PDA

View Full Version : DistRTgen: BOINC is up and running again



RSS
02-08-12, 03:30 AM
I'm happy to announce that we are finally up and running again at full speed! To those of you whom might not now, we had issues with our brand new server. We were suffering from performance degradation and frequent spontaneous reboots. Now we have migrated to a temporary loaner server while our original server is being tested to find the cause of the issues. Previously it were running NetBSD 6.0 with LVM to manage the space, but we think it's what's causing the problems and so the server will be reinstalled with NetBSD 5.1 without LVM after the testing phase has completed, to keep things simple. Until then, we hope you will help us test our temporary loaner server. Let's see how much it can take before going down Let the crunching begin!

More... (http://boinc.freerainbowtables.com/distrrtgen/forum_thread.php?id=69)

Fire$torm
02-08-12, 03:59 AM
OH yeah baby! I'm on this.

STE\/E
02-08-12, 04:22 AM
I'm gonna stay off it until I see the Wu's Validating, I picked up a few over night & every one of them were marked Invalid ...

spingadus
02-08-12, 04:54 AM
So far I've completed 3 valid tasks. No errors so far.

STE\/E
02-08-12, 05:02 AM
I think I know why mine are being marked Invalid, it could be because I have Client 7.0.11 Installed on all my Box's now. So I've screwed myself unless I go back to 6.12.34 or something like that. In order to go back I'll lose a Ton of Wu's in progress, so the moral is sometimes it's best to just leave things alone in BOINC instead of trying to do better in a Challenge, I don't have the time or the motivation to go back in Clients right now, it's all just too frustrating some times ...

STE\/E
02-08-12, 06:55 AM
I've brought the Problem up at the DIRT Forums https://www.freerainbowtables.com/phpBB3/post19839.html#p19839 it appears the Client 7.0.xx won't Validate until the DIRT Server is Upgraded to accept the new Client, Cruncher Pete is showing Validation errors too with v7.0.3, mine is v7.0.11. It may be a few days before they do it though ...

Slicker
02-08-12, 10:35 AM
I bet DiRT uses gzip compression of the output files. Somehow, they broke that in 7.x.x while adding compression support for something else. Yoiu know how it goes.... for every one needless enhancement, there have to be two bugs added to BOINC. I think they must be using "SCRUM" which, according to Wikipedia, "is a form of agile project management... does not require any creativity or intelligence"

Technically though, 7.0.x is still being tested. It took 34 versions before they got 6.10.x to work well, so my guess is that 7.0.x probably has a ways to go.

Important: I'm pretty sure that to downgrade to 6.x.x, you will need to abort all work, detach from all projects, and then do a complete uninstall, potentially deleting the data directory manually before attempting to reinstall 6.x.x because 7.x.x changes so much stuff that you cannot roll back to a previous version. They strongly suggested making a complete backup of the 6.10.x data directory prior to installing 7.0.x for a reason. Yep. I'm sure everyone did that, right? If you want to use bleeding edge software, prepare to bleed.

STE\/E
02-08-12, 10:43 AM
Important: I'm pretty sure that to downgrade to 6.x.x, you will need to abort all work, detach from all projects, and then do a complete uninstall, potentially deleting the data directory manually before attempting to reinstall 6.x.x because 7.x.x changes so much stuff that you cannot roll back to a previous version. They strongly suggested making a complete backup of the 6.10.x data directory prior to installing 7.0.x for a reason. Yep. I'm sure everyone did that, right? If you want to use bleeding edge software, prepare to bleed.

Their going to Update the Server ASAP, I'm running POEM, Milkyway & MOO Wu's with the 7.0.11 Client & the Wu's are Validating so it can be done. Quote from PowerBlade at the Forum:


I know we have to do it eventually, so we might as well have to do it as soon as possible and properly document the steps to do the upgrade.
Then it will make it easier for us to do upgrades in the future.

For now I just renamed the Original BOINC Folders & installed v6.12.34 & it's working on just 1 Box. I'll wait & see if they Upgrade in the next few days before switching any other back to an earlier version.

Mike029
02-08-12, 11:59 AM
Their going to Update the Server ASAP, I'm running POEM, Milkyway & MOO Wu's with the 7.0.11 Client & the Wu's are Validating so it can be done. Quote from PowerBlade at the Forum:



For now I just renamed the Original BOINC Folders & installed v6.12.34 & it's working on just 1 Box. I'll wait & see if they Upgrade in the next few days before switching any other back to an earlier version.

I'm on it as well. Let's see if we can break it. :o

Mumps
02-08-12, 02:22 PM
I think I know why mine are being marked Invalid, it could be because I have Client 7.0.11 Installed on all my Box's now. So I've screwed myself unless I go back to 6.12.34 or something like that. In order to go back I'll lose a Ton of Wu's in progress, so the moral is sometimes it's best to just leave things alone in BOINC instead of trying to do better in a Challenge, I don't have the time or the motivation to go back in Clients right now, it's all just too frustrating some times ...

