WARNING: This website is obsolete! Please follow this link to get to the new Albert@Home website!

Posts by ChertseyAl

1) Message boards : Problems and Bug Reports : WUs cancelled? (Message 111748)
Posted 26 Jan 2012 by Profile ChertseyAl
Post:

I waste many hours of computing for nothing :(


Like it says on the home page "Don't expect ANYTHING to work here".

At least you only lost a few WUs worth of time. I've lost about 20 WUs (so far), and one user lost 700 in one go last time this happened. It goes with the territory.

It's not a project for the credit collector. Best to assume zero credit for everything and treat anything you get as a bonus :) Plenty of work at Einstein if you want a more reliable project.

Al.
2) Message boards : Wish List : Low credit for such a risky project (Message 111372)
Posted 21 Nov 2011 by Profile ChertseyAl
Post:

You come into a shop ...


Slightly odd analogy :)

But then I live in the Utopian world where BOINC is open, relies on the generosity of volunteers and has a strong community spirit. I should take off my rose-tinted spectacles perhaps :)

Dude, I'm on your side.

Wipe the stats and get DA to ask the stats sites to remove the cached stats. That's a joke BTW :)

It was unfortunate that stats ever got exported. The best exit strategy for all concerned seems to be the path you've taken - Export for a while, then stop having warned everyone. Then we can all stay friends :)

Al.
3) Message boards : Wish List : Low credit for such a risky project (Message 111365)
Posted 21 Nov 2011 by Profile ChertseyAl
Post:
But I still like to understand what happened here, how & why people that are only interested in credit ever attached a host to this project. This project was never meant to attract this kind of attention. I'd like to avoid such misunderstanding in the future.


As I said earlier "The hardcore milestone collectors will crunch *any* project that's available to them".

It's not just this project, there's loads of projects out there not listed on stats sites that I/we are attached to. The difference is that at some stage you exported stats. Once that happened, the cat was out of the bag :) Don't know why you started exporting them - Maybe the feature got turned on during a server upgrade or something.

As for finding out about projects, you'll find putting a key phrase from the default text on the front page of your or pretty much any project in Google will find loads of BOINC projects that aren't listed. Also, word of mouth is very important among crunchers. Another source of info is from the WUProp project. That will list all projects that have been run and had stats gathered by it.

FWIW, my view is that stats are exported they are free to be used by anyone. Doesn't take much effort to look in the stats directory and see if there are any up to date tables in there; If there are, use them :) I know some people (OK, 1 person) who disagrees, but it takes all sorts ;)

How to avoid it happening in the future? Make the project invitation code only. Although I think BWT's get invite codes by deafult? Also, don't ever export stats. MOPAC seems to have kept user interest minimal by never exporting stats.

Oh, and thank for starting up the stats export again - Much appreciated!

Al.
4) Message boards : Wish List : Low credit for such a risky project (Message 111326)
Posted 20 Nov 2011 by Profile ChertseyAl
Post:
That's all fair enough, and you've certainly made it clear in enough places regarding credit :)

But just a couple of points:

The hardcore milestone collectors will crunch *any* project that's available to them (that includes me BTW, 110 projects and rising). I took the risk, got some credit, quite happy despite losing loads of pending credit a while ago.

Your user base was inherited from Einstein, so you automatically got a whole bunch of people who were bound to be interested. Maybe if you'd started out as 'invite code only' ... But I can see that would have taken more time/effort than this approach.

As for turning off stats export, how about either leaving it running until a clearly defined date, or run it just once more at some time in the future. Maybe set the cutoff time to the dealine of the WUs (2 weeks IIRC?). And announce it in big red capitals on the front page ;) Perhaps you could email it to everyone, but I don't know how much effort that would take.

Personally, I'm not bothered whatever happens. But I can understand the angst of others :)

Cheers,

Al.
5) Message boards : Problems and Bug Reports : Gravitational Wave S6 GC search v1.01 (SSE2) Error Messages (Message 111187)
Posted 29 Oct 2011 by Profile ChertseyAl
Post:
Still no new work here :(

Al.
6) Message boards : Problems and Bug Reports : Gravitational Wave S6 GC search v1.01 (SSE2) Error Messages (Message 111184)
Posted 28 Oct 2011 by Profile ChertseyAl
Post:
I sometimes see strange behaviour when trying to get work.

