> Fine...money talks. But when classic comes over with many many more users I
> doubt somehow that them servers are going to be able to catch up like they do
> for 100,000 users or so there is now. I'm sure one of those UCB maths
> undergraduates could do a little queuing theory and model this......could be
> interesting. My guess.......it will never catch up and it will go down the pan
> as users become disgruntled. Perhaps they need to consider dynamically
> allocated capacity as well as alternate and diverse routing?

But they are using the same line now as well, so nothing will change in regard of bandwidth, connectivity and this stuff.

Fine...money talks. But when classic comes over with many many more users I doubt somehow that them servers are going to be able to catch up like they do for 100,000 users or so there is now. I'm sure one of those UCB maths undergraduates could do a little queuing theory and model this......could be interesting. My guess.......it will never catch up and it will go down the pan as users become disgruntled. Perhaps they need to consider dynamically allocated capacity as well as alternate and diverse routing?

---

> Well my point is: there is more than 1 route. But I guess if you design for 1
> data then you have to accept the link fialing from time to time. Still....it
> might be worth further consideration.

As I read the article posted, campus wants $6k/month for above normal usage of their pipe, and Cogent is $1k/month.

So, intentionally, most of the project uses the Cogent bandwidth exclusively.

... and some things, like these forums, use campus bandwidth exclusively.

---

Well my point is: there is more than 1 route. But I guess if you design for 1 data then you have to accept the link fialing from time to time. Still....it might be worth further consideration.

---

> Not wanting to be too nosey but.....how are they able to provide web services
> from a host (klaatu - which is a transitioner also) on their network when they
> say the cogent link is unavailable.
>
> Hmmmm?
>
> If they have diverse routing and connections here they should be making better
> use of it?

AFAIK they have a smallband line via ordinary University connections, and a broadband through Cogent for the data transfer. Don't ask me numbers, but use of the fora-connection for data transfer would do no real benefit usually.

Look this old article

---

Gruesse vom Saenger

For questions about Boinc look in the BOINC-Wiki

> Not wanting to be too nosey but.....how are they able to provide web services
> from a host (klaatu - which is a transitioner also) on their network when they
> say the cogent link is unavailable.
>
> Hmmmm?
>
> If they have diverse routing and connections here they should be making better
> use of it?

From what I understand, the Cogent link is the primary datalink that serves as the I/O port from which all the data information is u/l and d/l. The proprietary link that Berkeley allows them to use, and what we used before Cogent, has been relegated to the website and forum communication line. Pretty much that's why we're able to post and read forum messages but not upload data information.

Also, if the old line system was barely able to handle SETI Classic's 3,500,000 or so users back in 2001, what makes you think it can handle present day numbers?(And yes, I know not all 3,500,000 crunched back then but a good portion did.)

---

Not wanting to be too nosey but.....how are they able to provide web services from a host (klaatu - which is a transitioner also) on their network when they say the cogent link is unavailable.

Hmmmm?

If they have diverse routing and connections here they should be making better use of it?

---

> One option that would help would be for the server status page to have some
> sort of "network through-put" meter.

It's here.

---

Turns out that problem just happened at 9AM this morning. Affects Seti Classic as well.

The other problem (and it IS a problem) started yesterday during the day and remains (from what I've seen) unacknowledged. I suspect that until the link problem is dealt with, no diagnosis of the older problem (from about 24 hours ago) will be possible -- although first step there would be for folks at the shop to become aware of that problem so they can look into and resolve it as well.


> May 3, 2005
> Our Cogent link has been down since around 9:00pm PST (16:00 UTC). Data
> servers are currently unreachable. The problem is currently being diagnosed.
>
> was it worth waiting for though ;)
>
>

---

If you look at http://fragment1.berkeley.edu/~cricket/inr-668-interfaces.html it is evident that the UCB network is/was having problems. No traffic since 09:00 ( or there abouts ) local time.

---

At Last....

May 3, 2005
Our Cogent link has been down since around 9:00pm PST (16:00 UTC). Data servers are currently unreachable. The problem is currently being diagnosed.

was it worth waiting for though ;)

---

> > One option that would help would be for the server status page to have
> some
> > sort of "network through-put" meter.
> >
> > If I am trying to upload (as I am now) and can't get through (& I
> can't at
> > the moment) and the server status page showed network traffic at 99%
> capacity,
> > I would know why I can't at the moment get any response- the
> system is
> > swamped.
> >
> > If the traffic was at 0% ditto- the link is down.
> >
> > If it was listed as '50%' then I have a problem, but it would seem SAH
> is
> > running OK.

