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

«‹›3 »

Alowishus: Jan 8, 2002; My name is Mud

Five months to fill a thread this time... come on folks, surely something about Linux is vexing you! Or has it just gotten that easy to use??

Previous Threads:
Archived (1 2 3 4) 5

General Help:

Hardware Help:

Distribution Specific Help:

Ubuntu

Debian

About Debian

Gentoo

Fedora

Fedora FAQ

# ¿ Mar 21, 2007 19:15

Adbot: ADBOT LOVES YOU

# ¿ Apr 25, 2024 22:43

Alowishus: Jan 8, 2002; My name is Mud

Postal posted:

Anyone know any good GUI frontends for Snort?

BASE? Or do you mean fat native desktop GUI?

# ¿ Mar 30, 2007 02:43

Alowishus: Jan 8, 2002; My name is Mud

Grabulon posted:

Let's say I have a directory /home/grabulon/photos with several subdirectories.
I simply want the users to start there and only be allowed to read from "photos" and its subdirs.

What's the best way to do this?

Set grabulon's home directory to /home/grabulon/photos and then enable chroot_local_user in vsftpd.

There are fancier ways as well, look at the chroot_list_enable (and matching chroot_list_file) settings, as well as passwd_chroot_enable for more pathing options.

# ¿ Mar 31, 2007 17:50

Alowishus: Jan 8, 2002; My name is Mud

Grabulon posted:

But doesn't this mean that they have to login as "grabulon"? I want to create new users with access to my home directory...

Oh well you could create more users who also have /home/grabulon/photos as their home directories. You'd need to make sure they had rights to the files in there, probably by making a group.

Give us some more specifics about what you're trying to do and hopefully we can give you better guidance.

# ¿ Apr 1, 2007 03:47

Alowishus: Jan 8, 2002; My name is Mud

fletcher posted:

Is there a way to see live traffic statistics for apache?

You might like mod_status that's built into Apache (but generally disabled by default). It gives you a magic web page that upon load shows RIGHT NOW what each Apache process is doing.

# ¿ Apr 30, 2007 23:36

Alowishus: Jan 8, 2002; My name is Mud

Stage1 installs aren't even supported by the Gentoo team any more. Go with Stage3, that should work fine.

# ¿ May 30, 2007 01:01

Alowishus: Jan 8, 2002; My name is Mud

coconono posted:

I need a linux distro that can host MediaWiki(and the stuff it depends on to run). The catch being is that the distro needs to be as lightweight as possible as I'll be running it in VMWare along with several other Virtual Machines(I'm thinking about ditching X server and all the toys it brings just to cut down on the headroom). Also it has to fit on one CD.

Debian. One CD to install (any further packages you need come over the network), very basic default install, and has the LAMP stack you need to run MediaWiki. And it's supported by VMware.

# ¿ Jun 4, 2007 19:32

Alowishus: Jan 8, 2002; My name is Mud

Washuu-Chan posted:

Is there a way to create a pair of virtual network interfaces (of ethernet type) that act as a "pipe"? I want to have ethernet packets sent on one of them to appear on another and vice versa.

Well what you want is a bridge... perhaps there's just a configuration problem with what you were doing? Your kernel does have to have bridge support compiled in or available as a module, and that module will need to be loaded. You were using 'brctl' to do the configuration, right?

# ¿ Jun 12, 2007 07:08

Alowishus: Jan 8, 2002; My name is Mud

cdrecord will do DVDs too. Use the -scanbus option to find your burner's device number, and then something like:

code:

cdrecord -v dev=6,0,0 driveropts=burnfree file.iso

# ¿ Jun 16, 2007 21:17

Alowishus: Jan 8, 2002; My name is Mud

Sister Miyagi posted:

I'm trying to compile ndiswrapper under Fedora7. I downloaded the kernel sources as described here, and they seem to have put themselves under /usr/src/redhat/BUILD/kernel-2.6.21/linux-2.6.21.

I think that doc is just wrong... or at least it's making things harder than they should be. This can get a little confusing, but let me try to explain...

Source RPMs (.src.rpm) are intended to be used to rebuild a package. You're not trying to rebuild the kernel package itself here, so you probably don't want the .src.rpm, and you don't need to be pointing anything to the RPM building area (/usr/src/redhat).

