Yesterday the number stuck around 143K give or take some, then this morning I see

Ready to send 1,279,362
In progress 1,713,462
Waiting for validation 199,788

Given that the new server was reducing this number by 6k/hr it should be down to zero (relative to new incoming units).

Hope nothings broke.

tony

---

Yeah i noticed that today also =) strange... :/

---

Yes, a very clear reversal of fortunes for the Validation Q -
it seems to have spent the last 10 or 12 hours adding several
thousand results per hour instead of clearing at that rate
as it had been since Friday. :(

Possible that an early-Sunday a.m. (CA time) cron job backup
or some such kicked in and took away resources from the DB or
or the validator? hmmm.

Dave

They probably remove WU's on the older server when queue got extremly large and are now inserting them back.

Well, what ever it is continues to happen at a good clip.

Last hour on the stat pages shows that the Q grew by
another 15,000. I think that's much faster growth
than what was occuring when the old machine was
trying (but not able) to keep up. So either the Q is
getting much more input, or having much less validation
done, than was happening before the change over.

Dave

another 48k and we're right back to where we started after the migration. Hope this upward trend does not continue.

Ready to send 1,275,094
In progress 706,226
Waiting for validation 210,906

The validator is still working.

This



Is a good barometer of what is happening :)

Dunc

---

In theory, the average number of WU and Results should be the same each day. But, with the connect every "x" being set differently, and people like me that "flush" our results daily, these numbers are not likely to ever be in balance.

So, minor fluctuations in results returned add turbulence to the process flow and may add to our troubles.

---

I agree, the validator IS working, but I have no idea why the number keeps growing. The most likely reason is what Paul told. I have no other explanation.

---

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

> I agree, the validator IS working, but I have no idea why the number keeps
> growing. The most likely reason is what Paul told. I have no other
> explanation.

Of course, maybe now the new server is spinning on the "garbage" collection spindle ...

---

Remember that count was out by around 150k to start with. I think it's found those "missing" WUs. A few hours ago, it looked like it was going very slowly yet I was getting credit at a phenominal rate. ;)

---

Be lucky

Neil

> Remember that count was out by around 150k to start with. I think it's found
> those "missing" WUs. A few hours ago, it looked like it was going very slowly
> yet I was getting credit at a phenominal rate. ;)
>
Yesterday the count hovered around 140K, 143K, 144K, for most the day. so you think that's some "Baseline"(for some reason), and then that would be the theoretical zero? Todays climb by 70K would be some normal fluctuation?

I think Paul's closer to the truth, and that the DB is undergoing some kind of internal function, and that the 70K is the inbound returns not being validated.

What do you think? anyone?

Could Berkeley have taken some of the WU offline for a while, when the servers were going nuts, and have over the weekend slowly incorporated them back in, and today they dumped the last of the load in?

It's going in the correct direction again, and I have seen a couple of my older WU that had all the correct responses etc., finally get validated.

And I hit 100K today! WOO!

---

My movie https://vimeo.com/manage/videos/502242

> > Remember that count was out by around 150k to start with. I think it's
> found
> > those "missing" WUs. A few hours ago, it looked like it was going very
> slowly
> > yet I was getting credit at a phenominal rate. ;)
> >
> Yesterday the count hovered around 140K, 143K, 144K, for most the day. so you
> think that's some "Baseline"(for some reason), and then that would be the
> theoretical zero? Todays climb by 70K would be some normal fluctuation?
>
> I think Paul's closer to the truth, and that the DB is undergoing some kind of
> internal function, and that the 70K is the inbound returns not being
> validated.
>
> What do you think? anyone?
>
Hi! Looks like missing WUs (look at thread about missing 150k: http://setiweb.ssl.berkeley.edu/forum_thread.php?id=11303 ) are found.
It used to be wrong number in stats before.

On the other hand: I am getting less credits per hour last few hours,
compared to yesterday.

It might be related with validating of 3 result/4 result WUs?
WUs with no proper flag in dB?

Whatever. Few more days of fast credit flow does not hurt.

73
Iztok

> Remember that count was out by around 150k to start with. I think it's found
> those "missing" WUs. A few hours ago, it looked like it was going very slowly
> yet I was getting credit at a phenominal rate. ;)

I agree, S-C. The "waiting for validation" queue is an estimate, and that counter was zeroed out a few times over the past week or two. I'm still seeing a steady flow of credits being validated, so I think that the validator is working fine, it's just this counter that's adjusting itself to now account for WU's that it's rediscovered. As I look at results records, many old "stuck" units have now been validated properly or had another WU sent out.

Paul's theory is certainly accurate too. There will always be fluctuations in the volume of WU's returned. With many people out there increasing their cache sizes recently, some of these results are coming home to roost in large numbers!

---

here's Roms' answer:

Rom Walton
Forum moderator
Project developer

Joined: Apr 27, 2000
Posts: 283
ID: 85465
Posted: 13 Feb 2005 18:47:17 UTC

The disk array that was holding the log files for the backend componentss, became full and caused all the components to stop.

Jeff went in and manully cycled the log files and moved them to a different array.

----- Rom

BOINC Development Team, U.C. Berkeley

I noticed the increase in "number to be validated" as it's listed by the counter on the server status page, myself.

http://setiweb.ssl.berkeley.edu/sah_status.html

Waiting for validation 176,821

as of that page right now. It was steadily dropping since the new servers came up. Does make me wonder if another validator (perhaps on another server or CPU) might help with getting these validated, especially as we now have a new server that according to the SETI team is under-utilized and not likely to not be up to the task itself...

Then again, someone does bring up a point with people having perhaps increased their cache sizes due to the difficulty people were having in getting WUs last week and so. How many of those WUs with a possible increase, might be waiting for the quorum? Honestly couldn't say, and not inclined to take a tabulation on every user's account for how many WUs they have pending, for which a quorum hasn't been reached. The people at Berkely probably have a better way to query their own database, to find such things out anyhow, then such a method which we could utilize... And yes, I know we're looking at an estimate while following the trend it displays as well.

---

> as of that page right now. It was steadily dropping since the new servers
> came up. Does make me wonder if another validator (perhaps on another server
> or CPU) might help with getting these validated, especially as we now have a
> new server that according to the SETI team is under-utilized and not likely to
> not be up to the task itself...

"under-utilized" means it could do more so it is up to the task. The "waiting for validation" queue has been dropping at up to 6,000 per hour from what I have seen. This means that the validator is validating all the results as they come in PLUS 6,000 per hour so there is still room for growth here. Add to that the fact that they can add another 2 CPUs and several GB more RAM to the thing and I think it is more than up to the task. Besides, the validator itself may not be the bottleneck. For some time now it has always been the database that has been the bottleneck in the system so adding another validator wouldn't even help. I'm not sure if this has changed with the new hardware or not.

---

A member of The Knights Who Say NI!
For rankings, history graphs and more, check out:
My BOINC stats site

The server status page seems to be stuck. It's not updated since 14 Feb 05:10 UTC. (10:10 pm eastern, Feb 13).

[As of 14 Feb 2005 5:10:07 UTC]

Ready to send 1,087,643
In progress 920,822
Waiting for validation 176,821

Patiently waiting

> How many of those WUs with a possible increase, might be waiting
> for the quorum? Honestly couldn't say, and not inclined to take a tabulation
> on every user's account for how many WUs they have pending, for which a quorum
> hasn't been reached.

the waiting for validation queue only shows WU's that are ready for validation. The transistioner only sends WU's that have a quorum to the validator for validation. So WU's that are waiting on results would not be in that list

---