Results ready to send 10. Need to make a few more work units please.

Thanks.

the problem is people are grabbing WUs faster than they can be put out there :P

i'm trying to clean up the database while letting it run so you guys can still get work. once it gets cleaned again, i'm going to keep the purge running as a daemon, so they DB will be working as fast as possible so results will get out to you all as quickly as possible. the only side-effect of this is that WUs will be removed from the DB almost as fast as they're reported... hopefully that won't be too big of a problem.

---

Dont let u to excite about some Guys getting bad if they Machine run out of Work Travis...
You Guys shud keep in Mind that Travis spend many Time too and its sure not his Job, so that he can't keep all the Time a eye on this.

It's seems working Good atm, im Wish you good luck Travis to fix this Problems!

Greetings

It happens yet again !

---

It happens yet again !

The assimilator is still generating work, are you guys all having trouble getting WUs?

---

The work will come back eventually.

Cheers,

Mike

---

[http://milkyway.cs.rpi.edu/milkyway/team_display.php?teamid=15]Click here to join the #1 Aussie Alliance on Milky Way.[/url]

G'day Travis,
There's a little trouble. When I click update sometimes it takes a while to respond and has trouble finding the server but doesn't time out. At the moment the response is:

24/06/2008 5:44:03 PM|Milkyway@home|Sending scheduler request: To fetch work. Requesting 81968 seconds of work, reporting 0 completed tasks
24/06/2008 5:44:08 PM|Milkyway@home|Scheduler request succeeded: got 0 new tasks

Cheers,

Mike

---

[http://milkyway.cs.rpi.edu/milkyway/team_display.php?teamid=15]Click here to join the #1 Aussie Alliance on Milky Way.[/url]

No work coming through and the pending queue is building up again

It happens yet again !

The assimilator is still generating work, are you guys all having trouble getting WUs?

Yessir. Similar messages as shown below (got 0 new tasks). Validator is having trouble keeping up/catching up as well.

---

Join the #1 Aussie Alliance on MilkyWay!

Just downloaded a WU as I was reading this thread.

Edit: and on another host got thi:
2008/06/24 11:41:04|Milkyway@home|Sending scheduler request: To fetch work. Requesting 224 seconds of work, reporting 1 completed tasks
2008/06/24 11:41:24|Milkyway@home|Scheduler request succeeded: got 0 new tasks

Oh well...

Edit2: And a third host just got one! Proof of the Uncertainty Principle, perhaps? ;)

Rod

Just downloaded a WU as I was reading this thread.

Edit: and on another host got thi:
2008/06/24 11:41:04|Milkyway@home|Sending scheduler request: To fetch work. Requesting 224 seconds of work, reporting 1 completed tasks
2008/06/24 11:41:24|Milkyway@home|Scheduler request succeeded: got 0 new tasks

Oh well...

Edit2: And a third host just got one! Proof of the Uncertainty Principle, perhaps? ;)

Rod

ok, it looks like my test of having the db_purge wait a day to delete WUs from the database wasn't enough. I'm gonna set it back to purging results as soon as they come back, and hopefully it will go back to running smoothly. Since the purge will have a bit of catching up to do, things might go a bit slow for a little bit, but hopefully will pick up when it's done.

---

ok, it looks like my test of having the db_purge wait a day to delete WUs from the database wasn't enough. I'm gonna set it back to purging results as soon as they come back, and hopefully it will go back to running smoothly. Since the purge will have a bit of catching up to do, things might go a bit slow for a little bit, but hopefully will pick up when it's done.

Hmmm...

I saw your comment earlier on the state of official BOINC documentation, and I can't say I really disagree with that opinion. ;-)

One thing which comes up over at the SAH boards fairly frequently for low count posters and newbies is why their results disappeared from their account listings, with the mistaken impression they may have lost credit somehow because of the purge. Given that the vast majority of people never even bother to ask a question when a BOINC project does something which seems strange to them, they just quit, this might be limiting your appeal to a larger range of participants.

I also saw that setting the purge daemon at 1 day wasn't cutting it when it came to backend performance, but zero is is not a very good solution either! I looked at the options for the daemon, and was wondering if you had tried setting it to a fractional day setting. It didn't say anywhere the value had to be an integer (but I wouldn't be surprised if that was the case). It would seem there should be some 'happy value' which would let the results hang around for more than a few seconds without causing the DB to grind to a halt eventually.

Of course, I'm assuming you're not just way underpowered for the host population you have running at the moment!

Alinator

ok, it looks like my test of having the db_purge wait a day to delete WUs from the database wasn't enough. I'm gonna set it back to purging results as soon as they come back, and hopefully it will go back to running smoothly. Since the purge will have a bit of catching up to do, things might go a bit slow for a little bit, but hopefully will pick up when it's done.

Hmmm...

I saw your comment earlier on the state of official BOINC documentation, and I can't say I really disagree with that opinion. ;-)

One thing which comes up over at the SAH boards fairly frequently for low count posters and newbies is why their results disappeared from their account listings, with the mistaken impression they may have lost credit somehow because of the purge. Given that the vast majority of people never even bother to ask a question when a BOINC project does something which seems strange to them, they just quit, this might be limiting your appeal to a larger range of participants.

I also saw that setting the purge daemon at 1 day wasn't cutting it when it came to backend performance, but zero is is not a very good solution either! I looked at the options for the daemon, and was wondering if you had tried setting it to a fractional day setting. It didn't say anywhere the value had to be an integer (but I wouldn't be surprised if that was the case). It would seem there should be some 'happy value' which would let the results hang around for more than a few seconds without causing the DB to grind to a halt eventually.

Of course, I'm assuming you're not just way underpowered for the host population you have running at the moment!

Alinator

I actually looked in the code to see if it could not be an integer... unfortunately it is. I think i'm going to rewrite it and recompile to have a better value (maybe 6 hours or so).

---

I actually looked in the code to see if it could not be an integer... unfortunately it is. I think i'm going to rewrite it and recompile to have a better value (maybe 6 hours or so).

LOL...

I'll bet that revelation was about as welcome a root canal! :-D

Just what you need right, more coding practice on something which really doesn't have anything to do with modeling the Milky Way Galaxy. ;-)

Alinator

As of about 09:00 UTC we seem to have the problem again i.e. no wu being downloaded and the validation queue growing.

As of about 09:00 UTC we seem to have the problem again i.e. no wu being downloaded and the validation queue growing.

looks like the validation queue is going back down. work is still being generated so i think once it gets through the validation queue it should be ok again. there might have been some kind of scheduled process that slowed it down, a backup or something.

---

As of about 09:00 UTC we seem to have the problem again i.e. no wu being downloaded and the validation queue growing.

looks like the validation queue is going back down. work is still being generated so i think once it gets through the validation queue it should be ok again. there might have been some kind of scheduled process that slowed it down, a backup or something.

yep, looks like it caught back up and is running normally again.

---

yep, looks like it caught back up and is running normally again.

Yep, must be. No ones posting. ;-)

---

[http://milkyway.cs.rpi.edu/milkyway/team_display.php?teamid=15]Click here to join the #1 Aussie Alliance on Milky Way.[/url]

Not posting because
Sat Jun 28 08:28:21 2008|Milkyway@home|[error] Error on file upload: can't write file /export/www/boinc/milkyway/upload/a/gs_607_1214630507_470607_0_0: No space left on server

But all seems as it has gone back to normal by now.

---

They just need to Flush the Toilet more so the Project runs smoother ... :)

They just need to Flush the Toilet more so the Project runs smoother ... :)

Well put!