Well, I have 2 systems on it.

My Win7 64bit PC with an ATI card and BOINC 6.13.6 is not validating. (No ATI tasks I can see, all CPU.)
My Ubuntu Linux 11-04 64 bit with an nVidia card and BOINC 6.12.32 is validating everything. (Both GPU and CPU.)

STE\/E
02-08-12, 03:20 PM
Well, I have 2 systems on it.

My Win7 64bit PC with an ATI card and BOINC 6.13.6 is not validating. (No ATI tasks I can see, all CPU.)
My Ubuntu Linux 11-04 64 bit with an nVidia card and BOINC 6.12.32 is validating everything. (Both GPU and CPU.)

I've got just 3 Box's on it & all Validating, 2 that I reverted back to 6.12.34 & a Linux Box with Ubuntu on it, I'm going to wait for now & see if they Upgrade the Server to accept work form the 7.0.xx Clients ...

Fire$torm
02-08-12, 06:18 PM
If I'm not mistaken upgrading the server software will put that project under the "CreditNew" system. OH joy..........

Bryan
02-08-12, 06:42 PM
I had to downgrade 3 of my machines that have ATI and Cudas in them to get them to validate. I do have 1 machine with 2 GTX 570s that is working without any problems running 7.0.11. Go figure.

Mike029
02-08-12, 09:29 PM
I had to downgrade 3 of my machines that have ATI and Cudas in them to get them to validate. I do have 1 machine with 2 GTX 570s that is working without any problems running 7.0.11. Go figure.

Is the ATI paying better than Moo? Or is Moo still the way to go?

Bryan
02-08-12, 10:58 PM
Unless something has changed, Moo pays much better than DirT for ATIs. In those 3 boxes I have the ATIs on Moo and the Cudas on DirT ;)

Mike029
02-08-12, 11:09 PM
Unless something has changed, Moo pays much better than DirT for ATIs. In those 3 boxes I have the ATIs on Moo and the Cudas on DirT ;)
+1
Unless this new Donateathome thing holds up. Beats Moo with my 5870's. I know you still have few of these running Bryan. I can't get it to run on my 5970 tho. I'll work on a fix tomorrow. May just be the drivers are a bit old on the 5970.

spingadus
02-08-12, 11:15 PM
I haven't done any gpu crunching with linux, only cpu. Is there a best distro for gpu crunching? You know, one where you don't have to fiddle with drivers too much? It will be my HD 6970 box.

Bryan
02-08-12, 11:59 PM
+1
Unless this new Donateathome thing holds up. Beats Moo with my 5870's. I know you still have few of these running Bryan. I can't get it to run on my 5970 tho. I'll work on a fix tomorrow. May just be the drivers are a bit old on the 5970.

You certainly could be right, but it won't work for me. I've never been able to put both ATI and Cudas on the same project. The scheduler in BOINC is so stupid it will wind up filling one card or the other over time till it reaches the project maximum and letting the other run dry. It is really frustrating to see the Cuda out of work and BOINC keeps asking for ATI wu or vice versa! I talked to Slicker and he said the problem was with the scheduler.

Fire$torm
02-09-12, 12:51 AM
I haven't done any gpu crunching with linux, only cpu. Is there a best distro for gpu crunching? You know, one where you don't have to fiddle with drivers too much? It will be my HD 6970 box.

Nope. Linux w/GPU always means fiddling with drivers. I have some links for ATI hardware on Linux including an overclocking utility w/GUI. I'll try and find them for you.

Fire$torm
02-09-12, 01:08 AM
As promised....

[Phoronix] A New Open-Source AMD OverDrive Utility For Linux (http://www.phoronix.com/scan.php?page=news_item&px=ODk1Ng)

3D Acceleration for your ATI card. (http://linuxhelp.150m.com/ati/ati.htm)

AMD Catalyst™ Proprietary Display Driver (http://support.amd.com/us/gpudownload/linux/Pages/radeon_linux.aspx?type=2.4.1&product=2.4.1.3.42&lang=English)

Gentoo Linux: About Gentoo (http://www.gentoo.org/main/en/about.xml) (Probably the fastest Linux out there)

Arch Linux (http://www.archlinux.org/) (Probably faster then Ubuntu)


Not really relevant but you might find these interesting..

20 Linux System Monitoring Tools Every SysAdmin Should Know (http://www.cyberciti.biz/tips/top-linux-monitoring-tools.html)

Monitor A Remote Linux System From Windows with GKrellM (http://www.websightdesigns.com/tutorials/gkrellm-mini-howto/?tutorial=gkrellm)

FrozenCPU Linux Inside Case Badge (http://www.frozencpu.com/products/933/bad-467/FrozenCPU_Linux_Inside_Case_Badge.html)

spingadus
02-09-12, 03:07 AM
As promised....

