> Matt Lebofsky on 2 Feb 2005 23:29:55 UTC
> The server status page is accurate at this point (it needed a kick today after
> stopping during the failed db migration on Monday). That said, it is true that
> we're down to a handful of workunits to send out. Not exactly sure why this is
> the case - we should have plenty in the queue. At any rate, I got two
> splitters working now to keep up with current demand and hopefully build a
> backlog over time.
> - Matt

Matt the Server status page can NOT be accurate at this point!

During the last 2 months we have been following the progress of the 2 largest teams (SETI Germany and OcUK) in the following thread.

The Cobblebeat Graph shows the daily returned cobblestones for both teams. Thus this graph shows you directly the Validator's activity. You can see normal activity at the beginning of January (in average 180,000 cobblestones returned daily by each team). After the 10th the Validator’s problem starts and you can see it never completely recovered.

Before the attempted migration 31-Jan there were about 150,000 results waiting for validation. According to the graph these are still outstanding. The question is: Where are these 150,000 results? Maybe you forgot about them after the failed migration?

This is not a complaint just double checking and reporting back when the sums do not add up.

Kind regards
Purdy & Poppy

Okay.. you're right. I should have said: "the numbers on the status page that *I* deal with are accurate." I know the results "ready to send" number is correct. I've never really been in the development loop regarding validation/credit but my current understanding is:

1. When that "waiting for validation" number is 0, everybody who deserves credit should have gotten their credit.

2. That number is currently messed up and we're not sure why. Not my department. Other people are trying to sort this out.

Of course, as of right now these numbers are all "0" but that's because we just had a massive server outage and all the stats collectors are getting back up to speed.

- Matt

---

-- BOINC/SETI@home network/web/science/development person
-- "Any idiot can have a good idea. What is hard is to do it." - Jeanne-Claude

Thanks for the info, Matt.

Did this outage identify and clear up the source of the mysterious database slowdown? Are we ready for another server migration attempt now?

---

This outage didn't clear anything up. It had nothing to do with currently known problems. If anything it made things worse. More in a tech news items shortly..

- Matt

---

-- BOINC/SETI@home network/web/science/development person
-- "Any idiot can have a good idea. What is hard is to do it." - Jeanne-Claude

Feeling like one of many crack addicts (basing this, of course, on popular, factual TV programming) effected by a DEA drug bust who's seriously Jones'in...

I need a fix of credits man... :D

Kidding of course.

This has given me a chance to get caught up on my Cancer research cycles.

---

> This outage didn't clear anything up. It had nothing to do with currently
> known problems. If anything it made things worse. More in a tech news items
> shortly..
>
> - Matt
>

Hi Matt,

I'm no database guy, but I'm wondering if MySQL is up to the task with such a huge database.

Did you evaluate PostgreSQL?

Regards Hans

---

> Did you evaluate PostgreSQL?

Given that the original Postgre was developed at Berkeley, I'm sure they did. ;)

I'm not sure that the size of the database is the problem. MySQL is handling some of the largest databases in the world. No, I think it's more a question of the frequency of accesses. MySQL is known to be fast. PostgreSQL and most others are more complex and slower. Having got this far, a change now would be a step into the unknown.

Hopefully, the new hardware will have a positive effect - assuming the team is actually able to complete the migration. Crossing fingers and everything else crossable. ;)

---

Be lucky

Neil

> > Did you evaluate PostgreSQL?
>
> Given that the original Postgre was developed at Berkeley, I'm sure they did.
> ;)
>

D'Oh :o)

> I'm not sure that the size of the database is the problem. MySQL is handling
> some of the largest databases in the world. No, I think it's more a question
> of the frequency of accesses. MySQL is known to be fast. PostgreSQL and most
> others are more complex and slower. Having got this far, a change now would be
> a step into the unknown.

True.


Regards Hans

---

In support of the original question: where are the 150,000 results waiting for validation?

Some simple math based only on my account: going back to 28 Jan 2005 I have 153 results listed as "success / over / pending" i.e. finished WUs. I have looked in detail at the WUs for three days 28 Jan - 30 Jan. Of the 44 WUs on these three days 23 have either 3 or 4 "success" results and ought to be in the validation queue.

Based on just those WUs and my percentage contribution to WUs crunched which is between 0.01% and 0.02% according to BOINCSTATS there should way way way more than 115,000 [23 * 5,000] WUs ready for validation. More like 300,000+ probably.

---

Now after the outage of Feb 3rd, we see that the results to be validated has once again grown to 42,000+ and it looks like maybe a 2nd Validator computer might be considered. I base this on the fact that the 1 validator computer does not seem to be able to keep pace with the returned results. What will happen when all the classic users migrate over here?

---

> Now after the outage of Feb 3rd, we see that the results to be validated has
> once again grown to 42,000+ and it looks like maybe a 2nd Validator computer
> might be considered. I base this on the fact that the 1 validator computer
> does not seem to be able to keep pace with the returned results. What will
> happen when all the classic users migrate over here?
>

Considering it was well over 100K, if this is all that is left, it is going to be close. I'm not sure what they will do when the lights go out for Classic users, but some of the equipment (IIRC this was mentioned more than a month ago) was slated to be switched over at the same time.

