Printable View
Save yourself some keystrokes.
This will update everything including the kernel, assuming there is a new one available. I can post a how-to if people want to compile and install a custom kernel for their Ubuntu rig based on the latest source code (currently 3.1.6 I think. I compiled mine yesterday.)Code:sudo apt-get update && sudo aptitude full-upgrade
Ok, how-to for setting up the remote BOINC management.
Assumptions:
You have installed Ubuntu server with OpenSSH
Procedure:
-Log into the remote machine with your SSH client of choice.
-Once you're logged in install BOINC from the repositories
the package boinc-manager is not required, ignore it.Code:sudo apt-get install boinc-client
-Next we need to open an editor and alter one of the config files to allow remote management of the client. I'll be using nano here but there are others.
You'll see something like this:Code:sudo nano /etc/boinc-client/remote_hosts.cfg
http://www.bitbenderforums.com/~dark...ote_hosts1.jpg
NOTE: Use the cursor keys to move around, don't use the mouse. This is a terminal text editor, not a gui application.
-Move the cursor down and add the host name or IP address (if you use static IPs on your network) of the system you want to be able to administer BOINC from where I have my chosen IPs entered.
-To save the file press ctrl-o (hold the control key and press the letter o) then press Enter to confirm the file name.
-To exit the editor press ctrl-x
-For BOINC to read the config file you need to restart the application
-Ok, now you're done. Exit the terminal and you can handle all the client management from the BOINC Manager on your desktop system, no need to play with the command line. Just open the BOINC Manager and click through Advanced > Select Computer and use the drop box to enter the hostname or IP of the remote system.Code:sudo /etc/init.d/boinc-client restart
I am, by the way, getting pretty ticked off over not being allowed to post attachments in any way, shape or form. :(
Wow, good post. I am still a unix noob. I know just enough to be dangerous.
If you like that you'll love this. ;)
Ubuntu is a bit of a bugger for not updating BOINC in the repositories but since it sprays the BOINC files around like hippo manure in the file system it can be a little daunting for a new guy wanting to update the client. Here's the skinny.
Open a terminal (via ssh/Putty or locally) and enter:
to stop the client andCode:sudo /etc/init.d/boinc-client stop
replace the version numbers with whatever version you're after. If you're not sure go here and look: http://boinc.berkeley.edu/dlCode:wget http://boinc.berkeley.edu/dl/boinc_6.12.34_x86_64-pc-linux-gnu.sh
This will download the file directly from Berkley to the current directory, which should be your home directory.
Next set the file as executable
and then run itCode:sudo chmod +x boinc_6.12.34_x86_64-pc-linux-gnu.sh
That will create a directory called BOINC in your home folder. Change into it.Code:./boinc_6.12.34_x86_64-pc-linux-gnu.sh
now we'll copy the relevant binariesCode:cd BOINC
and we restart the clientCode:sudo cp boinc* /usr/bin/
That should be it. You're now running the new client version. Since Ubuntu pretty much never update it you don't have to worry about it being overwritten, but if it is just do the same thing again to change it back or update to another newer version, or back to an older one if there's a problem.Code:sudo /etc/init.d/boinc-client start
An apt analogy. I once saw a hippo doing that at the San Diego zoo. Quite impressive.Quote:
Originally Posted by D_A
Buh dum PSSHHH! And the race is on for the last pun of 2011.Quote:
Originally Posted by shifty803
<bump>
I'm working on something that folk running Ubuntu-server should find useful with the upcoming (whenever it gets here) GPU client for HCC. Bumping the thread so I don't lose it.
Note: this is for an nVidia equipped machine. I don't have an ATi/AMD card to work this out on.
Ok, got it running on a remote machine.
