Anyone else finding the forums & web site going down for a few minutes at a time over the last few hours?

If it stays up long enough to post this reply.. YES!!!

It's been doing it since very early this morning...

---

"Time is simply the mechanism that keeps everything from happening all at once."

Nice to know it's not just me.

---

Grant
Darwin NT

Well looks like another problem rears it's head again.
My main rig was starting to do ok on the quota of 20, but will run out of work soon.
The reason Downloads not Happening.
Well happy 4th July.


Dave

---

Well looks like another problem rears it's head again.
My main rig was starting to do ok on the quota of 20, but will run out of work soon.
The reason Downloads not Happening.
Well happy 4th July.


Dave

Same problem. Project backoff due to constant HTTP errors.

Seems the servers are screwed after the crash.

---

- Jarryd

Same problem. Project backoff due to constant HTTP errors.

Seems the servers are screwed after the crash.

Network traffic graphs show the pipe is full, downloads occuring as fast as they are requested, hence the HTTP errors. When the traffic drops off, the errors will too.

---

Grant
Darwin NT

I hope this is a silly question...
Are the inter-computer communications within SETI@home riding on the same
network hub as the internet connection?

---

Janice

Happy July 4th all have a good one ;-)

---

I'm uploading fine. Just downloads are failing.

---

- Jarryd

If your tasks are all VLAR's then you will need to ride it out.
If you have a mix then Reschedule some of them to the GPU's.

Dave

---

For all the Americans out there - Happy Independance Day Guys

Am I reading the news post right?
Are the servers really only going to allow 20 tasks per PC per day, or have I read it wrong?
Im curious as one of my hosts hasn't been able to get any work since the outage and is currently crunching milkyway until it can get work...

Could do with an explanation as Im not that clear on how it works?

---

For all the Americans out there - Happy Independance Day Guys

Am I reading the news post right?
Are the servers really only going to allow 20 tasks per PC per day, or have I read it wrong?
Im curious as one of my hosts hasn't been able to get any work since the outage and is currently crunching milkyway until it can get work...

Could do with an explanation as Im not that clear on how it works?

Happy 4th July to you from Australia. :)

Back to the task at hand, after the last 3 day outage we had, a 20 WU limit was put in place to reduce the stress on the servers by specific hosts, and allowing more than just the big time crunchers access to work units, thus also keeping work units on the server ready to send out rather than dumping them all at once, meaning big time crunchers with 10 day caches don't get all the work and part time crunchers that have a half day cache can get work throughout the week when their machines need it.

The problem with this is some machines are getting CPU work only, and some are getting Astropulse only. Not to mention that the big time crunchers will not be able to keep their machines at 100% load through the next outage.

20 Work units on my basic home machine runs at 4 CPU WU's at a time, each taking about 1hr 30 mins. 16 GPU tasks taking about 10 minutes each, that's pretty much 20 work units in just under 2 hours.

Now there is a news post saying they're going to remove the limit Monday morning Berkeley time to allow big crunchers to fill up caches for the outage.

We'll see what happens though, i doubt all the big crunchers will get work.

---

- Jarryd

VLAR's will cause a huge problem for GPU crunchers mine included.
As a task finishes and reports if there are large amounts of VLAR's they will be sent.
If you are not using VLAR_KILL all your work will be CPU based.
Thus GPU's go cold.
There is a reverse to this of course, if there is a shorty storm CPU's will go cold.
I would prefer the second option.

Dave

---

Perhaps quota system is working WU's in the field now down to 3,258,508.

I have not seen it this low for a while, perhaps a few of the pendings are begining to validate at last.

Michael

I would say it's because nobody can refill his Cache. Every personally
Cache is now overwritten by the Server rules to 20 Workunits per Host.
So, the amount of Workunits out in the Fields will be reduced.

Helli

Well, looks like roughly a 12k/h deficit.
I find those spikes in up/downloads every 4 hours interesting.

Saying nobody can fill his cache is unfair to some 140,000 users with a RAC of 500 or less, to whom 20 WU would be enough for at least a 4 day cache.
That said the 27894 users with an RAC above 500 (thanks Robert Ribbeck, Scarecrow et.al.) are way below capacity and requirement.
So, an estimated 80% of active users got their piece of cake and will be happy - how about starting to give second helplings of the remaining cake?
At least that cake doesn't rot...

Seeing as it's Sunday and Monday a holiday in the US it looks highly unlikely to be enough if performing outage as scheduled.

Wonder if the electricity companies see a dip in power consumption?

I hope patience does not run out as fast as WUs... We'll need a lot more over the coming weeks.

---

Carola
-------
I'm multilingual - I can misunderstand people in several languages!

For those interested - my AP only cruncher has 37 tasks in progress right now so the limit doesn't seem to apply to AP.
http://setiathome.berkeley.edu/results.php?hostid=5097356&offset=0&show_names=0&state=1

---

yupp, but these 27,000 Users neets - i assume - 80% of all Workunits available.

Helli

No problems with downloads here. I turn in a WU, and I get one back right away.

The same with me. That is fine while the download servers are turned on, but at this rate I expect to enter the next 3 day outrage with about 4 hours of cache. And I increased my cache setting to 5 days, from 3.

Question: the effects of all this on us users is obvious, but what is the effect on the project? Compared to a (recent) typical week of unplanned outages, is S@H processing more work, less work, or about the same? And how does the processing speed compare to the flow of new work from the telescope?

---

yupp, but these 27,000 Users neets - i assume - 80% of all Workunits available.

Helli

From the RAC/users chart with added daily credits they need about 60%
EDIT: or perhaps 52% if the numbers are calculated differently *sigh*

---

Carola
-------
I'm multilingual - I can misunderstand people in several languages!

Question: the effects of all this on us users is obvious, but what is the effect on the project? Compared to a (recent) typical week of unplanned outages, is S@H processing more work, less work, or about the same? And how does the processing speed compare to the flow of new work from the telescope?

It will probably take a few weeks of this new cycle and the subsequent 'fine tuning' to assess the impact on Seti workflow of folks like me who are going to stick around but may not be able to get enough work to continue to crunch 24/7, and those who choose to leave for their own personal reasons.

---

"Time is simply the mechanism that keeps everything from happening all at once."