Gun Shy
28 Jun 2005 2:44:03 UTC
Topic 189432
(moderation:
Got the "latest" updates from E@H this evening including the looping download. Killed the app, rebooted, and tried again. Had one rig denied additional work as the daily quota of 1 was exceeded. Had a second rig go "daily quota of 3" was exceeded. These are prime crunchers. None had any quotas before this junk was received from the "server" Excuse me, but what is the gain for the project to hose two dedicated long time affiliates? I actually lost work in the process. Smell fish?
Gun Shy
)
There has been a daily quota for quite a while. You just have not been hitting it. The daily quota is there to protect the server from hosts that download and error out every WU very quickly. Each error or late result reduces your daily quota by one. Each successful result doubles your daily quota (up to some max).
Now the question is why you have had so many failures recently that your quota dropped to one.
BOINC WIKI
There is a very good reason.
)
There is a very good reason. Developer error. Look at the top thread in this list and learn all about it... It's called "New units not downloading". I've already had more than 10 boxes affected by it but the fix is in sight.
Cheers,
Gary.
John; I am having a lot of
)
John; I am having a lot of grief with your explanation. These rigs are not "erroring out" on anything but the garbage from the server. The two boxes I mention uploaded completed work units with positive results prior to downloading the "screaming looper". I have at least one additional rig with this evil download still sitting in the tranfers window. All but one of my boxes is still on 4.19. I had more pending work on these rigs before I foolishly rebooted and retried the download. Is there a cogent explanation as to how a rig can go from cruising day to day to a sudden quota of 1? Gary has a better handle on this than the developers. Are you comfortable with the acronym "CF"? It would appear that I am not alone.
Hey Tweakster, If you have
)
Hey Tweakster,
If you have any h1_nnnn (note the small h) style work units on any of your boxes then you'll have to get rid of them ASAP as you'll be wasting cycles. I'm using 4.19 as it works well for me (mostly :).) and it's a bit of a chore to get rid of the cancelled work, as there is no abort button.
See Bruce's posts and my questions over in the main thread on this and I've just posted there what I did to get things moving again. Hope this helps. The good thing is that Bruce has reset to 8 the daily limit and this has worked for me as all my boxes that were suffering from this are now back at 8 after being much lower.
Cheers,
Gary.
RE: Hey Tweakster, If you
)
Note: I'll continue during the next few days to reset the daily limits by hand for hosts that got h1_ workunits.
Director, Einstein@Home
Bruce; Here's the cleanup
)
Bruce; Here's the cleanup report: Box 202106 went daily quota of one at 5pm CDT today, suffering from h1_0078. It has 6+ WU's left. I think I was able to delete the file, but the server hasn't got a clue how to update it to health. Box 98411 suffers from h1_0101.5 and h1_0417.0, I was able to delete these nasties, but the WU's remain. Box 211430 is infected with h1_0441.5 and h1_0073.5. Box 102573, running version 4.45, carries two WU's from h1_0443 which I have set to aborted. Box 61409 suffers with h1_0430.0. I deleted the file but the 4 WU's remain in my que. Bruce, I am giving the project a full pull, stay away from the console and educate the freshman volunteers. It's tough managing a dozen+ machines on a diet of whoopsies.
Regards-tweakster
RE: Bruce; Here's the
)
I've again reset the quotas for all machines that got h1_ workunits. I'll do this at least a few times per day over the next few days.
Bruce
Director, Einstein@Home