Instead, what you want to do is build a separate package (ndiswrapper) against the kernel. So what you need is not the actual .src.rpm for the kernel, but instead the "kernel source" RPM, which is called "kernel-devel-2.6.21-1.3194.fc7.i686.rpm". If you install that, you'll get the proper kernel source and header files in /usr/src.

With most other packages the -devel just contains headers (.h) and libraries (.so) so this is less confusing... but for building kernel modules you need source code so that's what's inside -devel. It's just set up differently than the source inside the .src.rpm that's meant for building the actual RPM.

# ¿ Jun 20, 2007 16:29

Alowishus: Jan 8, 2002; My name is Mud

InferiorWang posted:

Is there a command to barf out a list of the hardware running on a machine?

Not necessarily a single one, but you can 'cat /proc/cpuinfo' to see the processor details, 'cat /proc/meminfo' for memory, and 'lspci' and 'lsusb' for onboard stuff and peripherals.

# ¿ Jun 25, 2007 16:34

Alowishus: Jan 8, 2002; My name is Mud

invid posted:

Any idea how can I go about configuring mailstats?

Are you running Sendmail? Or are you perhaps on a system which defaults to Postfix?

On current systems, sendmail.cf is generated from a file called sendmail.mc. If you just installed Sendmail but haven't configured it yet, it's feasible that sendmail.cf would be missing yet sendmail.mc would be present.

But I'm guessing you have another MTA... what distro are you using?

# ¿ Jun 27, 2007 17:15

Alowishus: Jan 8, 2002; My name is Mud

The VMware installer keys off some files that are already in place so you're going to have to delete them by hand to fool it into thinking it's a fresh install. Probably 'rm -rf /etc/vmware' will do the trick.

# ¿ Jun 29, 2007 16:50

Alowishus: Jan 8, 2002; My name is Mud

indigoe posted:

Although that wouldn't stop someone from hammering it with attempts.

DenyHosts will take care of that.

# ¿ Aug 11, 2007 20:21

Alowishus: Jan 8, 2002; My name is Mud

Harokey posted:

I thought I would pipe that into "cut" but it looks like cut only wants to read from a file, and won't go from stdin. Any other ideas?

cut should definitely work from stdin. This works for me:

code:

/sbin/ifconfig eth0 | grep HWaddr | cut -f11 -d' '

Not sure if using space as delimiter is ideal... something seems wrong about the MAC being field 11... perhaps there's a better tool or approach, but this will get you started.

Edit: ^^^ bang teapot is right on

# ¿ Sep 20, 2007 00:44

Alowishus: Jan 8, 2002; My name is Mud

aptitude happens to have a ncurses GUI, but it's not necessary. Just use the word 'aptitude' wherever you'd use 'apt-get' on the command line and it works exactly the same, with the benefits of some extra dependency tracking.

# ¿ Sep 24, 2007 23:26

Alowishus: Jan 8, 2002; My name is Mud

What do you mean, "won't create?" When e1000 is loaded, what do you see in dmesg? What does your /etc/conf.d/net look like? Did you symlink net.eth0 into /etc/init.d? When you try to start the network, what error do you see?

# ¿ Oct 12, 2007 15:20

Alowishus: Jan 8, 2002; My name is Mud

Sergeant Hobo posted:

One thing I was thinking of specifically in this context was that my clients (my desktop and my laptop) are both running Windows. Is there some kind of antivirus I can run on my Linux box that'll help fight off some of the virus crap and such out there?

ClamAV is your antivirus answer on Linux. But the question is how do you want it to protect your Windows boxes?

You could simply run a ClamAV scan of your shared drives every night in cron. If you want to get fancier, there is the vscan-clamav plugin for Samba that will run all traffic through a scan before serving it to or accepting it from your Windows boxes. If you want to be safe browsing and downloading, you could look into setting up a Squid proxy that runs web traffic through ClamAV as well.

Hope those are some ideas to get you started.

# ¿ Nov 18, 2007 03:12

Alowishus: Jan 8, 2002; My name is Mud

Sergeant Hobo posted:

I'm doing this in phases. First up will be basic DHCP serving and NAT stuff. If I can get that going, I'll get Squid, ClamAV and such set up. Somewhere in there, I will also be attempting (keyword there) to set up IPTables. Right now, that seems a little overwhelming but I'm sure I'll get it.