Starting from a machine that is already running a CUDA etc capable version of BOINC and Ubuntu-server (I'm working with 10.04 64bit)]
Install a CUDA/OpenCL enabled nVidia driver. (I'm running 280.something now and downloading 290.something as I type this).
copy it to your home directory on the server, then make sure the file is executable
now run the installerCode:sudo chmod +x NVIDIA-Linux-x86_64-290.10.run
answer all the questions and let it run, it should dump you back to the command line when it's done with no errorsCode:./NVIDIA-Linux-x86_64-290.10.run
next we need to set up a little something so that the nvidia modules load without the X server running
nano is a text editor and this just started creating a new file with it.Code:sudo nano /etc/init.d/nvidia_cuda
now copy and paste this into the new file AS IS
Press Ctrl+o (hold Ctrl and press the letter o) and then just press Enter. Next press Ctrl+x to exit back to the command line.Code:#!/bin/bash
PATH=/sbin:/bin:/usr/bin:$PATH
/sbin/modprobe nvidia
if [ "$?" -eq 0 ]; then
# Count the number of NVIDIA controllers found.
N3D=`/usr/bin/lspci | grep -i NVIDIA | grep "3D controller" | wc -l`
NVGA=`/usr/bin/lspci | grep -i NVIDIA | grep "VGA compatible controller" | wc -l`
N=`expr $N3D + $NVGA - 1`
for i in `seq 0 $N`; do
/bin/mknod -m 666 /dev/nvidia$i c 195 $i;
done
/bin/mknod -m 666 /dev/nvidiactl c 195 255
else
exit 1
fi
Now enter
this will set up the init script that will load the nvidia modules and create the processor nodesCode:sudo chmod +x /etc/init.d/nvidia_cuda
sudo update-rc.d nvidia_cuda defaults
Now reboot the systemWhen it loads back up it should see your CUDA enabled GPU (mine did at least)Code:sudo reboot now
credit where it's due, I got this from HERE
OpenCL has been supported by the nVidia drives since release 280.13 and they're currently up to 290.10 so check your driver releases.
Will your Nvidia guide work on ubuntu 11.10 desktop?
I think the instructions I followed were actually for 11.10 originally. For a desktop system you don't need to muck about with device nodes and such they're already there, just install the driver and you should be good. Install the driver, reboot and you should be ok. If it complains about the Nouveau driver the nVidia driver can put that in the black list but you'll need to get to the recovery console to get the nVidia driver installed afterwards which means getting to the grub menu. To do that you need to keep quickly tapping the left shift key as the machine boots, and I really do mean you have to keep doing it and do it QUICKLY. This menu is a bugger to get into. Once you're in the Grub menu, select the "recovery" option for the kernel you're using (VERY important to pick the right one, it should be the first recovery option from the top of the list) and when you get to the next menu pick the root terminal (very bottom) option.
You will now be in a terminal, don't even think about trying to start a desktop session or this won't work.
cd to the directory where you saved the nvidia driver and run it ( ./NVIDIA<blah blah blah>.run ) It will complain about being started in this mode but just go through it and answer all the questions. Once it's done type "exit" to get back to the last menu and then select "boot normally". That should be it done. If the driver is running then BOINC will see the CUDA/OpenCL device.
Worked like a charm, running on gpugrid now. One thing I ran into though was I had to delay boinc starting or it wouldn't see the gpus. It seems boinc loads before the graphics driver and boinc gives the message no gpus found.
Yes, my reply to this got lost in the forum update.
Here's the most elegant solution I found
This removes BOINC from the automatic startup sequence and re-adds it right at the very end so everything else will have time to load.Code:sudo update-rc.d -f boinc-client remove
sudo update-rc.d boinc-client defaults 99
Just had my Linux Mint for about a week and the Boinc I got from the repository. All seems well but I obviously need to get the latest and greatest.Quote:
Originally Posted by D_A
Can I use this method??
I believe so, yes. Mint is close enough to Ubuntu behind the curtains that this should all still work fine.
So I got Ubuntu 11.10 x64 installed and Boinc v7.0.20 (beta) installed and running but I have a few questions.
- When I run the client it shows that I'm running the x64 version as intended but when I run the manager and click on "about boinc manager" it tells me that I have the x86 version. What's going on? Will it affect anything? I'd like to fix it.
- How do I set the client to auto run on start-up? The first page of this thread is missing many images.
- How can I make a shortcut on desktop or taskbar (? thing on the left side with large buttons) for my boinc manager so I can easily open it at will.
- I tried pwnconfig in an attempt to slow down this Ultra Kaze 3000 ( o_o ) but it wouldn't work. Is there any other way to go about this on Ubuntu?
Honestly I'd ignore that. 7.0.20 is ALPHA software so if you're not ready and willing to join the alpha mailing list, contribute to it, report any and all bugs including those it may cause with other applications and expect any amount of work errors then don't run it. It's much safer and more reliable to run the current production version, whatever that is at any point in time.Quote:
Originally Posted by Gamekiller
Quote:
- How do I set the client to auto run on start-up? The first page of this thread is missing many images.
