VMWare problems/help thread

**p2501** · 11-14-2008, 11:34 AM

Ooookay, Marvin said it's better to take this into a new thread and I do think he's right. Well, where should I start. I followed The_Barnacle's thread while setting it up (thanks alot for that, very helpful

). I reconfigured it for my username and for 36362, for 2 CPUs and USB option. On the second try I put in 1024MB of memory to indicate (first try was without any indication), but no help. One of the VM's is always (well, not always, it ran a Wu fine through the night, but today I had to reboot and right after that it started to behave like this) showing this again.

My guess is that it's got to do with WU sizes, bigger WU needing more space. But no matter what memory I set it up with the upper limit shown in the pictures doesn't change, so I think it's got to do with the VM itself. Is this upper limit the limit of the virtual USB device (i.e. RAM disk)?

Did anyone configure it without this USB option? I tried around, even deleted everything and started anew, but noe help.

**DeanZ** · 11-14-2008, 11:50 AM

you have to reset the consoled to get the new memory option then it will stay

but now also seem to have

[12:32:38] Error: Could not write local file. Exiting.
[12:32:38] - Shutting down core

lol this is a lot of hassle but we can get through this

any ideas?

**~~Marvin_The_Martian~~** · 11-14-2008, 12:04 PM

I can't help I don't use the usb option and therefore don't have any experience ( as I can recall atleast ). I just set up the c2d with the vmware appliance and the free vm player, it's running a 2669 but I got problems monitoring it in fahmon, and since I don't have a network connection between that pc and this one I can't monitor it with fahwatch. It's at 6% though and using all ram I got on that machine for a single instance ( eg xp 32bit pro, 1gb ram all allocated ( +-500 if that going to vmplayer ) from 1gb total ).

Btw now I also know what you mean with having no option to adjust the ram for the vm, only vm workstation gives me that option. Think maybe you need to change the config to not allow big wu's and try to get some less demanding one's in ram that way ( though I think running without -advmethods would be the best option )

**p2501** · 11-14-2008, 12:33 PM

Originally Posted by DeanZ

you have to reset the consoled to get the new memory option then it will stay

I did that every time, without any impact on the outcome.

Originally Posted by Marvin_The_Martian

I can't help I don't use the usb option and therefore don't have any experience ( as I can recall atleast ). I just set up the c2d with the vmware appliance and the free vm player, it's running a 2669 but I got problems monitoring it in fahmon, and since I don't have a network connection between that pc and this one I can't monitor it with fahwatch. It's at 6% though and using all ram I got on that machine for a single instance ( eg xp 32bit pro, 1gb ram all allocated ( +-500 if that going to vmplayer ) from 1gb total ).

Btw now I also know what you mean with having no option to adjust the ram for the vm, only vm workstation gives me that option. Think maybe you need to change the config to not allow big wu's and try to get some less demanding one's in ram that way ( though I think running without -advmethods would be the best option )

Well there is an option to change RAM allocation, I currently set it to 2GB, let's see if that helps at all. If not I'll try it first without USB, then without adv.

**~~Marvin_The_Martian~~** · 11-14-2008, 12:45 PM

Where is that option? I select open the appliance, it just boots, the menu's in that screen are 'vmware player' and 'devices' but none of those let you select the memory allocation. The only memory allocating you can adjust is the one from the fah clients with the reconfigure option?

Ow nm, found it, you can edit the folding.vmx file for that

**p2501** · 11-14-2008, 01:10 PM

Originally Posted by Marvin_The_Martian

Where is that option? I select open the appliance, it just boots, the menu's in that screen are 'vmware player' and 'devices' but none of those let you select the memory allocation. The only memory allocating you can adjust is the one from the fah clients with the reconfigure option?

Ow nm, found it, you can edit the folding.vmx file for that

Either that , or...

Running without the USB option, and suddenly Fahmon works for the VM's? WTF?

