I've been getting an increasing number of errored WUs lately.
A week ago it was every now and then ... perhaps 1 in 10.
Now it's more like 1 in 5.
Windows, Ubuntu and GPU WUs, are all the same.
Anyone else with this problem?
I haven't seen any failures this month on Ubuntu or Windows, but I'm only going through about 1400 credit a day in WUs.
Yes, I am seeing an increasing number that seem to process and complete correctly, upload and report OK and are shown as success, but get zero credit and for some reason are marked invalid. Not as many as 1 in 5, but probably at least 1 in 20.
I think it may be my preferences as I've selected yes to running test applications.
Might be just that.
You need to turn off 'allow new tasks' to even see them error as they are removed within a minute.
Yes these invalid WUs have been issued since about the middle of June I think. There is a thread about it on the MilkyWay forums. All my ones that are marked invalid are the ps_sgr_210F5_3s_hiw types.
When I last counted on 22 June, I was getting 4 per hundred WUs downloaded that were marked invalid. This was about 30% of all ps_sgr_210F5_3s_hiw WUs.
I'm still getting some of these invalid, but I haven't checked closely how many. When I looked just now it was about one every 2nd page of 20 results, so about 2-3% of total WUs received. I haven't been watching closely though and perhaps have missed noticing if a larger than usual number of ps_sgr_210F5_3s_hiw have been issued recently.
So, who had the greatest fall in total credit this morning?
I just lost 16000 roughly...... and jobs erroring out. I said somewhere else that I do wonder at times if this project is more about persistence than science *grin*
From Milky Way
Quote
Server Crash
July 29, 2009
Our database crashed or got corrupted this morning and it's looking we lost a lot of user accounts and credits. For now if you lost your account just make a new one. We're looking into this right now and are trying to restore all that we can.
--Travis
Sometimes I think this project is more trouble than it is worth. biggrin
EDIT: BOINCStat's just updated (~18:00 UTC) and all my Milkyway numbers are shown as negative numbers. :thumbdown: Ooops! biggrin :rofl: biggrin
kaboom
-200,000 for me
thats a month or more of crunching for me haha
edit: Big losers in this massive market crash - DidierT with a whopping -9 million ..... this is the highest Ive ever been in the team for anything - negative credit :rofl: :rofl:
reedit: logging into Milky account, its all back where it should be ..... thank goodness!
hhaah -1,200,000 for me !! hahahha
that hurts..
I lost over 5 million, but most of it is back now. I say most of it - apparently anything that was crunched yesterday is going to be permanently lost -> about 140K in my case. :cry2:
That's unfortunate, I lost 20-25K myself.
I did a few Collatz WUs on the GPU after MilkyWay crashed, but then that was unreachable also. The server there seems to have problems now and again. After that my internet went down and refused to connect. I turned it off and it connected OK this morning. I'm glad I didn't have to ring the call centre in the Phillipines to ask for a tunnel reset. That is very time consuming because they insist you do everything on their script even though you know the problem is at their end.
This morning I find that not only are my MW numbers still negative at BOINCStats, now I am back to the previous team I was crunching for. ??? I'm not sure they can map their servers much less the universe. :jester:
Yes BOINCstats only does a full update once per day I think. Not sure when the next update is due, but possibly in about 3 hours or so it will be fine. Wouldn't want to lose you so soon after joining the team. :)
Some other stats sites have updated and show that you are now back in the BOINC@AUSTRALIA MilkyWay team.
http://stats.free-dc.org/stats.php?page=user&proj=mil&name=7486
http://www.boincsynergy.com/stats/boinc-stats.php?id=7486&project=mwah
Quote from: Dataman on July 30, 2009, 11:24:45 PMI'm not sure they can map their servers much less the universe. :jester:
That's got to be the quote of the month! biggrin Plus one for making me smile.
QuoteWouldn't want to lose you so soon after joining the team.
