing this issue.

I keep seeing references to the bug on Google but their fixes don't work or it says it's fixed in versions 7 years old.

If you log in to that machine with a different account does the trash work?

# ? Feb 16, 2013 01:14

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

Saint Darwin posted:

A user with Gnome on Red Hat 5 is getting miffed that his trash can icon (and the trash itself) never shows files he deletes. We've checked the .Trash folder and the files are going there, but he can't empty them without going CLI and deleting them by hand, which he does not want to do.

We've tried deleting the .Trash and recreating it, moving .Nautilus away so it's recreated on login, and tried to empty it in KDE. Nothing is fixing this issue.

I keep seeing references to the bug on Google but their fixes don't work or it says it's fixed in versions 7 years old.

NFS homedir? /home a separate partition?

https://bugzilla.redhat.com/show_bug.cgi?id=905048
https://bugzilla.redhat.com/show_bug.cgi?id=306721

Tried upgrading or downgrading (depending on whether or not you're on the most recent version) gnome-vfs2? Can you reproduce the bug?

# ? Feb 18, 2013 21:25

Bob Morales: Aug 18, 2006; ~~Just wear the fucking mask, Bob~~

I don't care how many people I probably infected with COVID-19 while refusing to wear a mask, my comfort is far more important than the health and safety of everyone around me!

I think it's kind of dumb that after you install the "MATE Desktop" group in Fedora, mate-screensaver doesn't get installed automatically.

# ? Feb 19, 2013 02:20

Suspicious Dish: Sep 24, 2011; 2020 is the year of linux on the desktop, bro; Fun Shoe

Report a bug. It's probably a packaging error.

# ? Feb 19, 2013 02:29

Not Wolverine: Jul 1, 2007

Would it be realisitc to install Chromium OS on my Eee PC? I think 1.6GHz Atom is about equal to a 1.5GHz Celeron, and I have 2GB DDR, so I think it's specs are kinda similar to a chrome book, although their screens are higher res. My biggest concern is Chromium just doesn't look user friendly, I mean the quick start guide talks about setting up and syncing a repository on your Ubuntu PC before installing, so I am pretty sure this is more a developer only OS. And apparently Chromium does not come with flash, which I think would be kinda necessary?

I want to play more with Chrome OS, but I am not quite ready to shell out $200 for a chrome book.

# ? Feb 19, 2013 03:10

Bob Morales: Aug 18, 2006; ~~Just wear the fucking mask, Bob~~

I don't care how many people I probably infected with COVID-19 while refusing to wear a mask, my comfort is far more important than the health and safety of everyone around me!

Colonel Sanders posted:

I think 1.6GHz Atom is about equal to a 1.5GHz Celeron

Depends on what Celeron you're talking about, but any recent Celeron would dust an Atom. Remember an Atom is roughly a Pentium III.

The current Chromebooks have a Celeron that will be pretty close to the E5300 at the top of that chart. The integrated graphics are also probably a hundred times faster than what is found in the Atom, as well.

# ? Feb 19, 2013 03:25

JHVH-1: Jun 28, 2002

Bob Morales posted:

Depends on what Celeron you're talking about, but any recent Celeron would dust an Atom. Remember an Atom is roughly a Pentium III.

The current Chromebooks have a Celeron that will be pretty close to the E5300 at the top of that chart. The integrated graphics are also probably a hundred times faster than what is found in the Atom, as well.

That is kind of apples and oranges though. That is like back when Apple was on PPC and people were comparing it to intel. Its a completely different architecture and better at different things. Atoms are small,low power/heat, and relatively cheap so you get a lot of bang for your buck.

EDIT: Wait, I think I just confused Atom and Arm? 12 hour work shift is getting to my brain.

# ? Feb 19, 2013 04:06

Qtotonibudinibudet: Nov 7, 2011; Omich poluyobok, skazhi ty narkoman? ya prosto tozhe gde to tam zhivu, mogli by vmeste uyobyvat' narkotiki

Is there any way to get the Google Translate IME for Linux? I /should/ learn proper Russian keyboard stuff, but auto-translit is just so much easier.

Plus my laptop keys are laid out in such a way that it's difficult to add Cyrillic stickers anyway.

# ? Feb 19, 2013 05:46

Adult Sword Owner: Jun 19, 2011; u deserve diploma for sublime comedy expertise

evol262 posted:

NFS homedir? /home a separate partition?

https://bugzilla.redhat.com/show_bug.cgi?id=905048
https://bugzilla.redhat.com/show_bug.cgi?id=306721

Tried upgrading or downgrading (depending on whether or not you're on the most recent version) gnome-vfs2? Can you reproduce the bug?

I'll try to reproduce but as far as I know, this is the only user having this problem. Everyone uses the UI heavily so it seems like if it was that bug, it would have come up before.

