<blockquote>Cogent they are not good on the service front however generous you feel towards them.</blockquote>

Didn't say I felt generous toward them. I recognise that aside from providing an opportunity to vent our ire toward Cogent here, there are constraints on what we (as outsiders) can do, and that there may be constraints on what Berkeley/SETI can do.

Then again, there are times I go off on tangents about lots of imperfections in this world myself. Heck, I've done that in these chat groups.

It is not clear to me that a massive phone campaign to Cogent would have positive results though.

---

<blockquote><blockquote>Cogent they are not good on the service front however generous you feel towards them.</blockquote>

Didn't say I felt generous toward them. I recognise that aside from providing an opportunity to vent our ire toward Cogent here, there are constraints on what we (as outsiders) can do, and that there may be constraints on what Berkeley/SETI can do.

Then again, there are times I go off on tangents about lots of imperfections in this world myself. Heck, I've done that in these chat groups.

It is not clear to me that a massive phone campaign to Cogent would have positive results though.</blockquote>
What Barry says here makes a lot of sense. Since we are not Cogent's direct customer, we have no right to call Cogent.

We do not want to do a Denial of Service attack by calling their NOC and complaining.

---

Sure....I have not mentioned any campaign btw.....new one on me. Oh and we are not outsiders: We are "stakeholders"....if the likes of us pull out...not that I am proposing anything like that....then the project will stop. That's reality.....and I don't mean us few here on the boards....I'm talking about loads of people walking away because of the "Cogent Factor". Its a serious risk to the success of the science here right now.

---

<blockquote>Sure....I have not mentioned any campaign btw.....new one on me. Oh and we are not outsiders: We are "stakeholders"....if the likes of us pull out...not that I am proposing anything like that....then the project will stop. That's reality.....and I don't mean us few here on the boards....I'm talking about loads of people walking away because of the "Cogent Factor". Its a serious risk to the success of the science here right now.</blockquote>
IMHO, the more serious risk is the fact that people seem to be incredibly sensitive to anything they don't perceive as perfect.

As "stakeholders" we can complain -- here in the forums, in E-Mail, postal mail, etc. to the SETI project. If we're fed up with Cogent, we should say that loud and clear.

... and we can even donate connectivity, or try to get bandwidth "donated" to either multi-home or flat-out replace the Cogent circuit.

Think of it another way: we are SETI's customers.

SETI is then provided bandwith through the Campus networking department, CNS. The SETI project is their customer.

While the Cogent bandwidth "belongs" to the project, it is undoubtably "purchased" by CNS.

Cogent gets bandwidth through UFO, and probably a lot of the UFO circuits run on other telco facilities.

So, let's say that an SBC fiber, leased to UFO is broken.

We bitch to SETI, SETI yells at CNS, CNS checks their part(s) of the circuit, and bitches out Cogent, Cogent beats on UFO, and UFO berates SBC.

In each case, each organization listens to complaints from one level upstream, and complains one level down.

As SETI "customers" we have no right to bitch at Cogent -- only to influence the project to dump them if we feel they are totally out of line.

On the other hand, BOINC works fine through these outages. I'm almost out of SETI right now, so I'll be doing Einstein and LHC until they're back, and the new 4.3x clients' scheduler will insure that I only crunch SETI for a few days to make up for lost time. If I was only crunching SETI, then I'd set my cache larger.

It's only the people who get bent when the project is down a couple of days.

---

<blockquote>On the other hand, BOINC works fine through these outages.</blockquote>
Yep.
I generally have about 4 days worth of work, which means i've got no problems with a 3 day outage (takes about 24 hours after an extended outage before i'm able to get new work).

---

Grant
Darwin NT

One problem with extended outages, even with folks who choose to maintain large caches is that they may have workunits coming up 'on deadline' -- and if the outage is long enough (1/2 day and up for some), the work processed gets tossed out as being overdue -- there doesn't appear to be a 'grace period' for submission of overdue units when the 'overdue' can be attributed to an outage (especially an extended outage).