No worries. You will not lose me, although you may want to throw me out. :rofl: :rofl: :rofl: I have a big mouth. :shock
A big mouth ...? That's something we value in Australia (well, I do..) .... you gotta say what you mean or you'll get something else instead...
Quote from: veebee on July 31, 2009, 06:32:40 AMyou gotta say what you mean or you'll get something else instead...
That's so true! :yahoo:
Hey Folks. Milky Way Wus are failing at an alarming rate. Thread on the web site. Just wasted hours/days maybe with the damn things. Set and forget my foot :furious: :furious: :furious:
What a nuisance, recently I lost a batch of yoyo Evolution workuniits that had reached 10 hours when my computer crashed and rebooted. No checkpointing, bah, humbug.
You may wish to give Moo! Wrapper (http://moowrap.net/index.php) a whirl. Currently GPU only with ATI Stream and NVIDIA CUDA applications for Windows and Linux. It's a wrapper for the distributed.net RC5-72 dnetc application, so similar to DNETC@HOME. Not on boincstats yet, but on Free-DC stats.
Very new project so bugs are still being worked out. High CPU usage and very slow performance on some multiple GPU configurations. However it is using the weird new BOINC credit system so tasks may grant more credit the slower they process. I had some corkers early but now seem to be getting much less per task than anybody else. :( Not to worry, it has still given me a fair bit.
I took out my 5970 and put a 5870 back in. It needed core selection changed to 3 in project preferences to overcome extreme sluggishness. Project forum provides more detailed information about the project.
Give it a go if you like, if it works on your configuration it may be able to compensate you for what you lost on MilkyWay. biggrin
Thanks Kashi. I'll give it a whirl shortly. I am liking Prime Grid a lot as it is very stable and pays pretty well. Note that DNA is having or has had similar problems to Milky Way. Times out after 2 min 47 seconds so I am guessing that WUs finished under that time should be OK. Might even play around with it tomorrow as well although I am getting a bit sick of MW and all the problems that arise.
Yeh I had Dotsch/UX running in a VirtualBox doing some DNA. The 64-bit Linux application is much faster than the Windows application. Its speed has attracted the attention of our old friend Saengar who has expressed his outrage at how much credit he is getting, so the current credit rate probably won't last long at DNA. ;D
Just finishing off some primaboinca tasks in my Windows BOINC cache now, then will give DNA another run in Linux and make hay while the sun shines. :thumbsup:
2011-07-14: MilkyWay@home - DOUBLE CREDITS Fundraiser - Request for Funding for MilkyWay@home
The big news! We have been planning a double-credit fund-raising event here at Milkyway@home. To explain the details, here's a letter from Prof. Heidi Newberg:
Dear MilkyWay@home volunteers,
I want to first thank all of you for helping us study the Milky Way galaxy by donating your CPU and GPU cycles to our project. Some of you have also contributed your computational expertise, or contributed cash or GPU cards to our projects. We appreciate all of these gifts. MilkyWay@home CPU/GPU cycles have contributed to defining the distribution of stars in the Sagittarius dwarf tidal stream. Stars in this stream were pulled off of the Sagittarius dwarf galaxy, which is in orbit around the Milky Way, by the Milky Way's gravity. In addition, novel computational techniques have been developed for asynchronous parameter searches using a heterogeneous volunteer network.
Over the next year we plan to modify MilkyWay@home to measure the overall distribution of stars in the Milky Way halo. The magnitude of your generosity has made it possible for us to dream bigger dreams. We are starting to transform MilkyWay@home to do n-body simulations of several dwarf galaxy tidal disruptions at the same time. We can then change the model parameters for the Milky Way's gravity and for the properties of the dwarf galaxies so that they match the properties of actual tidal streams in the Milky Way. This is a big job and a new field, and we would not be able to contemplate doing this if you were not volunteering the equivalent of one of the largest supercomputers in the world to this task. Presently, we have taken only the first steps towards this, by simulating the disruption of a single dwarf galaxy and varying only the parameters of the dwarf itself. We have quite a lot to learn in developing the correct techniques for solving this problem, but if we are successful we will eventually find out how much mass (mostly dark matter) is in the Milky Way, and how it is distributed in space.