**~~Marvin_The_Martian~~** · 11-14-2008, 01:15 PM

Troubleshoot.. change memory allocation.

yeah that's a logic place for the option

Tell me how you configured fahmon? I don't see the clients in my network, the clients run with nat just as the vm's on my quad, but on the xp rig I just can't see anything. Not even my own pc though.. must be setup issue within the xp install itself. though network sharing is on. I monitor with http:/x.x.x.x/ which works but fahmon can't take that as a client location

Ps are you sure you need that much ram?

**Shadowtester** · 11-14-2008, 01:24 PM

Sounds like your problem might be with the nofred folding disk setup which I never used I just used vmware player after I made a custom appliance using the website I linked to in the other thread to create the appliance. I then just installed Ubuntu 8.04 server edition from the downloaded ISO file and used finstall.

So basic procedure

1 Download free VMware player http://www.vmware.com/download/player/ and install
2 Setup custom VMware appliance to meet your system resources and your requirements you can use this website to do that http://www.easyvmx.com/easyvmx.shtml
3 Download Ubuntu 8.04 LTS 64 bit Server edition from this link http://www.ubuntu.com/GetUbuntu/download
4 Start the VMware player and setup your appliance you can point the cd drive to the Ubuntu iso so you do not have to burn a disk.
5 Turn On the VMware appliance and install Ubuntu Server edition
6 once you finish the install setup a user and password then login and do the following

Code:

 
apt-get update
apt-get upgrade
apt-get install ia32-libs
wget http://ra.vendomar.ee/%7Eivo/finstall
chmod +x finstall
./finstall -smp

Off the top of my head that should get you going I would pre-allocate the disk space for the VMware client with at least 2gb of space and I would set the memory to about 1.5gb so the client will have at least 1 gig of ram to use. To monitor the client you could install and setup samba but I an not up on that since I do not use samba I prefer NFS since I use a all Linux network.

**~~Marvin_The_Martian~~** · 11-14-2008, 01:35 PM

Now I remember having that 'uhuh' moment when thinking about your 1.5gb recommodation. It's because of the ubunto os having that much larger overhead then bare notfred's. With that I allocate 768 to the entire vm and it gives me more from that for the folding client ( or does that server edition have a bare shell when you first configure it? ).

Btw, 2 eue's just yet on the vm's same one each time, not enough memory. Going to try 1gb now, don't want to loose to much. I set the vm on the xp box up with 768 also but no advmethods ( it's still doing an a2 wu and I hope it doesn't chocke! ).

Odd that the wu's got completed without that memory problem before now, same project range

**p2501** · 11-14-2008, 01:40 PM

Originally Posted by Marvin_The_Martian

Troubleshoot.. change memory allocation.

yeah that's a logic place for the option

Tell me how you configured fahmon? I don't see the clients in my network, the clients run with nat just as the vm's on my quad, but on the xp rig I just can't see anything. Not even my own pc though.. must be setup issue within the xp install itself. though network sharing is on. I monitor with http:/x.x.x.x/ which works but fahmon can't take that as a client location

Ps are you sure you need that much ram?

Oookay, well it shows the VM's now again as hung, when they are not. I put in the name of the VM (\\fold-9e66\C\etc\folding\1\ for me, just change the bold to your IP or VM name), but it also worked for me with IP adresses. I don't know if I'll need that much RAM, we'll see. As Shadowtester indicated bigger WU's might need it.

Originally Posted by Shadowtester

Sounds like your problem might be with the nofred folding disk setup which I never used I just used vmware player after I made a custom appliance using the website I linked to in the other thread to create the appliance. I then just installed Ubuntu 8.04 server edition from the downloaded ISO file and used finstall.