I was also wondering why I can only get two units in my cache, even though I changed and updated several days ago my preferences to "connect to BOINC every 5 days". I wonder what I could be doing wrong?

Thanks,

Nick
> >
> > Listing things as 'on' or 'off' is not as informative as 'on, very busy'
> or
> > 'on. doing nothing', 'on, normally busy' or 'OFF'. A suggestion as to how
> busy
> > the system is gives more info.
> >
> > IMHO.
> >
> > Likewise, can not get through here either, for the last 12 hours on any
> computer. Yet, the "green light" is on, like it is running normal. But at
> least I got credit for many units that were pending last night.
>
> Nick
> >
>

---

> One option that would help would be for the server status page to have some
> sort of "network through-put" meter.
>
> If I am trying to upload (as I am now) and can't get through (& I can't at
> the moment) and the server status page showed network traffic at 99% capacity,
> I would know why I can't at the moment get any response- the system is
> swamped.
>
> If the traffic was at 0% ditto- the link is down.
>
> If it was listed as '50%' then I have a problem, but it would seem SAH is
> running OK.
>
> Listing things as 'on' or 'off' is not as informative as 'on, very busy' or
> 'on. doing nothing', 'on, normally busy' or 'OFF'. A suggestion as to how busy
> the system is gives more info.
>
> IMHO.
>
> Likewise, can not get through here either, for the last 12 hours on any computer. Yet, the "green light" is on, like it is running normal. But at least I got credit for many units that were pending last night.

Nick
>

---

One option that would help would be for the server status page to have some sort of "network through-put" meter.

If I am trying to upload (as I am now) and can't get through (& I can't at the moment) and the server status page showed network traffic at 99% capacity, I would know why I can't at the moment get any response- the system is swamped.

If the traffic was at 0% ditto- the link is down.

If it was listed as '50%' then I have a problem, but it would seem SAH is running OK.

Listing things as 'on' or 'off' is not as informative as 'on, very busy' or 'on. doing nothing', 'on, normally busy' or 'OFF'. A suggestion as to how busy the system is gives more info.

IMHO.

---

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

Do you think a reminder of the good old times will help, or the mentioning of useless send-again WUs just to keep the 'unwashed masses' happy in the past will help?

Yeah true. We will lose most of the classic users if they cannot see that things are reliable....and when they are not so that they can understand what is happening. That's one of classic users biggest complaints about the migration.....the other is its too complex....but I dispute that. Its all about sharing some information on a timely basis. As I have said elsewhere....you could cut the posts here on these forums by half if there was a decent news service.

---

> Give 'em time to wake up and get a coffee before kicking in high gear at
> work.
> If you want to have 24/7 babysitting at the servers, feel free to donate some
> Mega-$ to Berkeley (or get your beloved rich aunt to do so ;)
>
Everyone should indeed.

My point is that this is a global project - while some are just getting their first caffine-fix, others are taking a night-cap.

I fully realise that there are very limited project resources and stress that I am not having-a-go. Overall it is a fantastic project with great potential and moving in the right directions - the developers have done a great job with what they have.

From my own point of view, I've been around long enough to understand how and why things happen. I recognise my mood as 'frustrated' - I can hanlde it ;)
I only worry that others with less-rounded experience of SETI Classic, BOINC and BOINC/SETI, and indeed non-comercial projects, may not have the stamina to stay the course when it's a bit rough.

---

Give 'em time to wake up and get a coffee before kicking in high gear at work.
If you want to have 24/7 babysitting at the servers, feel free to donate some Mega-$ to Berkeley (or get your beloved rich aunt to do so ;)

---

Gruesse vom Saenger

For questions about Boinc look in the BOINC-Wiki

Agreed. As I said in another thread....people just want to know. Quote from an email to me from D Anderson 19th April 05 "My feeling is that the existing mechanisms -Questions and Answers pages, and News items for temporary problems - should be enough if we maintain and use them correctly." . I was asking about being better informed about problems and issues. So this is the view you have got to change before anything will happen. Do remember....there is only a few of them to keep things going though.

---

> We won`t get anything with this, this is not the first time they don`t keep us
> informed ... We always say "Please..." but over and over again same things
> happens, we don´t know anything,,,
>
Maybe, but we can always hope...

I thought I would stress the international aspect of these outages to see if that increases their awareness of the issue.... well I can try can't I.

---