A year ago, I was successful in getting an NSF grant funded that supports my astronomy research on the Milky Way, and a significant part of that grant was justified by the use of your volunteer computing cycles. We are still seeking grant funding that will support the computer science research side of MilkyWay@home. A few years ago when we were stretched for funding between grants, we raised about $7000 in private donations, most of which were from MilkyWay@home volunteers. The remaining portion of this funding is currently being used on summer salary to support Matthew Arsenault, who is working on converting our n-body code so that it will run on GPUs. Matthew recently received his BS degree from RPI, and has been working on MilkyWay@home code for more than a year. He will be entering our PhD program this fall. He will be supported as a teaching assistant during the coming school year, but I am again turning to the volunteers to help me support him in summer 2012. Currently, I do not have funding to guarantee that I can support him. The cost of a graduate student summer salary plus institutional overhead is about $10,000. Of course, we can always do more with extra funding, but students are my top priority. Matthew has done a fabulous job implementing our current n-body simulations.
We are therefore launching the following "thank you" and fundraising campaign. On September 1, 2011, we will run MilkyWay@home with double credit. Hopefully, this will at least make up for some of the credit lost when MilkyWay@home was not operational for software, hardware, or air conditioning emergencies. We understand that these have been frustrating to some users, but be assured that they were much more frustrating for us.
In addition, we will run MilkyWay@home with double credit for one additional day for every $500 raised between now and September 1, 2011. Donations can be made with paypal or credit card online at:
http://www.dudleyobservatory.org/MilkyWayAtHome/MilkyWayAtHome2.html
Payments by check can be made out to Dudley Observatory and mailed to: Dudley Observatory, 107 Nott Terrace, Suite 201, Schenectady, NY 12308, with MilkyWay@home in the comments line of the check, or with a letter that explains that the donation is to MilkyWay@home. For donations in other forms, you may contact me directly (heidi@rpi.edu). The donations are tax deductible, and you will receive a letter from Dudley Observatory that can be used for tax purposes. Donations will be used to fund graduate or undergraduate students, travel, publication, materials, or equipment costs related to Milky Way research and education. For gifts over $1000, donors may specify a person, family, or organization to be acknowledged in research publications which are partially supported by this donation, and to receive a signed, printed copy of the publication.
Again, we thank you for your support of our research project. We have been overwhelmed by the number of people who are willing to contribute.
Sincerely,
Prof. Heidi Newberg
Further questions and comments are welcome in this thread.
looks like a scam email.
I understand they need money as do many other projects but I do not agree with involving credits in this. The credits are meant to represent the amount of work done and are based on the formula of Andreas Przystawik that set credit at the same rate per flop as SETI adjusted for the extra resources required for double precision. Giving double credit to compensate for the large amount of completed work that received no credit due to dumping the database and rolling back the credit after the massive server crash 3 months ago I have no problem with. But giving double credit because of cash donations is not right, it devalues the concept of credit being based on actual work performed and opens the door to the horrid CreditNew which threatens the whole of BOINC. It has already stopped one project dead in its tracks and caused problems at others.
2011-07-18: MilkyWay@home - DOUBLE CREDIT FUNDRAISER CANCELLED
(Posting this to get it up as a news item it is also in the original thread below this one in the News section).
Dear MilkyWay@home volunteers,
It appears that I have made a mis-step in management of MilkyWay@Home, regarding double credits for fundraising. I believed that this had been done before. I have since discovered that not only I was incorrect, but it is also inconsistent with the BOINC credit system. As volunteers, you have donated enormous computational resources to my group, and on top of that have created a powerful community that not only helps us with software but also teaches us how to be responsible members of your community. We will be running double credit two days this week to make up for lost credit in the past, but we will not be running a double credit fundraiser in September. I apologize for any inconvenience this has caused. I will need a little time to understand what the right way is to do this, so stay tuned.