So basic procedure

1 Download free VMware player http://www.vmware.com/download/player/ and install
2 Setup custom VMware appliance to meet your system resources and your requirements you can use this website to do that http://www.easyvmx.com/easyvmx.shtml
3 Download Ubuntu 8.04 LTS 64 bit Server edition from this link http://www.ubuntu.com/GetUbuntu/download
4 Start the VMware player and setup your appliance you can point the cd drive to the Ubuntu iso so you do not have to burn a disk.
5 Turn On the VMware appliance and install Ubuntu Server edition
6 once you finish the install setup a user and password then login and do the following

Code:

 
apt-get update
apt-get upgrade
apt-get install ia32libs
wget http://ra.vendomar.ee/%7Eivo/finstall
chmod +x finstall
./finstall -smp

Off the top of my head that should get you going I would pre-allocate the disk space for the VMware client with at least 2gb of space and I would set the memory to about 1.5gb so the client will have at least 1 gig of ram to use. To monitor the client you could install and setup samba but I an not up on that since I do not use samba I prefer NFS since I use a all Linux network.

Yup, on all Linux private nets I'm always running NFS too (Vista brings that too btw

). What you're proposing here could be what I'll do next, if that try with USB disabled doesn't work out. I think so too that it looks like an error in the VM itself, because what it's complaining about is the size of the RAMDisk, not the memory, I guess. Let's first see if it survives the night.

I'd get my hand dirty with Arch, I had very good experiences with it and their crowd. Especially the ppc people, veeery nice they are. Sadly Gentoo seems to be dying, that was fun too.

**Shadowtester** · 11-14-2008, 01:40 PM

The server edition is just a bare command shell version of Ubuntu no gui desktop and it has been said that some of the larger smp wu's can require up to 1gb of ram to fold so you need to make sure the VMware client has enough ram to run the OS and leave 1 gb free for the FAH smp client which 1.5gb give plenty. Now if you don't use - advmethods and set the size of wu to small or normal you can get by with less ram but if you have it set to big you should have 1 gig available for the client to be safe.

**Shadowtester** · 11-14-2008, 01:48 PM

Originally Posted by p2501

Oookay, well it shows the VM's now again as hung, when they are not. I put in the name of the VM (\\fold-9e66\C\etc\folding\1\ for me, just change the bold to your IP or VM name), but it also worked for me with IP adresses. I don't know if I'll need that much RAM, we'll see. As Shadowtester indicated bigger WU's might need it.

Yup, on all Linux private nets I'm always running NFS too (Vista brings that too btw

). What you're proposing here could be what I'll do next, if that try with USB disabled doesn't work out. I think so too that it looks like an error in the VM itself, because what it's complaining about is the size of the RAMDisk, not the memory, I guess. Let's first see if it survives the night.

I'd get my hand dirty with Arch, I had very good experiences with it and their crowd. Especially the ppc people, veeery nice they are. Sadly Gentoo seems to be dying, that was fun too.

I have played with Arch as well and had no real problems with it just I am more comfortable with debian base Linux since thats what I know and Ubuntu is a very easy turn key distribution to work with and is kept very up to date its a very active project.

**~~Marvin_The_Martian~~** · 11-14-2008, 01:48 PM

Yup makes allot of sence, thanks for heads up. Think none of us really followed the linux threads. The 1gb per smp might also explain why jettasin has his probems with two windows smp clients under vista with 2gb.

I don't use adv methods but still get the a2 core, hope it won't choke during the night on the xp rig. Changing to 1gb for each vm on my vista right now but I won't be happy with setting it higher I need some ram on my daily os as well.

**Shadowtester** · 11-14-2008, 02:00 PM

Originally Posted by Marvin_The_Martian

Yup makes allot of sence, thanks for heads up. Think none of us really followed the linux threads. The 1gb per smp might also explain why jettasin has his probems with two windows smp clients under vista with 2gb.

I don't use adv methods but still get the a2 core, hope it won't choke during the night on the xp rig. Changing to 1gb for each vm on my vista right now but I won't be happy with setting it higher I need some ram on my daily os as well.

One thing I have noticed with the Linux SMP clients when I use -advmethods flag I have a lot less problems with losing wu's when I try to stop the client with "Ctrl c" or using the scripts from finstall especially on the a1 core projects. That problem was much worse on the earlier clients and honestly I have had no problems with using the -advmethods flag.