Sounds like a good plan, but your NAT and iptables are probably going to go hand-in-hand. I'll take this opportunity to recommend Shorewall for managing that firewall-type stuff. Yes, if you want, you can learn the arcane internals of iptables, but I personally don't see the point. Shorewall is a script that parses easily readable table-based config files and generates the iptables rules for you. It's hugely flexible, and makes management of more complex rulesets cake. And it's conveniently available in the Arch repository.

# ¿ Nov 18, 2007 18:24

Alowishus: Jan 8, 2002; My name is Mud

KS posted:

What's a good file system choice for a 2.2 TB raid array? It's for home data storage use. I formatted it ext3 and then did some reading that suggested there may be better options.

ext3 is fine. The biggest caveat is that a fsck of your 2.2TB volume will take an awful long time, but since it's a journaled filesystem the need for fsck will be rare. ext3 isn't necessarily the best performing filesystem, but for a home data store it's fine.

Depending on your distribution, the other filesystem that you might want to consider is XFS. It's made for large stores of data and generally tuned toward large file performance. That would be good for video and such but it's probably irrelevant because you won't be pushing those performance limits on a home server.

I say "depending on your distribution" because I'd only consider XFS if it's easily available. On CentOS or RedHat it's not. You can hack it in, but it's more trouble than just using ext3. But if you're on Ubuntu or another OS that provides the kernel code and support programs then go for it.

# ¿ Nov 25, 2007 22:04

Alowishus: Jan 8, 2002; My name is Mud

Harokey posted:

So I'm about 3 days into a fsck on a 500 gig drive with an ext3 filesystem. Is this normal? Should I be using a different file system maybe? Whats the best filesystem for an external drive?

Not normal. ext3 can take a while to fsck, but by "a while" I mean hours... even on 1TB+ filesystems.

3 days tells me something's physically wrong with that disk. Check 'dmesg' output for errors.

# ¿ Dec 6, 2007 07:50

Alowishus: Jan 8, 2002; My name is Mud

sund posted:

Does anyone have experience using USB security dongles in a guest OS using this fancy virtualization stuff? Am I dreaming or is this sort of thing possible?

Specifically which fancy virtualization stuff? You're running Windows virtualized on Linux? Many of the virtualization solutions can pass USB devices through, so it's potentially possible.

# ¿ Dec 14, 2007 06:10

Alowishus: Jan 8, 2002; My name is Mud

CentOS provides vsftpd right out of the box. Unless you need exceptionally fancy multi-host/multi-IP configuration, use the vsftpd that's provided and stay away from proftpd which has a spotty security history.

Your process will be a matter of 'yum install vsftpd' and then reviewing /etc/vsftpd/vsftpd.conf and possibly editing the userlist files. RedHat's vsftpd documentation is here and applies to CentOS as well.

# ¿ Jan 5, 2008 08:23

Alowishus: Jan 8, 2002; My name is Mud

fletcher posted:

How do I upgrade from PHP 5.1.6 to 5.2.5 on Fedora Core 6?

Best approach would be to upgrade to Fedora 8, which will get you to 5.2.4... and Fedora 6 isn't supported any more anyhow.

If you don't want to do that, you could try simply installing the PHP packages from Fedora 8. If those won't install due to dependencies, you could try grabbing the source RPM from Fedora 8 and rebuilding it (rpmbuild --rebuild) to see if that process generates FC6 installable packages.

If that doesn't work, check around some of the web hosting boards... there may very well be some folks maintaining PHP 5.2.x packages for older Fedora systems, as this is frequently an issue with webhosts who don't want to keep doing OS upgrades.

And last, if all else fails, take out the Fedora PHP RPMs altogether and just build your PHP of choice from source.

tripwire posted:

I feel really dumb about picking JFS as my filesystem when I installed ubuntu the other day. Is there anything cool jfs can do like that?

The coolest thing JFS can do that ZFS can't is run under Linux.