Best Wishes,
Prof. Newberg
011-07-18: MilkyWay@home - Make-Up Credits to Run Wednesday and Thursday
Hello,
We'll be running two days of double-credits this week on Wednesday and Thursday to make up for outages earlier in the summer.
Cheers,
Matthew N
They like to throw those double credits around. Not what the credit system was designed for.......
I'm not sure but I think this is the first time they have had double credits on purpose. The other times were due to a misconfiguration.
I agree that the current period of double credits was unnecessary. Travis decided to give this double credit to stop the whingeing about losing a few days' credit back in April. So it can be justified on that basis and in that sense I have no problem with it.
However the usual credit rate on MilkyWay ATI is still quite good, even if it isn't quite as much as DNETC or Moo!, so it would have been wiser and more discreet to just cop it sweet. Doubling credit for everyone, even those who did not lose a few days' credit, just draws attention to MilkyWay and may hasten efforts to bring GPU applications under the control of CreditNew.
The good thing is they canned the idea of double credit for donations and Prof. Newberg was generous in her praise and acknowledgement of the large donations of computer resources from the volunteers. Don't need a pat on the back all the time but it's nice to hear occasionally that your crunching efforts are appreciated.
What's done is done, so I decided to whack my 5970 back in for a few days so that I am not disadvantaged relative to other MilkyWay crunchers. It's a noisy wee beastie!
:rofl: I was rather sure MW could not stay up for 48 hour under a big load. I should have made a wager! Problem is I have 100+ wu's stuck waiting for upload. :cry2: Oh well, the double credits were awsome on 12 cards. :crazy
Cheers all.
:aus1:
Haha yes, always on the cards, although the server didn't seem to slow down at any time, so it was probably a big load of hot weather causing RPI to turn off their air conditioning again. Got three quarters of the 2 days before it stopped anyway and a fair whack of credits. I've only got 36 tasks waiting to upload.
Much cooler in Troy, NY today ... even some showers. Maybe some one will come in give it a kick today. :compbash:
Ah Troy, guess they need strong firewalls at RPI to guard against Trojan horses. >:D
Laocoön, follow'd by a num'rous crowd,
Ran from the fort, and cried, from far, aloud:
'O wretched countrymen! What fury reigns?
What more than madness has possess'd your brains?
Think you the Grecians from your coasts are gone?
And are Ulysses' arts no better known?
This hollow fabric either must inclose,
Within its blind recess, our secret foes;
Or 't is an engine rais'd above the town,
T' o'erlook the walls, and then to batter down.
Somewhat is sure design'd, by fraud or force:
Trust not their presents, nor admit the horse.'
Quote from: kashi on July 25, 2011, 01:18:33 PM
Ah Troy, guess they need strong firewalls at RPI to guard against Trojan horses. >:D
Laocoön, follow'd by a num'rous crowd,
Ran from the fort, and cried, from far, aloud:
'O wretched countrymen! What fury reigns?
What more than madness has possess'd your brains?
Think you the Grecians from your coasts are gone?
And are Ulysses' arts no better known?
This hollow fabric either must inclose,
Within its blind recess, our secret foes;
Or 't is an engine rais'd above the town,
T' o'erlook the walls, and then to batter down.
Somewhat is sure design'd, by fraud or force:
Trust not their presents, nor admit the horse.'
You never cease to impress me, kashi. +1 :worship
Server is back up but not running as yet.
:aus1:
I hav,nt crunched MilkyWay for a long time. I noted that I do not belong to any team, so I attempted to join Boinc@Australia and received the following error: Deprecated: Creation of dynamic property BoincTeam::$refcnt is deprecated in /home/boinc/boinc/milkyway/html/user/team_search.php on line 67.
Any idea what that means?
Wow... no idea sorry Pete. I've never seen that before.
It's been ages since I've run M@H.
I searched there forum and nothing came up but it looks like a server error. I would ask in the forum and see if you get a response. I have never seen it before but I think it is a PHP error.