**~~Marvin_The_Martian~~** · 11-14-2008, 02:05 PM

Yeah I got no problems with wu's in the vm's as I just suspend the vm never power down

I'm more worried about the memory usage. Ddr2 is dirt cheap now, think I'll order some extra for the missus rig right now.. checking my fav online etailer

**Shadowtester** · 11-14-2008, 02:15 PM

I am going to have to work part of the day Saturday but if I have time later in the weekend I will see if I can setup and install a fresh VMware client and document all the steps and setup the smp client to help get people going. The only thing is I use Linux as my host OS so some of the window screen shots might look a little different in Windows. The basic procedure should be the same I will have to read up on samba installation and setup so that I can include that as well for client monitoring. No promise that I will get enough time to do it this weekend but I will try.

**~~Marvin_The_Martian~~** · 11-14-2008, 02:19 PM

That would be great

Thanks for the help Shadowtester, your presence is a big asset to our forums and I can tell you it's appriciated

**p2501** · 11-14-2008, 10:35 PM

That is a very generous offer, Shadowtester! I think it would be much appreciated!

Both VM's are still running, again they survived the night...

**~~Marvin_The_Martian~~** · 11-15-2008, 01:07 AM

Btw if you know how to monitor an appliance folder with fahmon, please tell me. The frames get written to the log about every 90minutes so tracking is impossible. Unitinfo the same, queue.dat the same etc. I can only check the hanglog, and cpu usage to see if it's running and though it seems to be folding fine I wouldn't mind being able to monitor it more closely

**Shadowtester** · 11-15-2008, 01:35 AM

Originally Posted by Marvin_The_Martian

Btw if you know how to monitor an appliance folder with fahmon, please tell me. The frames get written to the log about every 90minutes so tracking is impossible. Unitinfo the same, queue.dat the same etc. I can only check the hanglog, and cpu usage to see if it's running and though it seems to be folding fine I wouldn't mind being able to monitor it more closely

You need to setup either Samba or NFS file sharing and then share the FAH folder over the network. After you do that you should be able to mount the share on your network and then add the client to your fahmon. Thats how I always did it I used NFS setup the shares and mounted them on my Host OS so when I ran fahmon I could monitor the clients like any other client on the network/local machine. Another option but much more complicated which I use to monitor my network over the internet would be to setup a FCI server on one of the Linux client and then setup FCI clients on all the other computers to upload their files to the server then you can setup Apache on the Linux server box to host FCI web pages for access from anywhere. FCI clients can be Linux, Windows, or OSX but the server has to be Linux its a great setup when you need to remotely monitor your folding and can not setup a file share easy its has a very low resource overhead.

That log getting written to about every 90 minutes was a error which was introduced in the latest smp client which really has no solution however its just a minor irritation it only effects the logs if any thing happens during the folding run which causes a error it will immediately catch up the log when it ends the run and it also catches up the log whenever if finishes a run and starts a new wu as well. Dr Pande thought the problem was in the way VMware handled the Linux Logs since the error is not present in non VMware Linux clients running the latest smp beta.

**~~Marvin_The_Martian~~** · 11-15-2008, 01:53 AM

That's how I monitor the vm workstation clients, which I set up using the iso. But the vmware appliance is a> only accesible using the ip for some reason, I can't see it on my network as '\\Diskless\FOLDXXXXXX\'. And even if I had the correct path, from what I been reading the problem is with the vm apliance just not writing to the log in time.

Code:

[22:02:35] + Processing work unit
[22:02:35] Core required: FahCore_a2.exe
[22:02:35] Core found.
[22:02:35] Working on Unit 01 [November 14 22:02:35]
[22:02:35] + Working ...
[22:02:35] - Calling './mpiexec -np 4 -host 127.0.0.1 ./FahCore_a2.exe -dir work/ -suffix 01 -checkpoint 15 -forceasm -verbose -lifeline 441 -version 602'