I don't exactly know the inner workings of this "cluster", but if the active DB unit is having bandwidth (aka I/O) issues that will slow everything accessing the DB down. It might be that the validator isn't the bottleneck. Hope the SETI team is keeping good notes, this whole little drama will make an interesting chapter in the history of SETI someday.

---

Still looking for something profound or inspirational to place here.

Before the I/O problem crept in, the validation process was pretty quick. Significant backlogs were fairly rare and usually attributable to a specific cause (i.e.: that server was brought down). You were almost always waiting on other crunchers to meet the quorum than for the validator to catch up. Plans to accommodate higher activity from the extra Classic users were meant to be dealt with by a hardware upgrade that’s still in the works. We’ve been plagued by an unknown performance hit to the system for weeks now that has to be resolved first, as the project team has already mentioned. Once that’s taken care of, we should have plenty of performance overhead to accommodate extra users and still have a responsive crediting process.

---

> In support of the original question: where are the 150,000 results waiting for
> validation?
>
> Some simple math based only on my account: going back to 28 Jan 2005 I have
> 153 results listed as "success / over / pending" i.e. finished WUs. I have
> looked in detail at the WUs for three days 28 Jan - 30 Jan. Of the 44 WUs on
> these three days 23 have either 3 or 4 "success" results and ought to be in
> the validation queue.
>
> Based on just those WUs and my percentage contribution to WUs crunched which
> is between 0.01% and 0.02% according to BOINCSTATS there should way way way
> more than 115,000 [23 * 5,000] WUs ready for validation. More like 300,000+
> probably.
>
Yes we agree with your calculation. We still don't have an official answer to the original question . . . Where are the 300,000 results?

Now we have another 250,000 results in the Validator's queue. Thus we have at this point in time a backlog of more than 1/2 million work units (each with 3 or more results) ready for validation.

---

Hi,
I'm new in this project, so I don't understand much of these calculations of credits. I have one question - I have done some packages (units)since I become a member of a SETI team, 27 so far,I think. But my credit is still zero. How much work I have to do to put some credit beside my name?

I'm also a member of ProteinPredictor@Home and I already have some credit on it. So I don't understand. Is there some problem with SETI project, or I did something wrong with my account?

Can anyone give me some explanation on this?

Thanks.

---

<img src="http://petrus.homeftp.org/bws/counter_big.php?id=7828882" /> <img src="http://boinc.mundayweb.com/one/stats.php/userID:553/prj:2/trans:off/.png" />

> Hi,
> I'm new in this project, so I don't understand much of these calculations of
> credits. I have one question - I have done some packages (units)since I become
> a member of a SETI team, 27 so far,I think. But my credit is still zero. How
> much work I have to do to put some credit beside my name?

The Boinc-Seti validates your results before giving the credits (so they know the result is a real, not a cheated one).

Currently the validator has falled behind as you can read from the other posts. I personally have over 40 results vaiting for validation.

What you need to do is wait, hopefully the validator will give you the results in the coming days. From "your account" at the Boinc-Seti webpage you can view your results. If you have there lots of zero granted credits then you might have problem with your computer (maybe its making computing errors). But currently as the validator is not up to date, you should have "pending" granted credits.

So, check that you are having pending credits and then just wait. Validator will catch up...

---

Look, that's what this tread is all about. The validation of credits is very slow at the moment. For the past few weeks we've all been waiting for our credits. We just have to wait, don't worry, you'll get your credits.

---

<img src="http://www.boincstats.com/stats/banner.php?cpid=d2319b8f0ad14565556d0ba45b64e779">

> If you have there lots of zero granted credits then you might
> have problem with your computer (maybe its making computing errors). But
> currently as the validator is not up to date, you should have "pending"
> granted credits.
>
> So, check that you are having pending credits and then just wait. Validator
> will catch up...

All of my granted credits are "pending". So, it looks like I have to wait.
Thanks a lot. I feel better now. :)

---

<img src="http://petrus.homeftp.org/bws/counter_big.php?id=7828882" /> <img src="http://boinc.mundayweb.com/one/stats.php/userID:553/prj:2/trans:off/.png" />

> Look, that's what this tread is all about. The validation of credits is very
> slow at the moment. For the past few weeks we've all been waiting for our
> credits. We just have to wait, don't worry, you'll get your credits.
>
Yes, Toni just explained to me. As I said, I'm a member since last week. So, there are some tehnical stuff I have to learn and understand how the system works. But one step at the time.
Thanks for your anwser.

---

<img src="http://petrus.homeftp.org/bws/counter_big.php?id=7828882" /> <img src="http://boinc.mundayweb.com/one/stats.php/userID:553/prj:2/trans:off/.png" />

No problem, I had the site opened so I haven't seen that the answer was already writen ;)

---

<img src="http://www.boincstats.com/stats/banner.php?cpid=d2319b8f0ad14565556d0ba45b64e779">

> Yes, Toni just explained to me. As I said, I'm a member since last week. So,
> there are some tehnical stuff I have to learn and understand how the system
> works. But one step at the time.
> Thanks for your anwser.

Have I the answer for you ... :)

with only 700+ pages (some quite small) the *IS* a Owner's Manual to read ...

But like the one for your car you probably won't read this one either :(

Anyway, there is a FAQ with sections on credit, was to feedback and complain to the author that he missed a question ...

---