Don't feel dumb... paired with LVM, JFS will serve you very well. Yes, there are some fancy things ZFS can do that no other filesystem can, but as far as day-to-day use goes you should be just fine with what you have.

# ¿ Jan 30, 2008 19:29

Alowishus: Jan 8, 2002; My name is Mud

toadee posted:

Everything I have read suggests that having the following in /etc/security/limits.conf should help:

But it doesn't, does anyone have any other ideas about this?

Is it possible that your libPAM doesn't support those settings? The audio FAQs I've read seem to indicate that's a possibility...

Also what kernel version are you running?

# ¿ Jan 31, 2008 06:28

Alowishus: Jan 8, 2002; My name is Mud

DirtyDiaperMask posted:

...which I can't figure out how to kill completely since killing it restarts gdm, and loads MythTV again.

code:

sudo /etc/init.d/gdm stop

# ¿ Feb 5, 2008 19:34

Alowishus: Jan 8, 2002; My name is Mud

do it posted:

My current IDE drive is booting Ubuntu, but I'm not sure how to get Ubuntu to recognize the new harddrives on the controller card. I don't need the fakeraid that the card has, I just want the drives to be available as storage.

It should just see them, very unlikely that you'll need drivers. It's going to see them as "sdX" devices instead of "hdX" like your current IDE drive. I'd try "dmesg | grep sd" and see what you can see... chances are you'll see lines referring to sda and sdb, which are your two attached drives.

# ¿ Feb 13, 2008 03:11

Alowishus: Jan 8, 2002; My name is Mud

do it posted:

So the harddrives are definitely not appearing which leads me to believe it's an issue with the card. I'm not sure how to go about debugging that, though.

Hmm... yes, definitely seems to be the case. I'd expect that the card offers some sort of hotkey to enter a setup/diag utility during your BIOS POST. See if you can find that, and then maybe it will be obvious from there. If given the choice, you want to have each drive configured either as a JBOD or as a single RAID0 drive (different controllers handle single disks differently.) Other key words that you'd want would be AHCI instead of RAID.

# ¿ Feb 13, 2008 06:50

Alowishus: Jan 8, 2002; My name is Mud

Rescue Toaster posted:

Is it straightforward to set it up so me & my roommate can both use VNC (are the kids still using VNC these days?) at the same time to connect to separate X instances logged in as ourselves?

It's definitely possible... each user can run their own vncserver and they each get their own desktop... you'll just have to connect to different ports ("screens"). But as Col said, NX is much faster displaywise, so your choice.

quote:

Is it practical to use a small compact flash (connected to IDE) or even a USB flashstick as the /boot partition so root and everything else can be on the RAID array? Does /boot get written to a lot - wearing out the flash?

Absolutely. /boot doesn't get written to at all unless you're upgrading the kernel or modifying GRUB's config. A lot of people run /boot unmounted or under autofs.

# ¿ Feb 15, 2008 17:16

Alowishus: Jan 8, 2002; My name is Mud

rugbert posted:

I need to know where it stores its web documents so I can get webalizer working but I cant seem to find them anywhere (and is also listed as a virtual package I asked about earlier).

Zope serves documents out of its own data store. That data store is kept in your Zope site's "instance home" under the 'var' directory with the name Data.fs. It's a big-rear end ZODB blob.

quote:

Im just trying to get webalizer to dum its output in the same place the documents served from.

You have a couple options... since you're running Zope behind Apache, you could us an Alias statement to make a special http://www.domain.com/stats URL that gets served from Apache while the rest of http://www.domain.com still gets served through Zope.

There are also external filesystem objects that you can put into your Zope site that are able to reference files on the actual filesystem, but those are generally better for static data... probably not appropriate for Webalizer reports since they change and grow with each run.

Third option is to find a way to push the Webalizer output into Zope's database itself. You generally can do that via FTP or WebDAV.

That said, I'd suggest the Alias approach as the ideal one... the other two would only be for exceptionally weird circumstances.

That help any?

# ¿ Feb 20, 2008 00:50

Alowishus: Jan 8, 2002; My name is Mud

rugbert posted:

so I added this line
Alias /webalizer "var/www/mysite.rugbert.com/webalizer"

If that isn't a typo, you should have a leading / in front of var. Since it's a directory alias, you might also want a trailing slash after the webalizer alias, and then be sure to visit http://mysite.rugbert.com/webalizer/.