[22:02:35] 
[22:02:35] *------------------------------*
[22:02:35] Folding@Home Gromacs SMP Core
[22:02:35] Version 2.01 (Wed Aug 13 13:11:25 PDT 2008)
[22:02:35] 
[22:02:35] Preparing to commence simulation
[22:02:35] - Ensuring status. Please wait.
[22:02:44] - Assembly optimizations manually forced on.
[22:02:44] - Not checking prior termination.
[22:02:49] - Expanded 4834508 -> 23977273 (decompressed 495.9 percent)
[22:02:51] Called DecompressByteArray: compressed_data_size=4834508 data_size=23977273, decompressed_data_size=23977273 diff=0
[22:02:52] - Digital signature verified
[22:02:52] 
[22:02:52] Project: 2669 (Run 2, Clone 129, Gen 20)
[22:02:52] 
[22:02:52] Assembly optimizations on if available.
[22:02:52] Entering M.D.
[22:02:58] Will resume from checkpoint file
[23:16:19] Resuming from checkpoint
[23:16:19] Verified work/wudata_01.log
[23:16:19] Verified work/wudata_01.trr
[23:16:19] Verified work/wudata_01.xtc
[23:16:19] Verified work/wudata_01.edr
[23:16:19] Completed 37519 out of 250000 steps  (15%)
[23:16:19] Completed 40009 out of 250000 steps  (16%)
[23:16:19] Completed 42509 out of 250000 steps  (17%)
[23:16:19] Completed 45009 out of 250000 steps  (18%)
[23:16:19] Completed 47509 out of 250000 steps  (19%)
[23:16:19] Completed 50009 out of 250000 steps  (20%)
[00:33:00] Completed 52509 out of 250000 steps  (21%)
[00:33:00] Completed 55009 out of 250000 steps  (22%)
[00:33:00] Completed 57509 out of 250000 steps  (23%)
[00:33:00] Completed 60009 out of 250000 steps  (24%)
[00:33:00] Completed 62509 out of 250000 steps  (25%)
[00:33:00] Completed 65009 out of 250000 steps  (26%)
[01:50:02] Completed 67509 out of 250000 steps  (27%)
[01:50:02] Completed 70009 out of 250000 steps  (28%)
[01:50:02] Completed 72509 out of 250000 steps  (29%)
[01:50:02] Completed 75009 out of 250000 steps  (30%)
[01:50:02] Completed 77509 out of 250000 steps  (31%)
[01:50:02] Completed 80009 out of 250000 steps  (32%)
[03:07:04] Completed 82509 out of 250000 steps  (33%)
[03:07:04] Completed 85009 out of 250000 steps  (34%)
[03:07:04] Completed 87509 out of 250000 steps  (35%)
[03:07:04] Completed 90009 out of 250000 steps  (36%)
[03:07:04] Completed 92509 out of 250000 steps  (37%)
[03:07:04] Completed 95009 out of 250000 steps  (38%)
[04:02:20] - Autosending finished units...
[04:02:20] Trying to send all finished work units
[04:02:20] + No unsent completed units remaining.
[04:02:20] - Autosend completed
[04:20:35] Completed 97509 out of 250000 steps  (39%)
[04:20:35] Completed 100009 out of 250000 steps  (40%)
[04:20:35] Completed 102509 out of 250000 steps  (41%)
[04:20:35] Completed 105009 out of 250000 steps  (42%)
[04:20:35] Completed 107509 out of 250000 steps  (43%)
[04:20:35] Completed 110009 out of 250000 steps  (44%)
[05:35:44] Completed 112509 out of 250000 steps  (45%)
[05:35:44] Completed 115009 out of 250000 steps  (46%)
[05:35:44] Completed 117509 out of 250000 steps  (47%)
[05:35:44] Completed 120009 out of 250000 steps  (48%)
[05:35:44] Completed 122509 out of 250000 steps  (49%)
[05:36:22] Completed 125009 out of 250000 steps  (50%)
[06:52:38] Completed 127509 out of 250000 steps  (51%)
[06:52:38] Completed 130009 out of 250000 steps  (52%)
[06:52:38] Completed 132509 out of 250000 steps  (53%)
[06:52:38] Completed 135009 out of 250000 steps  (54%)
[06:52:38] Completed 137509 out of 250000 steps  (55%)
[06:54:09] Completed 140009 out of 250000 steps  (56%)
[08:09:45] Completed 142509 out of 250000 steps  (57%)
[08:09:45] Completed 145009 out of 250000 steps  (58%)
[08:09:45] Completed 147509 out of 250000 steps  (59%)
[08:09:45] Completed 150009 out of 250000 steps  (60%)
[08:09:45] Completed 152509 out of 250000 steps  (61%)
[09:23:09] Completed 155009 out of 250000 steps  (62%)
[09:23:09] Completed 157509 out of 250000 steps  (63%)
[09:23:09] Completed 160009 out of 250000 steps  (64%)
[09:23:09] Completed 162509 out of 250000 steps  (65%)
[09:23:09] Completed 165009 out of 250000 steps  (66%)
[09:23:09] Completed 167509 out of 250000 steps  (67%)
[09:29:36] Completed 170009 out of 250000 steps  (68%)
[10:02:20] - Autosending finished units...
[10:02:20] Trying to send all finished work units
[10:02:20] + No unsent completed units remaining.
[10:02:20] - Autosend completed

