HELP!!!!!

I've installed Mandriva-2007-FREE-x86-64 on my AMD64 2800 and got Boinc to work just fine. When I installed the SAME cds on both my X2 4800 and AMD64 3700, I can't get boinc to work.

Here's what I've done:

OK, I deleted the BOINC folder.
Opened Console, entered:
[astro@localhost Desktop]$ sh boinc_5.6.4_i686-pc-linux-gnu.sh
use /home/astro/Desktop/BOINC/run_manager to start BOINC

then

[astro@localhost Desktop]$ cd BOINC
[astro@localhost BOINC]$ ./run_manager
./boincmgr: error while loading shared libraries: libXxf86vm.so.1: cannot open s hared object file: No such file or directory

then using ls command, I get:

[astro@localhost BOINC]$ ls -l /usr/lib/libXxf86*
ls: /usr/lib/libXxf86*: No such file or directory

Maybe I need root??

[astro@localhost BOINC]$ su
Password:
[root@localhost BOINC]# ls -l /usr/lib/libXxf86*
ls: /usr/lib/libXxf86*: No such file or directory
[root@localhost BOINC]# cd
[root@localhost ~]# ls -l /usr/lib/libXxf86*
ls: /usr/lib/libXxf86*: No such file or directory

I've googled and one site tells me the missing file is part of lib64xorg-x11...... So I used Easy URPMI and downloaded anything with an xorg or x11 name in it. Installed every lib I could find, and still no soap.

Interesting thing is, is that that file DOESN'T exist on my 2800 either, but it's working.

any ideas?

TIA

any ideas?

When I couldn't get 64 bit Ubuntu 6.10 to BOINC, I was able to solve that by

sudo apt-get install ia32-libs

which installed 32 bit libraries. Maybe you had some occasion to install 32 bit libraries on the first installation for something else, but not for these two new ones?

Pam

Doing a quick search on "libXxf86vm.so.1" gives:

libxxf86vm1-1.0.1-2mdv2007.0.i586.html Development files for libxxf86vm Mandriva Linux Official packages libxxf86vm1-1.0.1-2mdv2007.0.i586.rpm

XFree86 Video Mode Extension Library

Provides

* libxxf86vm1
* libxxf86vm
* libXxf86vm.so.1

Requires

* libX11.so.6
* libXext.so.6
* libc.so.6

Files

/usr/lib/libXxf86vm.so.1
/usr/lib/libXxf86vm.so.1.0.0


... which is all 32-bits.


Have you stumbled onto some problem for the type of display card in those systems?

Boinc itself should still work ok from the command-line with text only. A 'quick fix' that might be worth a try is to see if you can reinstall Xorg without any acceleration or any vendor-specific drivers.


Sorry, just wild guesses 'cos I'm still languishing in the world of 32-bit hardware :-(


HTH (a little at least)

Cheers,
Martin

---

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)

Hi Pam and Martin,

The X2 has S3 onboard graphics (as does the 2800), the 3700 has the Geforce 6200 pcie-x16 graphics card.

I've been reading as well, someone recommended getting ahold of the libXxf86vm file, unpacking it and copying the files to the /lib32/ location??

I'm not even sure where that is, let alone where to find the repository.

I tried Pam's suggestion, but it spat back reasons it wouldn't do it, bad command and so forth.

I think you're both right though, it has something to do with the lack of 32bit libs needed by boincmgr, but lacking on x86-64 mandriva, for the X window.

I wonder if I should boot this issue to the devs and see if Walt Gribben (linux guru) has anything to say???

It's also been suggested to me that I might use Boincview to run the CC on these computers. This defeats the purpose of the install so I'm not going to try that. Martin, you suggest the command line running of boinc. Now that's another new marvelous journey I'd have to undertake, since I don't know any of those commnands. LOL

Oh....I'm off to sit an ponder.

thanks

tony

Hey, I found these:

file:///usr/lib64/libXxf86dga.so.1
file:///usr/lib64/libXxf86dga.so.1.0.0
file:///usr/lib64/libXxf86misc.so.1
file:///usr/lib64/libXxf86misc.so.1.1.0
file:///usr/lib64/libXxf86vm.la
file:///usr/lib64/libXxf86vm.so
file:///usr/lib64/libXxf86vm.so.1
file:///usr/lib64/libXxf86vm.so.1.0.0

on the 2800. I'm off to reboot into linux on one of the others and see if it's the same.

There is NO libXxf86vm.so.1 in the /lib directory. I assume this to be the 32b library????

Ah HA,

The below files DO NOT exist on the X2 /lib64

How do I get them??????

Can I just copy them or do they need installing (insert some other linux term here)

Ah HA,

The below files DO NOT exist on the X2 /lib64

How do I get them??????

Can I just copy them or do they need installing (insert some other linux term here)

OK, very good detective work thus far...

Interestingly, I can only see the 32-bit versions in the Mandriva mirror repositories.

Any clues for what you have done differently for the installs across the machines?

You could just copy the files across into the right place provided you are running the same kernel on the respective machines, and provided that there are no other dependancies. Letting "urpmi" or the system install tools do their job is much safer unless you know what is happening.


