climateprediction.net home page
Hard Drive Crash.

Hard Drive Crash.

Message boards : Number crunching : Hard Drive Crash.
Message board moderation

To post messages, you must log in.

AuthorMessage
Firewagon

Send message
Joined: 30 Dec 12
Posts: 7
Credit: 162,529
RAC: 0
Message 45627 - Posted: 10 Mar 2013, 2:05:23 UTC

I recently had a hard drive failure. This was the 2nd time in 6 months. After much testing, drive has been deemed unreliable. Now my problem.... I was about half way through a 300 hour wu. Luckily the data files for boinc were not in an affected area of the drive so they were all recovered. My question is this.... Can I copy the wu to the new drive to complete it? otherwise it seems a terrible waste of cycles. Please note that the computer has also been upgraded....

Old system: Intel(R) Pentium(R) CPU G6950 @ 2.80GHz [Family 6 Model 37 Stepping 2](2 processors)

New system : AMD FX(tm)-8320 Eight-Core Processor [Family 21 Model 2 Stepping 0](8 processors)

My concern is that if I copy the files, the difference would cause the wu to error which would be an even bigger waste of cycles.

Question is simple... is it worth it?
ID: 45627 · Report as offensive     Reply Quote
old_user671679

Send message
Joined: 30 Jan 12
Posts: 38
Credit: 10,197,388
RAC: 0
Message 45628 - Posted: 10 Mar 2013, 4:19:14 UTC

You won't get any errors if you do it right, I've done it three or four times and the hardware doesn't matter. Might I suggest an Enterprise class hard drive, they have a longer validation process and warranty.

You need to find both the program files folder and the data folder, install the exact same version of BOINC then copy and past both folders back where they belong and it should work.

The full resolution ocean models are very sensitive to the computer rebooting without a proper shut down, what I'm saying is if the computer just froze, the model my error out no matter what you do. Good luck.
ID: 45628 · Report as offensive     Reply Quote
Profile mo.v
Volunteer moderator
Avatar

Send message
Joined: 29 Sep 04
Posts: 2363
Credit: 14,611,758
RAC: 0
Message 45634 - Posted: 10 Mar 2013, 15:50:20 UTC
Last modified: 10 Mar 2013, 16:05:59 UTC

You can transfer the models to another computer or drive if they both have the same operating system. So you can't take the model data from a computer with Linux and transfer it to a Mac, for example. But you can transfer model data from XP to Win 7 because they're both Windows.

I would want both the new and the old computer/drive to have BOINC 6 or both to have BOINC 7. I wouldn't want to move folders and files between one and the other. It wouldn't matter whether the old drive had one version of BOINC 7 and the new drive another version as long as it was also BOINC 7.

Moving BOINC data between Intel and AMD systems isn't a problem.

If you need to download BOINC 6 there's a web page where all the versions are listed and available.

I've done this successfully a few times but not recently. I'd install BOINC on the new drive. Exit from BOINC completely. I wouldn't touch the BOINC folder (this is the one that contains the Locale subfolder).

The BOINC Data folder is the one that contains the Projects and Slots subfolders.

I'd go to the new BOINC Data folder, double-click on it to open it up. Delete the entire contents. (It helps if you have an empty Recycle Bin before you start because if you make a mistake you can send everything back to where it came from.) To do this in the menu > Edit > Select all, then Edit > Delete.

You then need to get the contents of your old BOINC Data folder. Double-click to open this folder, then Edit > Select all, then Edit > Copy.

If you then go back to the new BOINC Data folder, you can right-click on it and select Paste.

You should then be good to go.

Where the BOINC and BOINC Data folders are located in each of the three OSs is explained in section 3 articles here for both BOINC 6 and BOINC 7:

http://boincfaq.mundayweb.com/index.php

If you make a mistake in the copying of the files it isn't really a problem as you can try again and again. But as Flashawk says, if problems with the old drive caused a computer crash, this may have crashed the model(s) and you won't be able to salvage anything unless you by chance have a backup made before the time of the crash.

Let us know what happens.
Cpdn news
ID: 45634 · Report as offensive     Reply Quote
Firewagon

Send message
Joined: 30 Dec 12
Posts: 7
Credit: 162,529
RAC: 0
Message 45638 - Posted: 11 Mar 2013, 3:30:10 UTC

Thanks for the info....

Unfortunately the BOINC program file folder was corrupted, it only had a few exe and txt with the logo file left. I was hopeful when I saw the program data seemed un-affeccted.

Oh well.. was worth a shot.

On the bright side, I am gonna have a few really cool coasters to put my coffee cup on to protect the finish on my desk. :)

@flashawk... I have been a big fan of Western Digital for many years (too many to count) they helped me by retrieving data I thought was lost for free and replaced a drive outside of warranty also(computer was hit by lightning). This drive unfortunately was initially part of a "discount store built bargain system" when I last upgraded my computer but was too lazy to design and build it my self. It wasn't even a SATA! I hate HITACHI!!!! Soon, (very soon) the 2nd drive from that system (all that is left of the original system)will be replaced by a Western Digital.
ID: 45638 · Report as offensive     Reply Quote
Firewagon

Send message
Joined: 30 Dec 12
Posts: 7
Credit: 162,529
RAC: 0
Message 45639 - Posted: 11 Mar 2013, 3:32:55 UTC - in response to Message 45638.  

Also... forgot to ask,,,,,

Is there any way for me to get the wu downloaded again?
ID: 45639 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 45640 - Posted: 11 Mar 2013, 3:59:42 UTC - in response to Message 45639.  
Last modified: 11 Mar 2013, 4:25:28 UTC

No.