When I get in I'll check the version numbers.

# ? Feb 19, 2013 14:19

peepsalot: Apr 24, 2007; ��PEEP THIS...
��BITCH!

Can anyone recommend a nice gui graphing calculator that can graph using log scale on an axis?

# ? Feb 20, 2013 07:14

Varkk: Apr 17, 2004

I haven't really used it but extcalc may be able to do what you want
http://extcalc-linux.sourceforge.net/

# ? Feb 20, 2013 21:40

fletcher: Jun 27, 2003; ken park is my favorite movie; Cybernetic Crumb

I was having trouble compiling something after moving from gcc 4.4 to 4.7 and finally found the fix, apparently I just needed to move around the order of the arguments in the makefile. I am curious though, like the other commenter in that stackoverflow thread, what the reasoning behind the change like this was, and if it's mentioned somewhere in a changelog or something. Also, how do I write the makefile so it will work for both gcc 4.4 and 4.7?

# ? Feb 20, 2013 23:23

hackedaccount: Sep 28, 2009

fletcher posted:

I was having trouble compiling something after moving from gcc 4.4 to 4.7 and finally found the fix, apparently I just needed to move around the order of the arguments in the makefile. I am curious though, like the other commenter in that stackoverflow thread, what the reasoning behind the change like this was, and if it's mentioned somewhere in a changelog or something. Also, how do I write the makefile so it will work for both gcc 4.4 and 4.7?

I don't know much about GCC, but maybe put both lines into your makefile and keep one commented out. For example, if you're compiling with 4.4 uncomment the 4.4 compatible line and comment out the 4.7 compatible line, and vice-versa for 4.7.

# ? Feb 20, 2013 23:35

ExcessBLarg!: Sep 1, 2001

fletcher posted:

I was having trouble compiling something after moving from gcc 4.4 to 4.7 and finally found the fix, apparently I just needed to move around the order of the arguments in the makefile.

C compilers (really, linkers) have historically been weird (annoying) about link order. The issue, particularly for static libraries is that the linker only does a single pass over object files specified on the command line, so source files and libraries have to be specified in the order of dependency.

Since your source code (foo.c) makes library calls, it depends on symbols defined in some library (say, libbar.a). So, you have to compile/link with "cc -o foo foo.c -lbar" so that your source code is read first, the undefined symbols found, and those symbols satisfied by libbar.a. If you flip things around ("cc -o foo -lbar foo.c") it won't work, as there's no symbol dependencies in libbar.a (no main function), and by the time the library calls in your source is reached, the linker has finished analyzing the library.

That said, gcc historically hasn't cared about link order for dynamic libraries. So if you have a libbar.so, "gcc-4.4 -o foo -lbar foo.c" works fine. However at some point since GCC 4.4, it added the single-pass ordering constraint for dynamic libraries as well, hence the need to change the order. Honestly I have no idea why, this is the first I've heard of it.

Either way, the "cc -o foo foo.c -lbar" order is pretty traditional. Does it not also work with GCC 4.4?

# ? Feb 21, 2013 01:14

fletcher: Jun 27, 2003; ken park is my favorite movie; Cybernetic Crumb

ExcessBLarg! posted:

Either way, the "cc -o foo foo.c -lbar" order is pretty traditional. Does it not also work with GCC 4.4?

Well what do you know, it does work when written that way in GCC 4.4! I don't know why I didn't try before opening my big mouth. Thanks for the info, still learned something new!

# ? Feb 21, 2013 01:59

Ninja Rope: Oct 22, 2005; Wee.

Ha, that explains a lot. I was always calling it gcc -hyphenargs filenames to group the hyphenated arguments at the beginning like the top of the man page shows (until I started running into this problem too). Thanks for the clarification.

# ? Feb 21, 2013 03:07

nonathlon: Jul 9, 2004; And yet, somehow, now it's my fault ...

Not exactly a Linux problem - but if I can't solve the Linux end, it's a no go. So I thought people here might be able to point me in the right direction.

The aim: I want all my project data in front of me on every computer. I'm a computational biologist, so I'm continually writing scripts, parsing datasets, editing manuscripts. And I'm plagued constantly by wanting _that_ dataset that's on that computer over there, or being unable to tell if I've got the latest version of something in front of me. So I figure I'm looking for a file syncing solution.

The environment: heterogenous. I prefer to do most work under linux or unix-like environments, there's the home Mac, work insists that we use Windows (no, I can't just quietly install Fedora, it's a disciplinable "security risk"), I have an Android tablet.

The parameters:

* Reliability: it's no good if 99% of files sync 99% of the time
* Cost: of course
* Security: I've got no personal information on there, but no need to be cavalier
* Space: probably a low amount of storage at first (a few gig) but certain to grow
* All files and info: solutions that exclude files or transform them are right out

People have suggested SparkleShare, SpiderOak and OwnCloud to me. I've installed SpiderOak and it seems to work fine, but opinions on the others would be welcome.