**Shadowtester** · 11-15-2008, 04:39 AM

That is correct Marvin it seems to be a bug which was introduced with the version 6.02 smp client and with the latest 6.23 smp client has got even worse Dr Pande is aware of the problem with VMware log handling however so far there has not been any fix for it as it only seems to effect the VMware clients not standard clients.

**~~Marvin_The_Martian~~** · 11-15-2008, 04:52 AM

Well that's not true imo because the clients I installed using the downloaded iso in workstation don't have the problem. It's really limited to my one client I installed on the c2d using the vmx appliance file

It's abit of a let down for me, I need to rely on windows taskman and the handlog to be sure it's running but getting any sencefull ppd information is tedious because I have to cut the time between the last frame reported and the first frame reported and devide it by the number of frames reported

I have a small app running in debug mode atm to do this but even so it will only update ppd or current frame after the vm spews out a number of frames in sequence.

Btw you will be able to answer this I think, all notfreds installs have this textfile 'cpuinfo.txt' is that the same with any native linux install? Asking for the eue parser, want to include the hw info as well and I don't want to look into getting shell acces to a linux box and running the command line options posted in the 'help with eue parsing for linux' thread as I think I will get a severe headache getting it done.

Kid in bed, dog asleep after a long walk and a shower so I'm working on 3 things at the same time

not my best forte but I want to finish the eue parser ( finish = no bugs and every feature I want in it done ) and I want the contest tracker done for XS, and with that I don't mean just the exported html pages with graphics but also a framework to include them in

Taking to much load on me

Will see when the ship begins to sing

Edit: btw above log is from a client with 640mb allocated and it's running the same project one off my vista boxes is doing with advmethods and big wu's, I'm hoping it will complete and chocke on the last few frames

**Shadowtester** · 11-15-2008, 05:44 AM

No my native Linux 6.23 smp client install does not have a cpuinfo.txt file only client.cfg and unitinfo.txt so that must be something to do with nofreds. As for your clients installed from workstation iso not having the problem I don't think that is the answer. I have had the log function work fine on some clients in VMware on one wu and on the next it would exhibit the problem so the problem is intermittent. I noticed the problem more on a2 core wu's than a1 core wu's but I had seen it on both and I have had both complete a wu without the error so its very hard to isolate the cause.

**p2501** · 11-16-2008, 12:52 AM

A little update: both VM's are still running, yay. They are both displaying this "attempt to access beyond end of device" error, though.

One of them caught a 5101 over night, which seems to be one of the slower WU. It doesn't have as many atoms as 2669 but still, it called for a1 core and this is why I'm netting just 1500PPD from this one I guess. The same VM has problems sending a finished WU to PG, bleh.

Thread: VMWare problems/help thread

Thread Tools

Search Thread

Rate This Thread

Display

VMWare problems/help thread

Bookmarks

Bookmarks

Posting Permissions