PS
Regular backups would have 'Saved' things for you.
Backups: Here
ID: 45640 · Report as offensive     Reply Quote
Firewagon

Send message
Joined: 30 Dec 12
Posts: 7
Credit: 162,529
RAC: 0
Message 45641 - Posted: 11 Mar 2013, 5:04:09 UTC

Lesson learned...... Discount (not discounted, there is a diff) is crap. You get what you pay for.

I never thought to include Boinc directories in my backups. That issue has been corrected though.
ID: 45641 · Report as offensive     Reply Quote
Firewagon

Send message
Joined: 30 Dec 12
Posts: 7
Credit: 162,529
RAC: 0
Message 45646 - Posted: 12 Mar 2013, 13:17:55 UTC

Just an update.....

I read in one one of the backup/restore instructions that the PROGRAM DATA directory only is needed not the PROGRAM FILES directory is required to restore when using boinc version 6 and later. I restored the PROGRAM DATA directory salvaged from the old drive and all the pending wu have reappeared for all projects I participate in. Considering the fact that I was using version 7.0.25(32bit) on INTEL prior to the crash and am now using 7.0.28(64bit) I am happy with this. It looks like I have not lost all the work.

The crunching seems to be going well.

I'll post when the wu's are done and validated.
ID: 45646 · Report as offensive     Reply Quote
Eirik Redd

Send message
Joined: 31 Aug 04
Posts: 391
Credit: 219,888,554
RAC: 1,481,373
Message 45649 - Posted: 12 Mar 2013, 17:07:50 UTC

save both folders - it doesn't cost much if you are trying to save and get restarted.

There's lots of info on saving and restarting - do a search on this forum.

Sometimes a save- restore works - sometimes not so good.

And great big thanks for donating your time and cycles.


ID: 45649 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 45651 - Posted: 12 Mar 2013, 19:26:07 UTC - in response to Message 45649.  

Sorry Eric, but the program part can always be recreated by just re-installing BOINC, either from the original zip download, or a fresh download.
This becomes an opportunity to also upgrade BOINC.

Hang in there Joel; if it's now running, let it.


Backups: Here
ID: 45651 · Report as offensive     Reply Quote
Firewagon

Send message
Joined: 30 Dec 12
Posts: 7
Credit: 162,529
RAC: 0
Message 46130 - Posted: 30 Apr 2013, 4:45:23 UTC

Well an update......

As you can see on my profile, the WU's I had when the drive crashed ended in error status which seriously sucks.... but hey it happens. But Ihave since had success with a WU downloaded after the problem.

I read somewhere in the boards that climate models are very "fussy" and can error out for strange reasons or no apparent reason. I recently noticed something my computer is doing only when boinc is running. I get bad pixels displayed, that is pixels with no color in various sizes in diff places on screen at diff times. The "bad Pixels" vary in size from what looks like maybe pin size to 1/32 in size. The area varies for no rhyme or reason. I can "force" it to happen by putting the system under heavier than normal load. This leads me to believe it is a video driver issue.. I use Catalyst 13.1 by the way because my games (WOW, FARCRY3, CRYSIS3) require the most recent or almost drivers available.

My new question is.... If the video driver is boinc unfriendly (new word), could that cause the WU to error?


FYI: since BOINCSTATS was having issues with identifying my real name (JOEL) as an account ID, I updated to an old gamertag -> FIREWAGON
�Any fool can know. The point is to understand.�
- Albert Einstein

I on the other hand am just confused.
ID: 46130 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 46131 - Posted: 30 Apr 2013, 5:40:39 UTC - in response to Message 46130.  

One of the major changes from BOINC V5 to V6 was the separation of the graphics from the crunching. This was to stop what you describe: a graphics failure causing the crunching to crash.

However, there have been other problems introduced by "the very latest driver version". Which version exactly seems to vary across time, and also with the manufacturer. The cure has apparently been to downgrade a bit.

And I only know about this from constantly reading the BOINC/dev boards, not because I keep updating them.



Backups: Here
ID: 46131 · Report as offensive     Reply Quote
Firewagon

Send message
Joined: 30 Dec 12
Posts: 7
Credit: 162,529
RAC: 0
Message 46136 - Posted: 30 Apr 2013, 22:57:39 UTC

Thanks.

After reading a lot more about it.... it is a video driver issue.... Boinc does not like the new driver, but this does not appear to be a critical problem. Other projects I participate in are completing successfully without error so I wont worry about it. It is just really weird when you walk up to the computer and see all these blank dots everywhere (less than 10%).

Whenever I do something resource hungry (games, Video/Graphics editing, etc...) I suspend computation anyways. The only time I forgot, I got a BSOD. The projects still completed OK, but that was scary...lol

For now I will just monitor the error rate more closely. Hopefully boinc will catch up to catalyst 13.1 before 14 comes out.
�Any fool can know. The point is to understand.�
- Albert Einstein

I on the other hand am just confused.
ID: 46136 · Report as offensive     Reply Quote
Profile mo.v
Volunteer moderator
Avatar

Send message
Joined: 29 Sep 04
Posts: 2363
Credit: 14,611,758
RAC: 0
Message 46137 - Posted: 30 Apr 2013, 23:53:15 UTC

Your CPU and GPU are not at all similar to mine, but something I've discovered may be of use. I find that if I let the computer run for too many days without rebooting I start to see abnormalities in the graphics eg coloured pixels among text. A reboot fixes this immediately. In fact, the best thing is to regularly reboot once a week before this occurs.

I never saw anything like this when I had a run-of-the-mill graphics card, but the current one that's good enough for crunching is more sensitive. Nothing to do with CPDN.
Cpdn news
ID: 46137 · Report as offensive     Reply Quote

Message boards : Number crunching : Hard Drive Crash.

©2024 climateprediction.net