Also you're going to need URLs matching that Alias line to be excluded from any of the rewrite rules you're using for Zope... see example below.

quote:

also - Should this line be in the apache.conf file or the site-enabled/referenced file?

Should be in the sites-enabled config snippet. That's included in apache.conf, but it's cleaner to put the Alias along with the rest of the site config, and if you're using VirtualHost definitions it's pretty much required.

Here's a code snippet for a similar situation I have on one of my Zope servers. In this case it's not webalizer, but the pipermail archives for a mailing list... but same thing - a big directory of HTML that's not served through Zope.

code:

Alias /pipermail/ /var/lib/mailman/archives/public/

RewriteEngine on
RewriteLog "/var/log/apache2/rewrite.log"
RewriteLogLevel 0

# Don't rewrite static content into Zope
RewriteRule ^/pipermail - [L]

# Use mod_proxy and Zope's VirtualHostMonster to serve Zope from behind Apache
RewriteRule ^/(.*) http://localhost:9673/VirtualHostBase/http/www.sample.com:80/site_main/VirtualHostRoot/$1 [P]

# ¿ Feb 21, 2008 00:41

Alowishus: Jan 8, 2002; My name is Mud

rugbert posted:

ok so in the apache2/site-enabled/mysite.rugbert.com file (where all the virtual host information is) I added

...

but it doesnt work. I just get the the 404 page the site uses.

I think you're still missing some parts... or I am not clear on the rest of your Apache config. First of all, is the 404 page you're getting served out of Zope? Sounds like it is... that's definitely an indicator that the rewrite isn't working. From the snippet you posted, I can see why - all you did was turn the engine on, but you didn't give it any rules.

But here's the thing - I would have expected RewriteEngine to already be on in order for Apache to sit in front of Zope. Can you post the chunk of Apache config where it's wired up to serve the Zope content? If it's not a Rewrite thing, I'd guess it's a ProxyPass setup.

# ¿ Feb 21, 2008 18:38

Alowishus: Jan 8, 2002; My name is Mud

Un-l337-Pork posted:

Ok, so I have a bunch of mounted logical volumes or something. This doesn't really look surprising to me, but on my previous install, my volume mounted on / would've contained the remaining free space. In this case, it looks like I should have ~140G or so sitting somewhere. This is how the machine came from Dell. We have done virtually nothing to it at this point.

Dell set you up with LVM. Your space is there, it's just unallocated. This is a good thing - it lets you take that remaining space and apply it where and when you need.

LVM has three layers:

code:

Physical Volumes (PV)
  Volume Groups (VG)
    Logical Volumes (LV)

In your case, your PERC 5/i is presenting one PV, /dev/sda. It looks like Dell created a single VG that encompasses all that physical space. Inside that VG, they created 5 (or 6, one is missing from your list, it's probably swap) LVs. If you were to type "vgdisplay VolGroup00" you'll get a report containing some lines like this:

code:

  VG Size               406.03 GB
  PE Size               32.00 MB
  Total PE              12993
  Alloc PE / Size       4126 / 128.94 GB
  Free  PE / Size       8867 / 277.09 GB

You should all of your ~160GB for "VG Size". Then under the Alloc Size, you'll see that maybe only 20GB is taken, and under Free Size will be your missing ~140GB.

To utilize that, you can either use 'lvcreate' to make some new logical volumes and mount them where you please, or 'lvextend' to add more space to one of the existing volumes. Of course once the LV itself is extended, you'll then have to expand the actual filesystem... resize2fs is the tool you'll want for that.

# ¿ Feb 21, 2008 18:46

Alowishus: Jan 8, 2002; My name is Mud

rugbert posted:

What sort of line am I looking for? I cant find any mention of zope in my apache.conf file nor did "/.fs" give me any results.

I'm not really sure what to tell you, outside of my guess that the line isn't actually going to say anything about Zope but will instead have some sort of Rewrite or ProxyPass verbage to it.

The usual setup is that Zope listens on a higher non-standard port (like 9673 on one of my Debian systems), and then Apache sits on the standard HTTP port 80 and listens for requests. It redirects most of those requests via proxy to the Zope system listening on the non-standard port. You need to figure out this mechanism so that Apache can be told not to redirect certain requests (the ones for your webalizer stats).