The commandline:

uname -a

will describe what kernel/system you are running.


If you're still stuck, I'll give Mandriva 2007 - 64bit a try tomorrow and see what I see with it.

You're welcome to the command-line invocation if you wish.

However, if it's only the display that is the problem, then you could instead ssh in from your working Linux box and start up the boinc manager from there using your working system display. Hence:

On your working system, login to your new box:

ssh username@ip.address.of.box

That logs you in, and then:

cd to/your/boinc/directory
./run_manager

That will then run the boinc manager on the target machine but produce the display on your working system at which you are sat. Could work depending on how/where those missing libraries are called.

You may need to install (on your target machine!) "ssh-server" and set for the service to start up (system control centre -> services, one tick box for "sshd"). Also, the security level needs to be at level 3 or lower, or you need to edit "/etc/hosts.allow". You will need ssh-client on the machine at which you are typing (there's no 'service tick box' to start it being as it is a client).


Good luck,

Cheers,
Martin

---

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)

I'm not sure what happened. Earlier I was picking an installing things and I thought it was done. I switched from win back to lin and started the MLCC and sheeebang, It started downloading 167 files/packages before I even started searching for xorg and xll. I'm currently at 96 of 167. I'll let this play out, I see lots of lib64 stuff in it. I keep doing a 'view' 'reload' to see if/when they show up, so I can attempt to do the same to the other machine. I'm in a holding pattern.

Since the usr/lib64/ of the 2800 has them, then I'd assume (correctly or incorrectly) that they're 64 files and they're sufficient to allow me to get a snidge further on the boinc running problem.

Thanks for all you assistance, It will be some time before I know, but will post either way.

tony

I'll do the uname -a after it's finished to see.

OK, I fought the good fight, but I lost. I now have those files in usr/lib64/ but it STILL won't work. I must have installed something which placed them there. I'd been installing all last nite and a good chunk of today.

as a side note,,,uname -a lists 2.6.17-5mdv for all puters.

I have learned how to find the floppy drive though, that's a bonus.

I'm going to try to figure out how to uninstall all this so I can install Suse 10.1 x86-64. Off we go again.

Lovely, Suse 10.1 found no hard disks. Must be to old to use Sata II.

Off surfing to find and download something else.

OK, so something must be different between your working Mandriva system and those other two that don't let the Boinc Manager start up...

(Sorry, on the command line...) You can use on your working system:

rpm -qa >full_installed_list_good.txt

and say:

rpm -qa >full_installed_list_bad.txt

for one of your other systems. That will output a text list of all the installed packages into the respective text files. (The ">" is a 'redirection' operator that sends output to the named file.)

Then copy the "good" list over to your bad system via some means. If they are networked and you have sshd running on the bad machine then you can simply use:

scp full_installed_list_good.txt your_username@target_ip_or_machine_name:.

The ":." there says dump the file in your user's home directory.


Then, running a comparison will show up what's different.

eg:

diff full_installed_list_good.txt full_installed_list_bad.txt

to list all the differences.


And your SATA disks should appear as "/dev/sdXXX" devices and should be automatically found.

Good luck,
Martin

---

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)

Update:

My troubles installing 5.6.4 and above on Mandriva free 2007 x86-64 have been resolved by installing 5.4.11_i686-pc-linux-gnu.sh. Not sure exactly why it happens, might be the change from GTK1 to GTK2 in version 5.5.x, or maybe something else. I found this out when another user reported he could use 5.4.11, but when he upgraded to 5.8.11 he was missing a "shared lib" (not the same one as I was missing). I tried 5.4.11 and it worked on all computers.

I reported this to Boinc Alpha.

Well, it's the weekend again, and a whole new issue has arisen. For one reason or another Mandrava free 2007 x86-64 wouldn't install on my last machine to go "dual boot". I suspect it's due to built in ON board HD video.

So, I broke out Suse 10.1 x86-64 and it installed straight away. 5.8.15 installed, and even used Augustines 5.8.11 64 boinc with it. NO problems. I attached, got and returned work. I thought I was finished.

THEN after a reboot, I found I no longer had a network. during configuration the default was IFUP for network config. If I tried to ping anything on my network it returned something to the effect "network unreachable". It would ping 127.0.0.1 though. Nothing else was reacheable. I disabled the firewall with the same outcome. I tried to reconfigure with "network manager" instead of IFUP, but go nowhere.

I ended up reinstalling linux, but this time picked, network manager instead of IFUP during the install, and disabled the firewall. It too worked great. I installed boinc got work let it run all nite getting and returning work, then I shutdown to go back to windows for the workday. After work, I restarted into linux and even reported a wu. I then took a nap, and when I awoke, the screen was filled with "uploading" status on a dozen wus.

When I hover the mouse over the "wired network" line of the network manager it reports an IP of 169.254.88.44 (last two digits changed on purpose). Nothing I did or more to the point know how to do would make it get the right IP for my WAP/4 port switch. It should be like 192.168.2.? with 192.168.2.1 as the gateway.