[Phoronix] A New Open-Source AMD OverDrive Utility For Linux (http://www.phoronix.com/scan.php?page=news_item&px=ODk1Ng)

3D Acceleration for your ATI card. (http://linuxhelp.150m.com/ati/ati.htm)

AMD Catalyst™ Proprietary Display Driver (http://support.amd.com/us/gpudownload/linux/Pages/radeon_linux.aspx?type=2.4.1&product=2.4.1.3.42&lang=English)

Gentoo Linux: About Gentoo (http://www.gentoo.org/main/en/about.xml) (Probably the fastest Linux out there)

Arch Linux (http://www.archlinux.org/) (Probably faster then Ubuntu)


Not really relevant but you might find these interesting..

20 Linux System Monitoring Tools Every SysAdmin Should Know (http://www.cyberciti.biz/tips/top-linux-monitoring-tools.html)

Monitor A Remote Linux System From Windows with GKrellM (http://www.websightdesigns.com/tutorials/gkrellm-mini-howto/?tutorial=gkrellm)

FrozenCPU Linux Inside Case Badge (http://www.frozencpu.com/products/933/bad-467/FrozenCPU_Linux_Inside_Case_Badge.html)

Awesome! Thanks Fire$storm :)

Mike029
02-09-12, 10:10 AM
You certainly could be right, but it won't work for me. I've never been able to put both ATI and Cudas on the same project. The scheduler in BOINC is so stupid it will wind up filling one card or the other over time till it reaches the project maximum and letting the other run dry. It is really frustrating to see the Cuda out of work and BOINC keeps asking for ATI wu or vice versa! I talked to Slicker and he said the problem was with the scheduler.

Updating the drivers did not work. I mentioned this in the other thread as well. Both of my cards are ATI's one is the 5970 which is I.D'd as a 5870 in Boinc and a true 5870. It will not use the gpu or the cpu I even shut down WU & FreeHal. Nothing.

Al
02-09-12, 10:51 AM
Did I miss something? Are you trying to run DiRT on an ATI card? I wasn't aware they had an ATI app.:confused:

Slicker
02-09-12, 01:30 PM
If I'm not mistaken upgrading the server software will put that project under the "CreditNew" system. OH joy..........

That's nothing a few lines of code doesn't fix. After the server goes through all the CreditNew shananigans, and then right before it writes the CreditNew value to the database, you just interject your own value.

Mike029
02-09-12, 02:15 PM
Did I miss something? Are you trying to run DiRT on an ATI card? I wasn't aware they had an ATI app.:confused:

Doh!! Wrong thread. Sorry

STE\/E
02-12-12, 04:26 PM
The Project has Updated the Server to 7.0.7, I tried some with 7.0.11 and they Validated ...

Duke of Buckingham
02-12-12, 05:53 PM
Nice I will restart the project. Thanks for the update Steve.

Bryan
02-13-12, 11:46 AM
I've noticed a problem since the server upgrade but I don't know if it is coincidental or a problem. I had downgraded to BOINC 6.12.34 when they had the validation problem and everything worked fine. Since the server upgrade, my 3 machines that have both ATI and Cuda cards keep asking DirT for ATI wu which of course it doesn't supply. So where the cache should hold at 100 wu it starts dropping. I don't know what happens if the cache actually runs dry but I had to detach/attach one machine that had dropped to 17 wu. It seems strange that 3 different machines would experience the same problem all of a sudden. My machine with 2 Nvidia GPUs holds at 100 wu.

Anyone else seeing this or is it just my problem?

Fire$torm
02-13-12, 02:46 PM
I've noticed a problem since the server upgrade but I don't know if it is coincidental or a problem. I had downgraded to BOINC 6.12.34 when they had the validation problem and everything worked fine. Since the server upgrade, my 3 machines that have both ATI and Cuda cards keep asking DirT for ATI wu which of course it doesn't supply. So where the cache should hold at 100 wu it starts dropping. I don't know what happens if the cache actually runs dry but I had to detach/attach one machine that had dropped to 17 wu. It seems strange that 3 different machines would experience the same problem all of a sudden. My machine with 2 Nvidia GPUs holds at 100 wu.

Anyone else seeing this or is it just my problem?

I'm still using BM 6.10.58 on all my quad systems, one nVidia & one ATI each, and I'm not seeing any of that. With all the problems you guys are having with the newer BM's I'm sticking with what I have. For me POEM w/OpenCL just isn't worth all that grief.

Bryan
02-13-12, 02:58 PM
I just deattached/attached and things started working correctly again. Powerblade said that he had just increased the cache size to 200 per computer. I'm wondering if maybe those 3 machines asked for Nvidia work, the project said it didn't have any, and then the BOINC scheduler screwed up and started asking for ATI wu.