It's also possible that Zope itself is serving the webpages and Apache has nothing to do with it. Do you not have access to the person who set this up in the first place? If you can find the zope.conf file for your site instance (likely somewhere under /etc), it may give some clues. If Zope is configured to listen on port 80, then it's doing all the web serving itself.

# ¿ Feb 29, 2008 20:51

Alowishus: Jan 8, 2002; My name is Mud

rugbert posted:

aha! It's located in the virtual host file

Ok good... well you may have already done this, but now that you know where the rules are, go back to my previous post with my rewrite examples and see if you can adopt those for your situation. At least you'll be able to test now and see some results...

# ¿ Mar 4, 2008 16:30

Alowishus: Jan 8, 2002; My name is Mud

Twlight posted:

We already have a mail server at work but I'm not sure how to configure sendmail to work with an existing email server.

You need to configure sendmail to use a smarthost (where the smarthost is your existing mailserver):

code:

define(`SMART_HOST',`[smarthost.example.net]')dnl

That has to go into sendmail.mc and then you have to generate a new sendmail.cf. If that's not something familiar to you, I really recommend 'yum install postfix' followed by 'yum remove sendmail'. Postfix isn't nearly as retardedly difficult to configure as sendmail is... you'll be able to just read through the comments in /etc/postfix/main.cf and make pretty good sense of everything ("relayhost" would be the Postfix parameter for this particular situation).

# ¿ Mar 4, 2008 16:34

Alowishus: Jan 8, 2002; My name is Mud

DEAD MAN'S SHOE posted:

I want to be able to run `mount -t cifs` as a user, on the fly. I don't want to have to specify the share in /etc/fstab (which is a stupid idea anyway).

If you were to put it in fstab with the "user" option, it should allow non-root users to mount it.

Are you saying it's stupid to put it in fstab due to the password info? If so, then there is a way to tell smbmount to reference a password in another file which can be restricted to root readable, though I'm not sure whether that will work in concert with the user option.

Another approach may be to use autofs.

# ¿ Mar 5, 2008 00:43

Alowishus: Jan 8, 2002; My name is Mud

DEAD MAN'S SHOE posted:

Not only the password issue but having to specify it prior in a root-only config file defeats the idea of being able to mount random shares on the fly.

Ah sorry didn't realize the goal was arbitrary SMB mounts.

Ok, first of all the sudo solution that Grey Area suggested can be configured to work without a password.

Second, are you *sure* that you changed the right binary files with the right chmod command? Straight from the 'mount.cifs' manpage - "It is possible to set the mode for mount.cifs to setuid root to allow non-root users to mount shares to directories for which they have write permission." That sounds like exactly what you want.

The third approach would be to install FUSE and the SMB module, since that's all user-space stuff and no root is required.

# ¿ Mar 5, 2008 23:14

Alowishus: Jan 8, 2002; My name is Mud

DEAD MAN'S SHOE posted:

Yeah, I thought I had.

I'm not certain of my switches here but I'm sure this worked on the previous Ubuntu release

Call /sbin/mount.cifs directly:

code:

$ mount -t cifs //server1/library /mnt/test
mount: only root can do that

$ /sbin/mount.cifs //server1/library /mnt/test
Password: <Samba Passwd>
$ mount
...
//server1/library on /mnt/test type cifs (rw,mand,nosuid,nodev,user=username)

Also the user has to own the mount point. Sorry this didn't occur to me earlier!

# ¿ Mar 6, 2008 07:25

Adbot: ADBOT LOVES YOU

# ¿ Apr 25, 2024 22:43

Alowishus: Jan 8, 2002; My name is Mud

Slow is Fast posted:

Apparently mythtv-setup isn't getting into the sql database properly so terminal is just barfing errors at me.

First verify that MySQL is actually running.

Second, is the MySQL server at 192.168.1.110 on the same machine, or a different one?

If on the same machine, try connecting to 127.0.0.1. MySQL's access control will prevent connections via any other address unless you go in and explicitly allow it.

# ¿ Mar 6, 2008 23:59

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

«‹›3 »