|
They were just (bad) examples. Most commands I want to run are not changing files like that. Edit: I found my answer: clusterit. The only problem so far is that it doesn't appear to loop back. So I would need to run it from one of the nodes that isn't part of the cluster. Which is better than nothing, unless someone knows how to make it loop back? Edit2: Just made a script to execute it locally, then on the cluster. Sir Sidney Poitier fucked around with this message at 15:34 on Apr 19, 2011 |
# ? Apr 19, 2011 05:38 |
|
|
# ? Jun 6, 2024 11:25 |
|
The other day I noticed that the hard disk light on my computer is constantly on (doesn't blink). Is there a utility that allows me to monitor actual reads/writes to the disk so I can figure out if something is actually writing to it or the light is getting incorrect signals?
|
# ? Apr 20, 2011 14:51 |
|
Modern Pragmatist posted:The other day I noticed that the hard disk light on my computer is constantly on (doesn't blink). Is there a utility that allows me to monitor actual reads/writes to the disk so I can figure out if something is actually writing to it or the light is getting incorrect signals? code:
|
# ? Apr 20, 2011 15:49 |
|
Bob Morales posted:
Awesome. Thanks. Looks like it's just the light since there is no activity. VVVV Well I'll be damned. That is quite useful. Thanks. Modern Pragmatist fucked around with this message at 16:06 on Apr 20, 2011 |
# ? Apr 20, 2011 16:00 |
|
While iostat is pretty good for determining how much IO is happening per device, iotop is better for isolating a particular process.code:
|
# ? Apr 20, 2011 16:01 |
|
I'm trying to automate a few things that I do regularly. I created a few scripts that work exactly as I'd expect them to: run a few things in sequence. However, I've run into something that I can't seem to script. Normally, I run, from an interactive bash shell: code:
code:
code:
How do I run something and have it detatch from a script?
|
# ? Apr 21, 2011 20:58 |
|
Any reason you can't just run the script like: ./script.sh &
|
# ? Apr 21, 2011 22:02 |
|
Well, I intend to put other stuff below it at some point, so that would entail creating a shell script to wrap a single command, so an extra copy of the interpreter, plus multiple files is a lot less clean. I'd assume there's a correct way to do what I need to do.
|
# ? Apr 21, 2011 22:12 |
|
Maybe I'm missing something here, but none of your examples had a & at the end of the line. I made a quick test and stuff like "cat /dev/urandom > /dev/null &" runs in the background from a script. The examples I see for nohup also have the ampersand at the end.
|
# ? Apr 21, 2011 22:48 |
|
Well, "> /dev/null 2>&1" was something that I remembered using years ago when I used Linux daily, so I pasted it in. It turns out that ">&" is totally different from "&", so "> /dev/null 2>&1 &" works. I was assuming that "2>&1" not only redirected STDERR into STDOUT, but also detached it, when that wasn't the case.
|
# ? Apr 21, 2011 23:16 |
|
There's actually three different things going on here - detachment, IO redirection, and backgrounding. '&' is "run in background". The command is still "attached" to the terminal it was started from, but other commands can be started in the meantime. In an interactive session, at least, background jobs can be moved to the foreground with 'fg' (and back to the background again with ctrl-Z 'bg'). '&n' is "file descriptor n". '> /dev/null' is "redirect fd 1 (stdout) to /dev/null", and '2>&1' is "redirect fd 2 (stderr) to fd 1 (stdout)". There's no relation to the sole '&' used to run a program in the background. Finally, 'nohup' detaches the program from the terminal so that it won't try to read from it or write to it and so that it won't get a NOHUP signal when the terminal closes.
|
# ? Apr 22, 2011 17:57 |
|
ToxicFrog posted:'&n' is "file descriptor n". '> /dev/null' is "redirect fd 1 (stdout) to /dev/null", and '2>&1' is "redirect fd 2 (stderr) to fd 1 (stdout)". There's no relation to the sole '&' used to run a program in the background. Also the order matters. If you do: $ foo >/dev/null 2>&1 It sends both stdout and stderr to /dev/null. If you do: $ foo 2>&1 >/dev/null It sends stderr to stdout and stdout to /dev/null. That's because it works left-to-right, and '&n' equates to the current target of the specified descriptor.