28/10/11 11:27:02|Albert@Home|Sending scheduler request: To fetch work. Requesting 3503 seconds of work, reporting 0 completed tasks
28/10/11 11:27:08|Albert@Home|Scheduler request succeeded: got 0 new tasks
28/10/11 11:27:08|Albert@Home|Message from server: No tasks sent
28/10/11 11:27:08|Albert@Home|Message from server: No tasks are available for Gravitational Wave S6 GC search
28/10/11 11:27:08|Albert@Home|Message from server: see scheduler log messages on http://albert.phys.uwm.edu//host_sched_logs/1/1020
28/10/11 11:27:08|Albert@Home|Message from server: No tasks are available for the applications you have selected.


This is just one host, another 7 are doing the same even when asking for many more seconds of work. According the the server status page there is plenty of work available, but I don't see the number of unsent WUs going down.

The log referred to says (trimmed):

...
2011-10-28 10:27:03.7620 [PID=28680] [send] work_req_seconds: 3503.08 secs
2011-10-28 10:27:03.7620 [PID=28680] [send] available disk 15.64 GB, work_buf_min 86400
2011-10-28 10:27:03.7620 [PID=28680] [send] active_frac 0.999925 on_frac 0.999697
2011-10-28 10:27:03.7625 [PID=28680] [send] [AV#364] not reliable; cons valid 1 < 10
2011-10-28 10:27:03.7625 [PID=28680] [send] set_trust: cons valid 1 < 10, don't use single replication
2011-10-28 10:27:03.7625 [PID=28680] [send] [AV#413] not reliable; cons valid 1 < 10
2011-10-28 10:27:03.7625 [PID=28680] [send] set_trust: cons valid 1 < 10, don't use single replication
2011-10-28 10:27:03.7625 [PID=28680] [send] [AV#432] not reliable; cons valid 0 < 10
2011-10-28 10:27:03.7625 [PID=28680] [send] set_trust: cons valid 0 < 10, don't use single replication
2011-10-28 10:27:03.7670 [PID=28680] [version] looking for version of einsteinbinary_BRP4
2011-10-28 10:27:03.7671 [PID=28680] [version] Checking plan class 'BRP3SSE'
2011-10-28 10:27:03.7671 [PID=28680] [version] reading plan classes from file '../plan_class_spec.xml'
2011-10-28 10:27:03.7676 [PID=28680] [version] parsed project prefs setting 'also_run_cpu' : true : 0.000000
2011-10-28 10:27:03.7677 [PID=28680] [version] Checking plan class 'ATIOpenCL'
2011-10-28 10:27:03.7677 [PID=28680] [version] No ATI devices found
2011-10-28 10:27:03.7677 [PID=28680] [version] [AV#443] app_plan() returned false
2011-10-28 10:27:03.7677 [PID=28680] [version] Checking plan class 'NVOpenCL'
2011-10-28 10:27:03.7677 [PID=28680] [version] No NVidia devices found
2011-10-28 10:27:03.7677 [PID=28680] [version] [AV#440] app_plan() returned false
2011-10-28 10:27:03.7677 [PID=28680] [version] [AV#432] (BRP3SSE) using unscaled projected flops: 1.49G
2011-10-28 10:27:03.7677 [PID=28680] [version] Best version of app einsteinbinary_BRP4 is [AV#432] (1.49 GFLOPS)
2011-10-28 10:27:03.7678 [PID=28680] [version] returning cached version: [AV#432]
...
2011-10-28 10:27:03.7700 [PID=28680] [version] returning cached version: [AV#432]
2011-10-28 10:27:03.7720 [PID=28680] [user_messages] [HOST#1020] MSG(low) No tasks sent
2011-10-28 10:27:03.7720 [PID=28680] [user_messages] [HOST#1020] MSG(low) No tasks are available for Gravitational Wave S6 GC search
2011-10-28 10:27:03.7720 [PID=28680] [user_messages] [HOST#1020] MSG(notice) see scheduler log messages on http://albert.phys.uwm.edu//host_sched_logs/1/1020
2011-10-28 10:27:03.7720 [PID=28680] [user_messages] [HOST#1020] MSG(low) No tasks are available for the applications you have selected.
2011-10-28 10:27:03.7721 [PID=28680] Sending reply to [HOST#1020]: 0 results, delay req 60.00
2011-10-28 10:27:03.7723 [PID=28680] Scheduler ran 0.022 seconds


Same thing happened last week, but it fixed itself before I had chance to report it.

Al.






This material is based upon work supported by the National Science Foundation (NSF) under Grant PHY-0555655 and by the Max Planck Gesellschaft (MPG). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the investigators and do not necessarily reflect the views of the NSF or the MPG.

Copyright © 2024 Bruce Allen for the LIGO Scientific Collaboration