- How can I make a shortcut on desktop or taskbar (? thing on the left side with large buttons) for my boinc manager so I can easily open it at will.
I take it you didn't use Aptitude to install from the repositories and update the client but rather just grabbed the client straight from Berkley. What you need to do is add the launcher script (run_client) to the session manager (you might call it "startup applications") so it starts when you log in and add a shortcut to the panel or whatever to launch the manager. You could, (I haven't done this) add a script to /etc/init.d that calls the run_client script when ti boots into run level 3 (I think that's the one you want) but that's not exactly beginner stuff.
I have no experience with the Unity interface (mobile phone interfaces are for phones and tablets at a pinch and have no place on the desktop imho. I stopped using Ubuntu on my desktop because of that monstrosity) so I'm afraid I can't be more precise than that.
A quick google to find what either is tells me that you need to have lm-sensors properly configured before trying to run pwmconfig.Quote:
- I tried pwnconfig in an attempt to slow down this Ultra Kaze 3000 ( o_o ) but it wouldn't work. Is there any other way to go about this on Ubuntu?
Useand get the modules loaded before trying to run pwmconfig. Other than that you might try checking your BIOS settings to see if those fans can be set to automatic control, some only run at fixed speed no matter what you try.Code:sudo sensors-detect
Ah, been running the 7.0.x series on my main rig without issue (though it's Win 7) so I figured it'd be alright. So far so good. *crosses fingers*Quote:
Originally Posted by D_A
I used the method from the first post of this thread to install it except using the newer version of Boinc. Alright, got it running on start-up and shortcut added. As for the Unity interface, I agree, but not an issue if not a main rig. :p:Quote:
Originally Posted by D_A
I've already completely set up lm-sensors and loaded the modules before trying. I've already tried setting the bios to auto fan control but it doesn't do anything and there aren't any adjustments. :(Quote:
Originally Posted by D_A
Thank you for the help!
Happy to help. ;)
As for rigs other than this one, I typically use Ubuntu Server purely for ease of administration. The lack of a gui also appeals to me, especially for a headless rig, as there's no point in wasting cycles on a gui that's not going to be used.
As for the 7.x series on linux ... I discovered that 7.0.18 would trash CEP2 and HPF2 units almost instantly and reliably on Slackware 13.37 x86_64, Ubuntu 10.04 x86 Desktop, 11.10 x86_64 Server and 12.04 i386 Server beta. Other work seemed to process ok. I posted to the alpha list and WCG forums but never got any response so I went back to 6.12.43.
Did anyone see my white stick or my dog :D
My story goes like this: was running the repository download in mint
Did the "If you like that you'll love this"
now I am thrashing around in the dark. Until now I always used boincmanager but although system monitor would suggest things are running I cannot start boincmanager (at least from where I used to in the graphical interface)
So, Can I connect via a port from my main rig? or Is there another way to start boincmanager? or do I have to download boincmanager again this time from berkley?
Not critical until I need to stop or pause or make changes I guess but... Heeeeeelp :)
You still need to set up the remote access files just like before, that hasn't changed. Mint is basically just Ubuntu with a better front end, so everything you learned administering Ubuntu still applies.
Putting together another box..... tried to "sudo apt-get install ia32-libs" and for the past 24 hours I get a '404 not found' for a bunch of the files needed.
Thoughts anyone?
A voyage of discovery for me... I discovered "apt-get update" which seems to update the repositories from which the various bits and pieces are collected.
Just a guess here but when items are updated I would say that instead of the new being put where the old were they are each given a new address too
Got ia32-libs loaded now
Still struggling to start manager though. eventually I must get around to learning how to do this remotely :rolleyes:
There are two repositories at work here. Firstly the main ones, next the update repos. If the package in the update repository is newer than the one in the main repo then during a system update it is downloaded and replaces the original.
On the local system, when a package is updated it is first removed, while leaving the data files in place, and then the new package is installed. If you purge a package it removes everything.
Ahh, Oldchap, your in good hands with D_A! :up:
I must agree with you there MM :)
So D_A, why do think it would be that when I download from the repository I can run Manager but when I "wget and then sh" even though I see the "use /home/*username*/BOINC/run_manager to start BOINC" trying to do so results in ...nothing at all?? The client will run though but I don't know how to do all the attach and configure in a text only environment.