|
# ? Apr 22, 2011 19:22 |
|
Is SElinux of any benefit to keep on? It is a real headache, this server is going to be for the whole company so I would like as much security as I can get, but this is just getting in my way constantly.
|
# ? Apr 25, 2011 00:56 |
|
If you run a web server yes. SElinux is not that difficult to learn and pretty easy to maintain once you get the hang of it. If your server can't be accessed externally though then who cares.
|
# ? Apr 25, 2011 02:00 |
|
Pram posted:If you run a web server yes. SElinux is not that difficult to learn and pretty easy to maintain once you get the hang of it. know any good guides to learning it
|
# ? Apr 25, 2011 02:41 |
|
I reference these when I have to deal with it: http://wiki.centos.org/HowTos/SELinux http://www.centos.org/docs/5/html/5.1/Deployment_Guide/rhlcommon-chapter-0017.html
|
# ? Apr 25, 2011 02:54 |
|
Pram posted:I reference these when I have to deal with it: Yeah I am just not going to use it, it seems to cause me a lot of headaches
|
# ? Apr 25, 2011 04:46 |
|
I like the Red Hat docs on it: http://docs.redhat.com/docs/en-US/Red_Hat_Enterprise_Linux/5/html/Deployment_Guide/selg-overview.html Running a strict policy on an internal server is probably paranoid. But if it gets in your way, it'll get in the way of whoever you're worried about. If it's really a risk worth worrying about, you might want those kind of hurdles in the way. Targeted policies have pretty solid man pages for whatever service you're protecting (e.g. man httpd_selinux). What you're complaining about is why people reply "disable it" to questions like yours.
|
# ? Apr 25, 2011 05:48 |
|
bort posted:I like the Red Hat docs on it: Yeah I was thinking that, maybe once I start making 30k or more I will look at it
|
# ? Apr 25, 2011 16:56 |
|
Corvettefisher posted:Yeah I was thinking that, maybe once I start making 30k or more I will look at it Thats the spirit.
|
# ? Apr 25, 2011 18:05 |
|
I'd like to also point out that "disable it" won't be a good option on Red Hat platforms in the future now that Fedora is actually leveraging SELinux's capabilities model to provide RBAC instead of doing dumb things like making ping setuid root.
|
# ? Apr 25, 2011 18:49 |
|
Does any one have any experience with btrfs? I was reading this: https://btrfs.wiki.kernel.org/index.php/Using_Btrfs_with_Multiple_Devices and I would like to know if I can mix devices with partitions? So I have created a btrfs file system with mkfs.btrfs /dev/sdb, can I then add /dev/sda4 , or something to that effect, without screwing up /dev/sda1,2,3, which are different file systems?
|
# ? Apr 25, 2011 20:53 |
|
Quick question that hopefully someone has a direct answer to, since all my googlin has left me blank. Is there any live/install linux distro that will boot an older model sparc processor, and mount drives left from a solaris install? Long story short of why I have to do this is I might have to do some quick and fast and mean work tomorrow morning rooting a solaris 7 sun machine to get an old rear end scantron to work.
|
# ? Apr 25, 2011 21:26 |
|
^^^ Maybe http://www.martux.org/ (Found with google search, never used it) e: I'm pretty sure the Sparc gentoo livecd can do what you need. Kaluza-Klein posted:Does any one have any experience with btrfs? The device is the entire drive, including the partition table and all partitions. If you format the entire device, you will wipe out the partitions. Regardless of the filesystem. taqueso fucked around with this message at 22:21 on Apr 25, 2011 |
# ? Apr 25, 2011 22:17 |
|
Not sure what filesystem it needs to mount, but FreeBSD has a SPARC64 Live CD. If the machine is 32-bit SPARC that will be trickier. OpenBSD is only one of two *nix distributions that still has 32-bit SPARC support, and I can't remember the other: OpenBSD 4.8 here.
|
# ? Apr 25, 2011 22:24 |
|
Pram posted:Thats the spirit. I forced HTTPS logins, complex passwords, encrypted the databases and server, set up antispam, IDS, and virus scans on documents. I think that is okay for right now, prior to my doings we did have any of this. Dilbert As FUCK fucked around with this message at 22:45 on Apr 25, 2011 |
# ? Apr 25, 2011 22:39 |
|
poo poo, I have no clue why I didn't think of gentoo or a BSD. . . Just a clue to how loopy my head is these days.