I use Google for a lot of stuff (email etc.) so it seems natural to look at Google Drive. The extra storage is cheap enough, but I've found it incredibly unreliable. (Some files just refuse to sync or take ages to appear. The sync seems to be a bit stupid.) On Ubuntu there's no native client, but Insync uses your Google space and seems to work well - although there are a few reports of it not syncing or excluding files.

I've used DropBox for this for a few projects and it's worked well. The Linux client is good. Dropbox appear to sanely sync. But extra space isn't cheap and their patchy record on security gives me pause.

Recommendations?

# ? Feb 21, 2013 13:08

Bob Morales: Aug 18, 2006; ~~Just wear the fucking mask, Bob~~

I don't care how many people I probably infected with COVID-19 while refusing to wear a mask, my comfort is far more important than the health and safety of everyone around me!

Get a bunch of friends to sign up for DropBox and get the referral storage?

# ? Feb 21, 2013 14:35

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

Bob Morales posted:

Get a bunch of friends to sign up for DropBox and get the referral storage?

SpiderOak also does this, and is (in many ways) an all-around better solution. Fewer security problems, mostly open-source libraries, etc.

# ? Feb 21, 2013 15:34

Lawen: Aug 7, 2000

Box.net gives away 5GB with their free account, no official Linux client but if you have root and can install WebDAV (davfs2) you should be able to make it work. If Dropbox has been working well for you though, either referral chain and do their promo stuff to get some more free space or cough up the $10/month for more space.

The other option, if your home Mac is always on and your Internet connection isn't terrible, keep everything in a git (mercurial/svn/cvs/whatever) repo on your home Mac, setup dynDNS and port forwarding, and pull your files down from there. It won't auto-sync but it would be secure, free, and you'd be assured of always working with the current version of the file.

# ? Feb 21, 2013 15:39

ToxicFrog: Apr 26, 2008

outlier posted:

Recommendations?

I've been wrestling with this for a while (although I've been looking specifically for self-hosted stuff, so no spideroak). Thoughts on what I've tried:

- SparkleShare: great for small stuff. It's git-based, though, and size will increase massively (and performance will tank) if you're slinging around a lot of large binaries. There are some plans to fix this in 2.x (don't store history on the client, auto-prune old history on the server, use git-bin for large files) but no timeline for when 2.x will be released.
- OwnCloud: needs a full LAMP stack on the server; resource hog on both client and server; client runs like poo poo even on fast machines and reacts to unreliable network connections by crashing. I do not recommend this. (This was 4.0; I don't know how much 4.5 fixes.)
- SyncAny: exactly what I want but hasn't been released yet and probably never will be.

For the moment I've settled for just manually running Unison to sync things. I'm toying with the idea of writing a daemon around it that auto-syncs on filesystem changes.

# ? Feb 21, 2013 16:08

nonathlon: Jul 9, 2004; And yet, somehow, now it's my fault ...

Yeah, I got a few big recommendations for SparkleShare but the git thing gives me pause. Problems with big binaries and all of that git history. DropBox is looking better kid only for the fact that it "just works".

# ? Feb 21, 2013 17:20

Bob Morales: Aug 18, 2006; ~~Just wear the fucking mask, Bob~~

I don't care how many people I probably infected with COVID-19 while refusing to wear a mask, my comfort is far more important than the health and safety of everyone around me!

evol262 posted:

SpiderOak also does this, and is (in many ways) an all-around better solution. Fewer security problems, mostly open-source libraries, etc.

Don't care about any of that stuff. If I want it secure I'll encrypt it before I put it anywhere.

The thing DropBox has going for it is that it's dead simple.

# ? Feb 21, 2013 17:51

spankmeister: Jun 15, 2008

Lawen posted:

Box.net gives away 5GB with their free account, no official Linux client but if you have root and can install WebDAV (davfs2) you should be able to make it work. If Dropbox has been working well for you though, either referral chain and do their promo stuff to get some more free space or cough up the $10/month for more space.

The other option, if your home Mac is always on and your Internet connection isn't terrible, keep everything in a git (mercurial/svn/cvs/whatever) repo on your home Mac, setup dynDNS and port forwarding, and pull your files down from there. It won't auto-sync but it would be secure, free, and you'd be assured of always working with the current version of the file.

Git isn't really suited for binary files, might as well just use rsync then.

# ? Feb 21, 2013 18:16

Adbot: ADBOT LOVES YOU

# ? Jun 12, 2024 18:25

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

Bob Morales posted:

The thing DropBox has going for it is that it's dead simple.

Exactly what about Dropbox do you think is simpler than Spideroak?

# ? Feb 21, 2013 18:18

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

«‹›1041 »