It feels like a simple step I have missed but I don't see it.
When you install from the repo the manager and everything else gets installed in locations that are in your path. There's a hidden file in your home folder that defines several paths to where stuff might be installed. /usr/bin is the default place for programs that you as a regular user are allowed access to so that's where the boinc, boinccmd and boincmgr files are put. THe script /etc/init.d/boinc-client not only starts the client but defines where the data directories are.
The script run_manager does exactly what it says. It runs the manager, not the client. The catch is that the manager requires an X environment to run. It's a graphical application, so without a graphical interface (in Unix terms an "X server") it won't run. That's also assuming you have all the required libraries and such on the system.
If you have a GUI environment, go to the BOINC folder, open a terminal and enterThis will output what libraries are required by the program "boincmgr", what versions it has found on your system and if any are missing or the wrong version. Copy and post the output here if you want any explanation of the output.Code:ldd boincmgr
If this is going to be a headless machine you could just alter the remote_hosts.cfg file to allow the IP address on your local network of the machine you want to use to manage it and do it all from the BOINC Manager on your main rig (Windows/Linux/Mac/Android doesn't matter for this, and yes I did mean to put Android in that list).
Another alternative is to install from the repositories and then use the files extracted from the Berkley download to update it. It's not hard and I can give you copy/paste code line by line.
In case anyone wants the commands:
I'll assume you're starting with a fresh system.
Download the installer file to your home directory and run it.
Open the newly created BOINC folder and open a terminal there.
If the terminal opens in your home folder just cd BOINC to get to the right one.
If there are any major issues just post 'em up.Code:sudo su
apt-get update
apt-get install boinc
/etc/init.d/boinc-client stop
cp boinc boinccmd boincmgr /usr/bin/
/etc/init.d/boinc-client start
exit
With Linux mint everything was very simple to setup. My ppd went up after moving to Linux from W7 64.
Mint is pretty much Ubuntu without that disasterous Unity interface. I'm running Mint 13 with the XFCE desktop right now.
Cinnamon Mint Maya (the no codecs version) is what is on the 2600K
Now have the same on a 3770K but the MIVG has an issue with not allowing dual channel ram (broken pin). It remains to be seen how ram impacts this one but bandwidth is down by almost 50%
Thanks D_A
I am losing it for sure.
I have just completely re-installed the latest version of Mint Cinnamon on my crawling box and thought I should take a look at running gpu on it.
intalled ia32-libs, installed 7.0.42 from BOINC using the routine outlined by D_A in post 105
Now I find this:
So I am asking you linux adepts.... What did I miss?Quote:
m@EX58-UD5pa ~/BOINC $ sudo /etc/init.d/boinc-client start
sudo: /etc/init.d/boinc-client: command not found
m@EX58-UD5pa ~/BOINC $
Hey Old Chap,
Did you install the 64 bit version of Linux 14 Mint Cinnamon? I tried that on my Q9550 rig and I couldn't get past the initial install menu before everything went black on my monitor.
Yep, the actual install seemed to go fine but I think I missed something for BOINC
Did you check the md5 for your download?
I actually bought a DVD copy from On-Link during a Black Friday type sale.
The only thing I can think of is that for some reason the basic install doesn't have the right drivers for my graphics card(Nvidia 9600GT).
That command will only work if you install from the repositories. If you install from a Berkley download then you either need to create that script yourself or add it to the startup programs using the session manager (or equivalent). It should be in the Settings menu under Session and Startup ... but I'm using XFCE not Cinnamon, so it might be worded slightly differently.Quote:
Originally Posted by OldChap
Ok, caught the first error myself (I remember now it is apt-get install ia32-libs) so now I have the client running and it sees the card
The next issue is that run_manager does not work. I have only ever used the manager before and until I fix my network with an upgrade it needs to stay that way
If you're using a Berkley install then you have to use the run_manager script in the BOINC directory. That script is auto-generated when you run the install script so if you've moved the BOINC directory from where it was originally created the run_manager script will be wrong. You can edit the path in it with whatever text editor you prefer and just save it.
If, on the other hand you're using the repository install then the run_manager script won't work. Period. You need to use boincmgr from the command line.