I've never had a problem with 6.12.34 so this was something new that just happened.

Al
02-13-12, 03:07 PM
Just ran out of DiRT on my 2-570's and am getting this error message:

2/13/2012 2:59:00 PM | DistrRTgen | [error] State file error: missing file 10084250 mysqlsha1_mixalpha-numeric-symbol32-space#1-7_0_40000x500000 - 21835000000_0_0
2/13/2012 2:59:00 PM | DistrRTgen | [error] Can't handle task 10084250 mysqlsha1_mixalpha-numeric-symbol32-space#1-7_0_40000x500000 - 21835000000_0 in scheduler reply

I've reset, it won't let me remove/detach...nothing. BM 6.12.34 (64)

coronicus
02-14-12, 02:01 AM
Just ran out of DiRT on my 2-570's and am getting this error message:

2/13/2012 2:59:00 PM | DistrRTgen | [error] State file error: missing file 10084250 mysqlsha1_mixalpha-numeric-symbol32-space#1-7_0_40000x500000 - 21835000000_0_0
2/13/2012 2:59:00 PM | DistrRTgen | [error] Can't handle task 10084250 mysqlsha1_mixalpha-numeric-symbol32-space#1-7_0_40000x500000 - 21835000000_0 in scheduler reply

I've reset, it won't let me remove/detach...nothing. BM 6.12.34 (64)

aye all my cpu boxes are doing that wether xp or windows vista/7 using boinc 6.12.34

Al
02-14-12, 06:54 AM
The very new install of win7 was still working. On the xp I had to reinstall/repair boinc 6.12.34, detach and then re-add the project. On the older win7 install of 6.12.34 I had to upgrade to the beta 7 of boinc....but everything is working now.

Mike029
02-14-12, 07:34 AM
The very new install of win7 was still working. On the xp I had to reinstall/repair boinc 6.12.34, detach and then re-add the project. On the older win7 install of 6.12.34 I had to upgrade to the beta 7 of boinc....but everything is working now.

Did an uninstall reinstall od 6.12.34 and still not pulling down cpu wu's on my ati box. Keeps asking for both cpu and ati work. I even created a new location for cpu only and still nothing.

Bryan
02-14-12, 04:40 PM
Did an uninstall reinstall od 6.12.34 and still not pulling down cpu wu's on my ati box. Keeps asking for both cpu and ati work. I even created a new location for cpu only and still nothing.

I had a problem with the scheduler asking for ATI wu even though the project runs Cudas. I redid 6.12.34 and it didn't help. I finally did a detach/attach and now the caches are full and staying that way. It now asks for the right type of wu.

Edit: that was happening on all 3 of my boxes running both ATI and Cuda cards. The dual Nvidia didn't have any problems.

Mike029
02-14-12, 04:41 PM
I had a problem with the scheduler asking for ATI wu even though the project runs Cudas. I redid 6.12.34 and it didn't help. I finally did a detach/attach and now the caches are full and staying that way. It now asks for the right type of wu.

Edit: that was happening on all 3 of my boxes running both ATI and Cuda cards. The dual Nvidia didn't have any problems.

I'll try that.

Fire$torm
02-16-12, 02:42 PM
Heads up guys, I'm getting CPU work finally. Had the same error you guys are seeing a few times. The detach/reattach seems to work for a time. Anyhow grab'em while you can. :D

Maxwell
02-23-12, 03:26 AM
DIRT is back down - make sure you have backups set accordingly...

Mike029
02-23-12, 10:35 AM
I'm only getting cpu work. I guess they ran out of cuda work.

Cappy
02-24-12, 12:15 PM
2/24/2012 11:10:08 AM | DistrRTgen | No tasks are available for Distributed Rainbow Table Generator (distrrtgen)

krazy k
02-24-12, 01:04 PM
Good thing I changed it to 10 days.

Cappy
02-26-12, 01:42 PM
just talked to quel over at dirt and he just cued the next cpu tables so

should be seeing cpu wu's shortly :)

Cappy
02-26-12, 03:04 PM
well hell if i knew asking for some cpu tasks would have taken them down again i wouldnt have asked for anything lol....

DAMN !!!!!!

ok PS: lol time is 3:32 pm EST their back up and running again and have CPU tasks .

krazy k
02-27-12, 10:13 PM
well hell if i knew asking for some cpu tasks would have taken them down again i wouldnt have asked for anything lol....

DAMN !!!!!!

ok PS: lol time is 3:32 pm EST their back up and running again and have CPU tasks .

I've had cpu tasks for a week, but gpu for only a couple days. I'm running out, and it was said that this is a hit and miss project. So it's safe to assume that they're out of work?