Is there something like IRrenew, IPconfig, or something which might make it link to the proper hub? I have a NAT enabled WAP/4p switch attached to a speedstream dsl router. This computer is cabled to the switch from the onboard lan.

TIA

Tony,

I'd recommend just setting your network to static instead of DHCP, as it seems you're configured currently.

The reason I think this will work better - the IP you said you got is a so-called APIPA IP (169.254.0.0, Automatic Private IP Addressing)- one that your host assigns itself when it cannot grab one via DHCP.

This IP is in a guaranteed private range, usually with a subnet mask of 255.255.0.0.

Other hosts with APIPA addresses will be able to ping you and transfer data via TCP, provided you know their address. However, you will not be able to go anywhere else.

So by setting your IP statically, you could circumvent this issue.

Still, I thought only MS Windows hosts did this, but it seems more recent Linux versions have added this behaviour as well. It works well for small workgroups without much hassle to setup (just connect cables, go, no need to even know what TCP is, much less how to configure it).

HTH,
Simon.

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information

Yet more fun adventures!

Shame I didn't document what (one) library I loaded up that got Boinc running on my Mandriva x84_64...

...Is there something like IRrenew, IPconfig, or something which might make it link to the proper hub? I have a NAT enabled WAP/4p switch attached to a speedstream dsl router. This computer is cabled to the switch from the onboard lan.

Sounds like you might have networking "zeroconf" getting confused...

Can you set to get your IP by dhcp, and turn off hotplug if that's an option?


(Meanwhile, I've just recovered a system from a failed HDD... Quite worrying to only have just one copy of some of my data for a few days until oncemore getting a backup going!)

Happy crunchin',
Martin

---

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)

I'm not a network guru by any measure. I have to the best of my ability to determine, set the network to configure IP and hostnames from DHCP, yet it won't see it to get the proper IP. It has the same mac addy as is used by windows, so I assume it's attempting to configure the right network card (built in to asus board). I've tried every config option available in the configuration menu I could think of. I even went back from "network manager" to IFUP, and no go. The "hotplug" option has never been enabled, according to the menu.

If I go to "static" addy, I think that I'd have to set it up in the wap and each of the 7 attached computers. I've tried to do (static) just the one, but that didn't work either (IP 192.168.2.5, Subnet 255.255.255.0). I can't ping 191.168.2.1 (gateway). No puters on the network can ping this host either.

It's kinda of strange that it DID work just fine, until I took a nap, and now nothing. reboots don't help either.

I'm not a network guru by any measure. I have to the best of my ability to determine, set the network to configure IP and hostnames from DHCP, yet it won't see it to get the proper IP. It has the same mac addy as is used by windows, so I assume it's attempting to configure the right network card (built in to asus board). I've tried every config option available in the configuration menu I could think of. I even went back from "network manager" to IFUP, and no go. The "hotplug" option has never been enabled, according to the menu.

If I go to "static" addy, I think that I'd have to set it up in the wap and each of the 7 attached computers. I've tried to do (static) just the one, but that didn't work either (IP 192.168.2.5, Subnet 255.255.255.0). I can't ping 191.168.2.1 (gateway). No puters on the network can ping this host either.

It's kinda of strange that it DID work just fine, until I took a nap, and now nothing. reboots don't help either.

You shouldn't need to mess around with the rest of the network - leave well alone, I say.

Have a look at the DHCP settings on the router. It'll usually show you the first address it's going to allocate, and the maxmimum number: so that leaves it up to you to work out the final one. E.G. starting address 192.168.2.10, maximum 50 addresses, means that you should avoid allocating anything up to 192.168.2.60 manually.

Then pick a number in the same subnet, but outside the DHCP 'scope', and that should co-exist happily with the other DHCP-assigned addresses for the other machines. You could even leave the Windows installation for that machine on DHCP client.

Just remember that if you set the linux IP to static, you also have to provide the gateway and DNS settings yourself - there's more to DHCP than just assigning the basic IP address.

The lan settings on the belkin wap/router showed:
IP addy 192.168.2.1
subnet 255.255.255.0

DHCP range WAS 192.168.2.2 - 192.168.2.100
lease time "forever"
Local Domain name "belkin" -optional

I changed the range to 192.168.2.2 - 192.168.2.14

I'd set the IP to 192.168.2.15 and subnet to 255.255.255.0 and it didn't work, I'm unsure about setting the DNS.

I'm off to reboot back into linux.

I opened YAST2. logged in, clicked on "network card". The screen offers two choices, User Controlled with Network manager, and Traditional Method with ifup.

I pick user controlled, and click "next".

It lists ASUSTeK MCP51 Ethernet controller and IP addy DHCP.
there are no other options at this point but to "add, edit, remove"
I click next.
I see network manager icon running then it stops and displays:

IP 169.254.44.88
hardware addy matchest that of windows
subnet 255.255.0.0
Broadcast 169.254.255.255
active: YES
Carrier detect: supported
Bandwidth 100Mb/s

If I click on KDE menu, System, Monitor, then Network interfaces, it only shows:

name lo
IP 127.0.0.1
network mask 255.0.0.0
type loopback