Please note that I'm assuming you have a graphical environment on this machine. If it's a server install then the manager won't work since there's no X server for it to run in. In that case you need to use terminal commands to administer BOINC.
Let me try to clarify:
I did the following after getting Mint 14 cinnamon running. and getting the fglrx driver from the repository...... It should look familiar :)
apt-get ia32-libs This was a mistake ...more on this later
replace the version numbers with whatever version you're after. If you're not sure go here and look: http://boinc.berkeley.edu/dl
This will download the file directly from Berkley to the current directory, which should be your home directory.
Next set the file as executable
and then run itCode:sudo chmod +x boinc_7.0.42_x86_64-pc-linux-gnu.sh
That will create a directory called BOINC in your home folder. Change into it.Code:./boinc_7.0.42_x86_64-pc-linux-gnu.sh
now we'll copy the relevant binariesCode:cd BOINC
and we restart the clientCode:sudo cp boinc* /usr/bin/
Client did not run...asked the question above then re-read what I had asked and realised that it should have been apt-get install ia32-libsCode:sudo /etc/init.d/boinc-client start
Now the client would run but I could not start manager so I asked again then went searching for others with this problem too.
Result! Found mention of this and looked to see if 2 files were loaded one called libwxgtk and another called libssl. libwxgtk was missing so I got it from the repository.
Now, from inside the BOINC folder hitting the start_manager worked as it should.
With manager working I have now attached to wcg and have downloaded GPU workunits...only running a single thread for tonight ...it is late enough. I will look at app-config tomorrow.
Thanks for your input D_A. I can always depend on you to make me think about what I am doing.
Let's start with a little something that you should have done BEFORE all that:
Those instructions are for upgrading a repository install to the current client from Berkley.Code:sudo apt-get install boinc*
:confused: < oh look they made a smiley just for me.
Incidentally, where does the app_config go in Linux?
Depends. In a pure Berkley install I'd expect to see it in the BOINC directory (/home/<username>/BOINC). In a repository install it'd go in the data directory, though I'm not sure what the correct path is for that off the top of my head.
good thread, ive decided a while ago that dedicated future builds will only run linux :)
thanks :up:
Not quite. There's not nearly enough silver hair on the smiley. ;)Quote:
Originally Posted by OldChap
So did you get it working or am I talking in short-hand again and just making things worse? (I've been known to do that so don't take it as sarcasm or anything)
Those are the three strands of hair remaining after any me vs Linux session. What makes it worse is the fact that once I see the light and start understanding things a little I really do like the way Linux works
Yep, I now have my crawling rig back up and running and it is currently crunching GPU with a single 5870 which I have overclocked using AMDOverdriveCtrl. I currently has 4 concurrent wu's using app_config.
The bothersome thing is that, were it not for a couple of issues I had with my previous install, re-reading this thread I get the impression I could have done everything I needed to do just by a command you described as "full upgrade" but hey this way I have a clean install to mess up again when trying stuff out. :up:
D_A, I may not say so every time but I do appreciate the time you take to explain stuff to me
Next is the update of my main PfSense router..... How are you with networking ;) :rofl:
I have a major project to do to decide if my major will be in network engineering (Cisco CCNA) or Unix systems administration. I've covered every required and optional subject in both majors aside from the project for which I have to decide which is my major and minor. (in other words, I know enough to know just how much I don't know ;) )Quote:
Originally Posted by OldChap
/var/lib/boinc-clientQuote:
Originally Posted by D_A
Thanks. I only have my Slackware machine running these days and it's a pure Berkley install so I couldn't check.
boinc detects ati when ran from my home dir but not when i use init.d/boinc-client start
not sure why...investigating later
fixed with this
http://boinc.berkeley.edu/dev/forum_thread.php?id=6307
What programs are reccomended to check cpu, gpu, and mem overclocks in a debian based os? Currently doing it pretty blindly and would like to push for more.
Have psensor for cpu load/temp (athlon II x3 400e)
AMDOverdriveCtrl (i386) for gpu temp/oc.
No way of seeing per core loading/throttling (not likely with the current temps) or gpu load.
Thanks
The whole point of these is that you set 'em and forget 'em. Worrying about anything but temps pretty much defeats the purpose.
For thermal monitoring I use lm-sensors. On a server install (no GUI) you check the readings from the command line with the command "sensors". On a desktop install you can install something like sensors-applet to monitor those and any other available temps.
