Asteroids@home - new to me! Whose crunching it?

[networkman]

^^^ Thank you!

As for WU cache, my settings on my primary rig are under Tools --> Computing Preferences --> Network:

Min work buffer 3.00 days Max additional buffer 3.00 days

So max should be 6.00 days.

Actual WUs right now in cache are 208.

So 208 / 4 CPUs = 52 WUs per CPU @ 2.5 hrs per proc = 130 hours of total time per CPU / 24(hrs/day) = 5.4166 days of work per CPU in cache.

It seems reasonably accurate for me. Maybe all your are base WUs belong to me?

 

[Assimilator1]

Lol, not heard that in a while!

Well I've set mine to 3+2 days & currently have 67 WUs 'in progress', 3 d/l failed, 8 finished, 4 crunching, so that's 52 cached, way under 10hrs worth

Is their a file stored within BOINC that could affect cache levels? I'm just wondering if a file got left over from an old version.......

 

[Sunny129]

my cache is set to 2 days, and i have 310 WU's in progress at the moment. i can't recall whether or not my cache has been set to 2 days for a while now, or if i just lowered it from 3 days...as a consequence, i don't know if my work buffer is currently settled in around the ~310 task mark, or if its on the decline to plateau at a lower number. that said, this is for an 8-threaded 3770K CPU that crunches a WU in 26 minutes per thread. i didn't do the math to see if my "tasks in progress" quantity is in line with my work buffer settings, but if the numbers seem off, its probably b/c CPU resources have to be balanced between 2 other CPU projects as well (LHC@Home and Test4Theory@Home).

even if the number fof tasks in progress doesn't jive with my work buffer settings, the point is that i'm able to download plenty of A@H work just as NWM is able to...which makes me think that it might be a client issue on your end. i'm sure there is an .xml file in the BOINC data directory that contains a variable that helps correct the amount of work a host gets, though i have little idea which one it may be...my guess would be the client_state.xml file, but who knows. you ought to start a thread on the general BOINC message boards, b/c the developers and testers over there will be able to point you in the right direction.

 

[networkman]

Boinc version? My BOINC Manager say 7.0.64(x86)

Also of note, all of my rigs are set for a single project - nothing is splitting time with another project.

 

[Assimilator1]

Boinc version? My BOINC Manager say 7.0.64(x86)

Also of note, all of my rigs are set for a single project - nothing is splitting time with another project.

Same here.

Sunny
Cheers, will look into it

 

[Sunny129]

well i woke up this morning to a bunch of pending A@H uploads, which hopefully means that the server migration is currently under way as planned...

 

[petrusbroder]

It is ... I have been able to upload (but not to report) quite a few WUs.

 

[Sunny129]

just noticed the same thing...all those pending uploads i had before are now ready to report, but are not reporting...no worries though, since this is "scheduled" maintenance...

 

[Assimilator1]

Yea except I haven't had the chance to resolve this damn cache issue & my client is crunching its last WU

 

[Assimilator1]

Anyone know when the servers are back up?

 

[Sunny129]

apparently not soon enough...it was silly of me to think that what was supposed to only be a 6-8 hour server migration would only get stretched into a 2-3 day ordeal. my host is crunching its last 7 A@H tasks before it runs out, and god only knows when the server will be up and running again. as it is with so many projects, we unfortunately cannot get a status update b/c both the server status page and the message boards are all on the same server, and so they go down whenever the project server goes down.

 

[Assimilator1]

apparently not soon enough...it was silly of me to think that what was supposed to only be a 6-8 hour server migration would only get stretched into a 2-3 day ordeal.

Particularly over a weekend .

I've switched back to F@H for now.

 

[networkman]

...it was silly of me to think that what was supposed to only be a 6-8 hour server migration would only get stretched into a 2-3 day ordeal.

Welcome to the similar experience of playing an online MMO like Dungeons & Dragons Online! I kinda figured the developer of this project was rather like the devs at Turbine and thus even a planned outage of several hours would morph into at least a few days. :\

Looks like I've got about a days worth of WUs left before the rigs go idle.

 

[Sunny129]

i don't think Asteroids@Home is a university-funded project...in fact i'm pretty sure the project server is just a machine at the head developer's house, hence the frequent server outages LOL...

 

[networkman]

I'm quite sure you're right about this project. I was only referring to the experience of the actual length of the outage compared to the projected length.

Here at my work, I don't get much leeway in projected lengths of outage - it's either working in its upgraded state in time for when we open to the public, or it get's restored to pre-upgrade and we re-schedule.

As a result, a LOT more planning goes into scheduled down-time.

 

[Assimilator1]

Any news as to when it'll be back online?

 

[networkman]

Nada so far.

Also not sure if the dev will know to address the issue of work units hitting their deadline; I have myriad work units with deadlines of 9/23 and 9/24 that are obviously past.

 

[Sunny129]

i was wondering about deadlines too...fortunately the oldest completed tasks on my host aren't due until 9/26. all the the tasks due on 9/26 and 9/27, and some of the tasks due on 9/28, have already been uploaded and are "ready to report"...the rest are still trying to upload. i wonder if tasks that have been uploaded, but not yet reported, will fail to earn credit if not returned by the deadline?

 

[Assimilator1]

Hmm, don't know TBH.

I googled 'A@H server........ ' & this thread was the 2nd search result, lol.

 

[Sunny129]

Server migration finished
Migration of the server is finished. I am very sorry it took so long but there were more problems during the migration than I expected. Server should now be stable and I hope that there will be minimal failures in the near future.
I also changed deadline of all sent results to 15.10.2013 so you shouldn't lost any of your work done by the migration.

25 Sep 2013, 21:09:52 UTC

uploads seem sluggish as hell, but i guess that's to be expected after an outage like this b/c everyone has loads of tasks to either upload or report...hopefully that network bottleneck will disappear soon. tasks that are "ready to report" won't report, and the associated error i see in the BOINC event log is " server can't open database."

 

[Sunny129]

tasks are now uploading, reporting, and downloading without a hitch...and the new downloads are fast and error-free! well, perhaps i should watch a few more rounds of downloads before i conclude that the download errors are gone for good...

 

[GLeeM]

They uploaded and reported OK, but more are ending up as Validate Error then good?

 

[Sunny129]

i've got validate errors too. for me its more like a 33% error rate (at least with the last batch of reported tasks). there's a thread on the A@H forums about the project validator that seems to address both inconclusive validations and validate errors...Kyong's working on finding and fixing the problem, but it hasn't been solved yet...

 

[Assimilator1]

Woohoo it's back , hope they get the validator sorted soon.

 

[Sunny129]

the project is currently out of work. hopefully that will get resolved by Monday...