|
# ? Apr 26, 2011 01:27 |
|
Kaluza-Klein posted:Does any one have any experience with btrfs? You can but unless sdb and sda4 are the same size it might not work like you want it to.
|
# ? Apr 26, 2011 01:29 |
|
I have two physical drives in my "pool" now. Ubuntu or Gnome or Nautilus is confused though. In my fstab I mount one drive in the pool, and btrfs magically finds and uses both drives, but I guess the desktop isn't aware of this. It sees the second drive and lists it as one of the "devices" (like removable storage) in nautilus. Is there any way to stop this? I assume if I attach another bare drive to this pool, it too will appear here. Am I making any sense?
|
# ? Apr 26, 2011 15:41 |
|
Having a samba nightmare. Running Fedora 14. What I'm hoping to do is setup a share with anonymous r/w access for some Win 7 and Xp/2003 machines which are on a AD domain. Furthest I can get is having the share visible but not accessible (attempting to access the share gives me the generic "not accessible" error). This happens on both machines. Hoping to get a little bit of handholding on this - I've looked around and tried dozens of different variations in the smb.conf file, changing file permissions/owners but something is sailing over my head. Can anybody post a basic config that should allow what I'm looking for? At this point I don't really care where the share is located, I just want to get something up and running. I've heard that Win 7 complicates things, so I'd even settle for getting access from the xp/2003 machines. Crackbone fucked around with this message at 16:51 on Apr 26, 2011 |
# ? Apr 26, 2011 16:49 |
|
Can you goto System->Administration->Disks and see if you can turn of auto-detect?
|
# ? Apr 26, 2011 16:49 |
|
Crackbone posted:Having a samba nightmare. Not accessible is a permissions error. Check the logs for more info, and try using SWAT if you're having a lot of trouble. I can post my smb.conf once I'm at work, as that machine is running Fedora 13 and hosts a share as you describe.
|
# ? Apr 26, 2011 17:15 |
|
TheGopher posted:Not accessible is a permissions error. Check the logs for more info, and try using SWAT if you're having a lot of trouble. I can post my smb.conf once I'm at work, as that machine is running Fedora 13 and hosts a share as you describe. What logs?
|
# ? Apr 26, 2011 17:27 |
|
Crackbone posted:What logs? look in /var/log/samba/
|
# ? Apr 26, 2011 17:36 |
|
Crackbone posted:Samba stuff Remember that in addition to the permissions specified in smb.conf file, you need to make sure that you have the actual folder permissions set properly. Here is my smb.conf (Version 3.5.8-74.fc14) code:
|
# ? Apr 26, 2011 18:37 |
|
Modern Pragmatist posted:Remember that in addition to the permissions specified in smb.conf file, you need to make sure that you have the actual folder permissions set properly. If I'm reading this correctly though, that share requires a user to provide an user/pass to connect? I'm wanting an r/w share without the user being forced to provide credentials. I guess the problem is I don't know what the proper way to implement that is. I've diddled around with several different things like guest only, map guest = bad user/pass, force user, etc.
|
# ? Apr 26, 2011 19:42 |
|
This is one I have that is writable by anyone [archive] comment = Archive Data path = /exports/archive public = yes writable = yes write list = @staff hide unreadable = yes
|
# ? Apr 26, 2011 20:02 |
|
The following works on my Fedora 13 machines with samba-3.5.8-74.fc13.i686.code:
# ls -ld share drwxrwxrwx. 24 $user root 4096 Apr 19 17:05 share I probably could do better with those permissions, but I'm not too concerned as this is not a public-facing terminal and I was having difficulties getting it to work.
|
# ? Apr 26, 2011 21:06 |
|
TheGopher posted:The following works on my Fedora 13 machines with samba-3.5.8-74.fc13.i686. Thanks for the config, I'll give it a shot. Did you have any issues with SELinux settings? Coworking is swearing up and down that's part of the problem.
|
# ? Apr 26, 2011 21:16 |
|
|
# ? Jun 6, 2024 11:25 |
|
Forgive my ignorance but is there a simple shell command that would delete all files with the extension .jpg and .ini in a directory and all its sub directories?
|
# ? Apr 26, 2011 22:17 |