For clock speeds there's something called (something like) CPU frequency scaling monitor, and the system monitor can track various CPU loads and graph them for you in the task bar. I'm not sure about GPU load monitoring, but sensors-applet can monitor GPU temperature if you configure lm-sensors (sudo sensors-detect) after installing the proprietary GPU drivers. It can also monitor per-core temperatures.
Temps aren't an issue, I'd just like to make sure the overclock is stable enough to do science without errors.
The only way to do that is to creep the clocks up until you start getting errors (check your results status on the My Grid page at WCG) and then back off, just like you'd have to in any other operating system. CPU load monitors won't provide any useful help in that situation.
I'm only mentioning this because I read about it somewhere but look for something called conky to monitor gpu.
So I installed Ubuntu 12.0.4 LTS on my Q9550 rig. I downloaded and ran the BOINC installerQuote:
Originally Posted by D_A
Typed sudo su
Typed apt-get update-Bunch of downloads and setting up lines appeared
Typed apt-get install boinc-Setting up lines appeared.
Then a line *Setting up scheduling for BOINC core client and children: appeared
followed by chrt: failed to set pid 15164's policy: No such process , which appeared 3 times
Then "Could not write to /proc/15164/oom_adj:"
followed by
"Setting up boinc-manager (7.0.27+dfsg-5ubuntu0.12.0.04.1)...."
and then it stalled right there
Did I miss something, some setting or command?
I've seen that before but don't remember what causes it. I'll have a dig and get back to you.
Try removing and reinstalling BOINC.
From the command line:
sudo apt-get remove boinc boinc-manager
sudo apt-get install boinc boinc-manager
The boinc client and manager used to be separate packages. Either they've been combined into one now, which could cause issues for those running server installs, or you added a wildcard at the end of one of those commands.
Thanks DA. I used those two commands and then went back and repeated the installation. This time it worked. Though now I can't seem to get GPU workunits, even though my graphics card is recognized. I'll have to figure that one out today. Thanks again!
You also need to go to your profiile on the WCG MyGrid page and enable GPU work. If you've done that, start a new thread and post whatever messages you're getting in your BOINC logs.
What repository are you talking about where you got the libwxgtk and libssl files?Quote:
Originally Posted by OldChap
Let's start with which version of BOINC you installed. Depending on the version you could well have the wrong version of libssl on your system.
Alternatively, open a terminal and enter:and thenCode:ldd boinc | grep ssl
and post the output.Code:ldd boincmgr | grep gtk
Installed Linux Mint 14 Cinnamon.
Installed AMD CCC 13
Installed BOINC 7.0.52
Installed the 32 bit drivers to allow 64 bit Linux to run it
After entering ldd boinc | grep ssl:
libssl.so.1.0.0 right arrow /lib/x86_64-linux-gnu/libssl.so.1.0.0 (0x00007f366647a000)
After entering ldd boincmgr | grep gtk:
libwx_gtk2u_html-2.8.so.0 right arrow not found
libwx_gtk2u_adv-2.8.so.0 right arrow not found
libwx_gtk2u_core-2.8.so.0 right arrow not found
libgtk-x11-2.0.so.0 right arrow /usr/lib/x86_64-linux-gnu/libgtk-x11-2.0.so.0 (0x00007fa8807fe000)
Ok, the wxgtk stuff isn't there so you'll have to install it.
Code:sudo apt-get install libwxgtk2.8-0
Thanks. That worked like a charm.Quote:
Originally Posted by D_A
*bump*
Didn't this used to be a sticky?
Here's some useful material regarding *nix and GPU crunching
http://boinc.berkeley.edu/wiki/GPU_computing
Latest ATi Linux drivers:
http://support.amd.com/us/gpudownloa...eon_linux.aspx
Latest nVidia Linux drivers:
http://www.nvidia.com/object/unix.html
Many thanks to OC and DA. I actually got boinc running in Mint. Can you guys tell me where the cc_config and app_config files go? Kinda struck out on that front.
EDIT: Found where the cc_config goes. DUH!
/var/lib/boinc-client/projects/www.worldcommunitygrid.com/app_config.cfgQuote:
Originally Posted by nanoprobe
/var/lib/boinc-client/cc_config.xml
if i understand this correctly you no longer need to have graphical desktop open to crunch with GPUsQuote:
AMD Catalyst™ 13.4 Proprietary Linux Graphics Driver Release Notes
New Features:
The following section provides a summary of new features in this driver version.
