Is it only me that has noticed that not credit seems to be getting granted (or very little)??

In the past I have never seen the 'waiting for validation' (on the server page) get into double figures. It's as I type 101,678 or there a bouts. This morning it was 60,000 odd.

Results are getting returned, but nothing seems to be happening beyond that.

How long before the storage space for 'un-validated' results fills up? It may be huge, but must be finite. When it fills up, what will happen? I am not going to let Boinc return any results till the backlog clears, as I don't want my returned results to 'get lost in a major server crash' caused by a full HDD...


Any-one any views/answers/comments?

---

Foamy is "Lord and Master".
(Oh, + some Classic WUs too.)

AFAIR it was up to some 200K after the another outage.

Simply a bit backlog, patience is a virtue ;)

---

Gruesse vom Saenger

For questions about Boinc look in the BOINC-Wiki

>Simply a bit backlog, patience is a virtue ;)

I beg to differ about it being "a bit of a backlog".

100,000 compaired to the normal 2-6 is more like 'Gridlock'.

A backlog suggest something that has built up, but is in the process of clearing. The current situations is growing, not clearing.

But I am patient...

I have 5-6 days WU's cached.

Not worried, just curious.

---

Foamy is "Lord and Master".
(Oh, + some Classic WUs too.)

It is a backlog. There are reasons for the backlog:

They just broght the replication server back online. This will take time to replicate everything over to that server, because it is slower than the main server.

The spike of traffic the last few days has been huge, and the validate does take some time to validate each result. If you had 200000+ computers pushing data at you, you'd get backlogged also.

Seeing this number at 100000+ is nothing to worry about. Things are getting validated. My credits have gone up the past couple of days, abit slow, but it is happening.

Just let everything get back to normal. This could take a week or more. Last time I think it took almost 10 days before the validator was back near 0.

---

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

It's going down now. While I was reading this thread, it went down from over 106,000 to about 104,000. I guess that's how much it did between two updates (ca. 10 minutes) of the Server Status page. So, if it continues in this pace, we'll reach zero in about.... 9 hours.

---

When they switched over to the new server months ago(that was a multi day outage), the Waiting to validate number was as high as 278,000. After, they came back, the new server not only handled all the newly arriving WUs, but also reduced this number by 6,000/hour.

Be patient, keep watching.

>
> A backlog suggest something that has built up, but is in the process of
> clearing. The current situations is growing, not clearing.
>

A quick look on the server-status-page shows 2 of the validators is running on Kosh, this is the replica-database-server.
The 2 other validators shares Koloth with 2 transitioners and the file_deleter.

A Transitioner is responsible for generating new "results" for newly-split wu, for errored-out results and past deadline-results if needed. Also responsible for setting "need_validate"-flag when enough results for a wu to try validation, and some other jobs.


After any longer outages, both the replica-database and the Transitioners will work harder, using more cpu-power than normal. This means the validators will get less cpu-power than normal, till things catches-up.

Hang with me, cause I am going into never neverland and not to see Jocko either, OK we have 50,000 hosts banging on the server to download 10 results each, that is, "damn ran out of digits" Ok used a calculator", that equals 500,000. So the whole ball of wax is moving at a crawl all the way thru the system so be patient.

---

Semper Eadem
So long Paul, it has been a hell of a ride.

Park your ego's, fire up the computers, Science YES, Credits No.

I just noticed something in my browser (I use Opera 8) yesterday, when I visited the Server Status page: in the status bar that pops up when a page is loading, it said "Connecting to bla-bla-bla... (#70 in queue)" or something like that. I thought that was funny, so I reloaded the page several times and it said #13 and so on...

So, one can actually see how many are in front of you in the queue! I didn't know that was possible (but then again I'm no TCP/IP guru). That'd be a neat feature in BOINC Manager too! Or at least in a 3rd party tool like BoincView.

---

Well as of 1:30EST, the que was at 150K and going upwards. I did not see my stat's move in a few days so I started researching this. Funny, Seti runs only 6 computers ( must be real beast ) but I wonder if maybe we should start a donation fund for them to get another one. I have donated in the past ( back in 2001 ), and I trust them to do what's right. DOes anybody have a clue to what type of server they really need ?

Rangepup

> Well as of 1:30EST, the que was at 150K and going upwards. I did not see my
> stat's move in a few days so I started researching this.

>Funny, Seti runs only 6 computers ( must be real beast )

actually they´re not really beast more like little sheeps :-)

>but I wonder if maybe we should start a
> donation fund for them to get another one. I have donated in the past ( back
> in 2001 ), and I trust them to do what's right. DOes anybody have a clue to
> what type of server they really need ?
>
> Rangepup
>

I don´t think that we need to start a donation for new hardware cause when seti classic is shut down these servers (i think) will switch over to seti2.

---

Join BOINC United now!

I noticed this morning, UK, that my credits etc hadn't risen. So I checked the server status and at 07:00 UTC approx that the waiting to validate was 128K. I checked again at midday and it ha risen to 135K and just now it's up to 152K.

But, one of the noisy units that I've just completed, which had already been thru the granted credit stage updated almost immediately.

Now at 155K and therefore still rising, this is not a backlog, it is an ever increasing jam!

---

I am getting credits. They are validating, just taking their time.

As it has been stated, they added the replicator back online. Since the replicator is slower than the main database and will take time for the replication to be finished. This allows them to do backups without taking the main database down.

When that replication finishes, the validator will catch up. I had replicating servers at a job once, it went down cause of a hard drive failer. It took almost a week for it to fully replicate after we got it back online.

Things are working. Watch your credits closely, I bet you will get a trickle of credits every so often. I've gotten nearly 1000 since Friday.

---

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

> Now at 155K and therefore still rising, this is not a backlog, it is an ever
> increasing jam!
>

At 160K now, and two of the four validators are down. 'kosh' is off-line.

Want to take bets on how high the backlog will go?

---

> > Now at 155K and therefore still rising, this is not a backlog, it is an
> ever
> > increasing jam!
> >
>
> At 160K now, and two of the four validators are down. 'kosh' is off-line.
>
> Want to take bets on how high the backlog will go?
>
>
>

I'm seeing 155K, maybe a minute after your post, everything still up

Server time: [As of 1 May 2005 19:00:08 UTC]

[edit] when i hit refresh, i got the same, sorry [endedit]

---

No matter where you go, there you are...

> At 160K now, and two of the four validators are down. 'kosh' is off-line.
>
> Want to take bets on how high the backlog will go?

I think, it will go up above 200K again, but so what?
That's no problem whatsoever, and I hope they don't bother to look for it, as they have important things to do.

---

Gruesse vom Saenger

For questions about Boinc look in the BOINC-Wiki

I see Two validators on kosh have gone 'Not running' recently. Perhaps someone at Berkeley is looking at this.

---

Many of my uploads right after the outage are still not credited, while more recent uploads seems to be credited at once. I suspect that the validate queue is a LIFO stack. ...or maybe just random? (Perhaps Matt or someone at Berkeley could comment on this) If so, I guess users would be more happy with a FIFO stack scheme.

LIFO = Last In, First Out
FIFO = First In, First Out

---

> I see Two validators on kosh have gone 'Not running' recently. Perhaps someone
> at Berkeley is looking at this.

More likely, will be looking at this when they return to the lab on Monday morning. With the other two validators on koloth showing green, I wouldn't want to pull them in from their weekend!

---