bad day


log in

Advanced search

Message boards : Number crunching : bad day

Author Message
Profile bj
Send message
Joined: 11 Oct 00
Posts: 163
Credit: 50,429,449
RAC: 1
United States
Message 1200731 - Posted: 28 Feb 2012, 23:26:36 UTC

Had two computers go T.I. this morning. Got the Windows 7 back up and running and was able to save the data.
Have to apologize to all the wingmen on my XP computer (#6056052) but that hard drive completely failed. There was no way in getting any data off of it.
Reimaging a new drive now but all the work on that computer is lost: it will be resent but probably will hang around for awhile
Sorry about that.

bj
____________

Claggy
Volunteer tester
Send message
Joined: 5 Jul 99
Posts: 3960
Credit: 31,817,790
RAC: 10,146
United Kingdom
Message 1200765 - Posted: 29 Feb 2012, 0:35:02 UTC - in response to Message 1200731.
Last modified: 29 Feb 2012, 0:52:12 UTC

Had two computers go T.I. this morning. Got the Windows 7 back up and running and was able to save the data.
Have to apologize to all the wingmen on my XP computer (#6056052) but that hard drive completely failed. There was no way in getting any data off of it.
Reimaging a new drive now but all the work on that computer is lost: it will be resent but probably will hang around for awhile
Sorry about that.

bj

There's always ways of restoring a lost host, having an old copy of the client_state.xml helps, but really isn't needed,

Once you're got an OS installed, and Boinc Installed, don't attach to any projects, copy account_setiathome.berkeley.edu.xml from one of your other hosts to your new Boinc Data directory,
copy a client_state.xml from one of your other hosts (with NNT set or network suspended) to your new Boinc Data directory,
edit the client_state.xml with notepad and change the <hostid> to your original host id, change the <rpc_seqno> to 2355 (just need a slightly higher value than 'Number of times client has contacted server' from the host details page),
delete everything from the client_state.xml except the <host_info> <time_stats> <net_stats> <project> <active_task_set> sections (you may delete the <active_task> sections from within the <active_task_set> section,
also leave in place the last nine lines or so,
close notepad saving the client_state.xml now start up Boinc, Boinc will repopulate all the CPU/GPU info with the current info, and will have the original host id,
eithier unset NNT or unsuspend Network usage, and ask for work, the server should resend it,
reinstall the Lunatics apps when you're ready (you could do that before starting Boinc)

Claggy

Profile bj
Send message
Joined: 11 Oct 00
Posts: 163
Credit: 50,429,449
RAC: 1
United States
Message 1200815 - Posted: 29 Feb 2012, 3:04:51 UTC

Claggy

Thanks for the info.
Will have to wait til my new drive comes in as I have to have the same type of drive for to re-image it.
I'll try that and hope I get the same work so as not to screw the works up for the others.
The old drive is completly hosed. Was hoping to get into it but to no avail.

bj
____________

rob smith
Volunteer moderator
Send message
Joined: 7 Mar 03
Posts: 7650
Credit: 44,579,410
RAC: 74,452
United Kingdom
Message 1200888 - Posted: 29 Feb 2012, 7:50:18 UTC

Its the season for crunchers failing - my main cruncher decided to develop a mobo fault on Monday and just stopped. New mobo, processor and memory installed, os (windoze xp pro) wouldn't see mouse or keyboard. Fixed keyboard, still no joy with mouse (but everything worked properly when I booted a trial-no-install ubuntu, so tried os repair, and that failed wiht a blue screen part way through (having "proper;y stuffed" the os on the disk, so now its find a recent xp pro install disk....
In the meantime I've got this old warhorse (4 core, 32 bits of ubuntu, so not that old...) to talk to a browser so I can talk to the world.....
____________
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

Claggy
Volunteer tester
Send message
Joined: 5 Jul 99
Posts: 3960
Credit: 31,817,790
RAC: 10,146
United Kingdom
Message 1200900 - Posted: 29 Feb 2012, 9:18:03 UTC - in response to Message 1200815.

Claggy

Thanks for the info.
Will have to wait til my new drive comes in as I have to have the same type of drive for to re-image it.
I'll try that and hope I get the same work so as not to screw the works up for the others.
The old drive is completly hosed. Was hoping to get into it but to no avail.

bj

I just tought of something, rather than snipping all those bits out of the client_state.xml, Just change the <hostid> and <rpc_seqno> then fire up Boinc, and 'Reset' the project, that will get rid of the Wu's from the donar host,

Claggy

Profile bj
Send message
Joined: 11 Oct 00
Posts: 163
Credit: 50,429,449
RAC: 1
United States
Message 1200981 - Posted: 29 Feb 2012, 16:05:45 UTC

Claggy

Was curious about something.
The drive is back-up everynight onto my other drive. But it adds the new files in the folder so what I end up with is all the work that was processed since the last time they were deleted.
Was wondering if I copied all those files into the folder; what would happen to the ones that have already been done? That way I would have the work that haven't been done but would also have work that was done and uploaded in the folder.

bj

____________

Claggy
Volunteer tester
Send message
Joined: 5 Jul 99
Posts: 3960
Credit: 31,817,790
RAC: 10,146
United Kingdom
Message 1201104 - Posted: 29 Feb 2012, 21:14:13 UTC - in response to Message 1200981.
Last modified: 29 Feb 2012, 21:15:29 UTC

Claggy

Was curious about something.
The drive is back-up everynight onto my other drive. But it adds the new files in the folder so what I end up with is all the work that was processed since the last time they were deleted.
Was wondering if I copied all those files into the folder; what would happen to the ones that have already been done? That way I would have the work that haven't been done but would also have work that was done and uploaded in the folder.

bj

O.K, If you have the Whole Boinc Data Directory backed up, this is what you need to do:

With no Internet connection, Restore your Backed up Boinc Data Directory, then install Boinc, fire Boinc up, then reset the Seti project, now shut Boinc Down,
copy all the Wu files from your Backed up setiathome directory to your setiathome directory,
now edit the client_state.xml and make the <rpc_seqno> 2 more than 'Number of times client has contacted server' from the host details page,
Startup Boinc, and reconnect the Internet connection, now ask for work, the Server will resend the work, oldest first, and skip the downloads,
If you monitor the Task pages for that host, you'll notice the tasks get new sent times dates as they get resent,
beware of trying to get VLARs resent to Nvidia GPUs, If the next set are VLARs, set NNT, change your project preferences (Use CPU, Not Use Nvidia GPU),
click update, make sure the scheduler contact was successful, then unset NNT and ask for CPU work,
rinse and repeat, changing Use CPU & Use GPU as needed,
Note: you don't have to wait 5 mins and 3 secs between attempts, as long as you don't have any backed off downloads you can ask again immediately,

Claggy

Message boards : Number crunching : bad day

Copyright © 2014 University of California