RHEL 6.4 Production Support
OpenCL Console Mode Support
Kernel 3.7 and 3.8 Support
sadly that driver doesn't work for me cause 7870 XT is not supported (even tho they're supported in 13.3 beta :shrug:)
I have not had a chance to try this but when searching I found the following info on ghacks.net http://www.ghacks.net/2010/03/11/wha...on-of-apt-get/
I found it interesting and thought to just paste the necessary here with thanks from me to the author.
Quote:
First I want to compare the two apt-get options: upgrade and dist-upgrade. The difference between these two is very simple. The upgrade option is used only to install all of the newest versions of the packages already installed on your machine. This is what happens when you see the upgrade upgrade packages through the Update Manager (You know, when you are made aware of notification in your panel and then Update Manager opens to show you which updates are available for your system.)
The dist-upgrade option is quite different from upgrade. But just what does dist-upgrade do? Simple. The dist-upgrade option not only will upgrade all of the currently installed packages on your system it will also handle the dependency changes with new versions of packages. In other words, dist-upgrade will remove obsolete packages from your system, whereas upgrade will not. This function is out of necessity for upgrading from one distribution release to another.
Quote:
But doesn't it upgrade my distro?
Not necessarily. Although, by itself, dist-upgrade, will prepare your system for a distribution upgrade, the only way it will actually perform the upgrade to a new release is if you have changed your /etc/apt/sources.listfile to reflect the change. In other words, you have to add the repositories for the new distribution in the sources.list file before this can happen.
Quote:
Well that sounds simple. Yes it is. But it is not the recommended plan of attack for upgrading to a new distribution. Although it will work, there is a better way...by command. The command do-release-upgrade was created specifically for upgrading the operating system to the latest release via the command line. Of course, before you issue this command you will want to check to see if an upgrade is available with the command sudo do-release-upgrade -d. When you do this you you be informed if a new release is available. If there is a new release available issue the command sudo do-release-upgrade and the upgrade process will begin. Of course this is a rather lengthy process and there are risks involved.You could lose packages you have manually installed (that are not supported by the new release) or (worst case scenario) your system could become unstable.
Quote:
Distribution upgrades are always tricky. I have had them go exceedingly well and I have had them go very awry. It's a gamble. But now you know exactly how to manage the task from the command line. You are now, officially, dangerous. ;)
Clicked on run manager after updating an install on an old disk, which I had not tried to run before the update.
It opens fine but shows as Disconnected
What to do to get it to connect?
EDIT: Never mind it is a permissions issue. That disk must have been set up to run BOINC as root
Hey D_A and Co.
How to properly use chown or chmod to give myself permissions to copy a BOINC folder to a stick with a view to retaining this particular machine install after I try a later version of Mint with a view to resolving a Kernel panic issue I have just now
As above it seems to be owned by root just now and I need to undo that.... not just to copy the folder but to run it later.
chown -r oldchap /var/lib/boinc-client
That is not where mine ended up on any of my installs but I get the picture
checked and found:
m@M ~ $ ulimit -n
1024
want to increase this to lots more (say 10 times bigger)
Would one of you adepts save me the pain of finding the answer on how to do this ;)
The aim of the game here is to increase the number of concurrent network connections which I think is covered by this. I expected to find 1024 somewhere in etc/security/limits.conf. but when I didn't see it I got to thinking.... ask one of the guys :)
Thanks
Edit: Incidentally this is so that I can efficiently dual purpose a crunching rig
ulimit -n displays the maximum number of open file descriptors.
I don't think this is the command you're looking for.
Here's an article about it: http://www.itworld.com/operating-sys...-limits-ulimit
Yeah, this has not been an easy find. I decided to trial changes to the limits.conf file just to see what happens. I am basing this on comments HERE
Do you think a re- boot is in order? Up until now all I did was log out and back in then re-start wcg etc. Testing cannot commence until around 2 am here when my internet connection gets quieter.
Re-booting anyway then we will see if I got this right
OK added this:
root hard nofile 10240
Myname hard nofile 10240
myname soft nofile 10240
after a re-boot the new limits show up in ulimit -n and ulimit -a. time will tell if this is enough.
I think the answer from the thread you quoted has what you need. Something to do with the port range and reuse timesQuote:
Originally Posted by OldChap
Quote:
Maximum number of connections are impacted by certain limits on both client & server sides, albeit a little differently.
On the client side: Increase the ephermal port range, and decrease the fin_timeout
To find out the default values:
sysctl net.ipv4.ip_local_port_range
sysctl net.ipv4.tcp_fin_timeout
The ephermal port range defines the maximum number of outbound sockets a host can create from a particular I.P. address. The fin_timeout defines the minimum time these sockets will stay in TIME_WAIT state (unusable after being used once). Usual system defaults are:
net.ipv4.ip_local_port_range = 32768 61000
net.ipv4.tcp_fin_timeout = 60
This basically means your system cannot guarantee more than (61000 - 32768) / 60 = 470 sockets at any given time. If you are not happy with that, you could begin with increasing the port_range. Setting the range to 15000 61000 is pretty common these days. You could further increase the availability by decreasing the fin_timeout. Suppose you do both, you should see over 1500 outbound connections, more readily.
Added this in my edit:
The above should not be interpreted as the factors impacting system capability for making outbound connections / second. But rather these factors affect system's ability to handle concurrent connections in a sustainable manner for large periods of activity.
Default Sysctl values on a typical linux box for tcp_tw_recycle & tcp_tw_reuse would be
net.ipv4.tcp_tw_recycle = 0
net.ipv4.tcp_tw_reuse = 0
These do not allow a connection in wait state after use, and force them to last the complete time_wait cycle. I recommend setting them to:
net.ipv4.tcp_tw_recycle = 1
net.ipv4.tcp_tw_reuse = 1
This allows fast cycling of sockets in time_wait state and re-using them. But before you do this change make sure that this does not conflict with the protocols that you would use for the application that needs these sockets.
On the Server Side: The net.core.somaxconn value has an important role. It limits the maximum number of requests queued to a listen socket. If you are sure of your server application's capability, bump it up from default 128 to something like 128 to 1024. Now you can take advantage of this increase by modifying the listen backlog variable in your application's listen call, to an equal or higher integer.
txqueuelen parameter of your ethernet cards also have a role to play. Default values are 1000, so bump them up to 5000 or even more if your system can handle it.
Similarly bump up the values for net.core.netdev_max_backlog and net.ipv4.tcp_max_syn_backlog. Their default values are 1000 and 1024 respectively.
Now remember to start both your client and server side applications by increasing the FD ulimts, in the shell.
1 step at a time, see my post above. Maybe that way I can run down the real reason for a lack of scalability when adding nodes
Edit: Made no difference.
Found some info and downloaded ethtool. Increased tx and rx buffers to max but with 100Mb connection currently @ 8Mb :( I must wait.
If this also doesn't help then port range and re-use is next
Memory failure..... Not the machine... Mine :shakes:
What was the fix for "apt-get install ia32-libs" when that stopped working??
try sudo apt-get install lib32z1
http://www.xtremesystems.org/forums/...=1#post5218990
Would someone please show how to set up remote_hosts.cfg and gui_rpc_auth.cfg from cli linux
use nano should be self explanatory
Code:sudo nano remote_hosts.cfg
Ha !! Superb !!! :yepp:
As Guides go that one is up there :clap: :clap: :clap:
Thanks Mick
No problem, happy to help.
would there be use for a headless USB bootable BOINC distro that runs in RAM and syncs settings upon startup/exit?
There might. I'm happy with the way my headless machines are set up but I'm sure there are others who would run such a thing.
Tell us more... particularly size, hardware requirements etcQuote:
Originally Posted by Evantaur
it's just something i've been thinking a long time, not sure about the size but considering it would be absolute minimal installation it wouldn't be huge. would need to have enough ram to hold those WUs thoughQuote:
Originally Posted by OldChap
I played a bit with Fatdog 700 but it was still in alpha and I am unclear how one saves the whole setup once everything is going the way you want it. Runs well though and is very small and naturally runs 100% in memory
I figure that once I get it figured out the stick could be imaged for all to use. am waiting until it gets out of beta first though.
i'm thinking about using bittorrent sync so all boinc files are stored online, but need to be sure it's 100% secure first (maybe use extra encryption too just to be sure)