In any event, in terms of tolerance -- when a supported community is in the 100,000+ range, folks really should not be particularly surprised that some number of them have expectations of 'large project reliability' -- ie getting stung by multiple extended (3 hour plus) outages in a 10 day period is likely to be considered problematic by some portion of the community -- especially those who are quite active.

In the past 10 days we've had at least 4 outages of 6 hours or longer. One was for the database maintenence (planned), one was due to a collection of problems the next day (firewall and other problems), and the last two have been due to link failures.

Many folks have an expectation of 95% or even 99+% reliability -- based in large part on their experience with other internet related connectivity. While Seti has also been much higher than 95% over the long haul, for the past 240 hours, the reliability numbers are under 90% -- so we should expect a higher 'noise level'.

At a guess, the on site SETI folks not only expect a higher noise level from us, but also expect to provide a higher over all reliability level to their 'served community'. I have got to guess that the on site SETI folks would look back at the past 10 days plus and flag it as 'unacceptable' and be seeking means to avoid this in the future.

---

<blockquote>
At a guess, the on site SETI folks not only expect a higher noise level from us, but also expect to provide a higher over all reliability level to their 'served community'. I have got to guess that the on site SETI folks would look back at the past 10 days plus and flag it as 'unacceptable' and be seeking means to avoid this in the future.
</blockquote>

I sure hope so. For the first time in days I have to use the proxies again :-(

---

<blockquote>One problem with extended outages, even with folks who choose to maintain large caches is that they may have workunits coming up 'on deadline' -- and if the outage is long enough (1/2 day and up for some), the work processed gets tossed out as being overdue -- there doesn't appear to be a 'grace period' for submission of overdue units when the 'overdue' can be attributed to an outage (especially an extended outage).</blockquote>

If you set your cache to 10 days, and you are running SETI, worst caseyou have four days to report work before it goes overdue.

Personally, I think the sweet spot is at about 6 days of cache if you only crunch SETI (and something like 0.5 days if you crunch two or three projects).

... and I don't remember many outages over a couple of days.

It usually takes a couple of days to fully recover after an outage, so if you cache 6 days of work you should be able to get everything "up" with 4 days to spare.

---

<blockquote>Sure....I have not mentioned any campaign btw.....new one on me. Oh and we are not outsiders: We are "stakeholders"....if the likes of us pull out...not that I am proposing anything like that....then the project will stop. That's reality.....and I don't mean us few here on the boards....I'm talking about loads of people walking away because of the "Cogent Factor". Its a serious risk to the success of the science here right now.</blockquote>

when will cogent stop blinking and light the building

i think i've had enough now it will be seti classic with einstein and lhc

<blockquote>
i think i've had enough now it will be seti classic with einstein and lhc</blockquote>
Classic and SETI/BOINC are on the same Cogent circuit.

---

if i'm not mistaken the trickle server for CPDN is down due to the same ISP, not saying that Cogent is their ISP, i mean that some of the traffic for CPDN goes thru Cogent lines. The last few times that Cogent has been down I've had problems getting CPDN's site or any results posted thru the trickle server to the "results page"....

again, i'm not sure, nobody has mentioned anything here other than SETI connections. i haven't done a tracert as i've got plenty of work to do for several projects.

anyone's thoughts?

---

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

<blockquote><blockquote>
i think i've had enough now it will be seti classic with einstein and lhc</blockquote>
Classic and SETI/BOINC are on the same Cogent circuit.</blockquote>

I'm using SETI classic (I'm using OS/2 and I don't think there's a BOINC for it yet) and I've had serious lack of connection for the past couple of days - it's up now, but whatever caused it, it wasn't BOINC specific! I reckon I've lost about 30 machine-hours of SETI working because of this. Is there a cache facility available for SETI classic? I know originally they announced they would only issue a WU when the previous one was returned, which is a bit shortsighted when you are working in a less than perfect World...

Cheers,
Howard, St.Albans, England

<blockquote>Is there a cache facility available for SETI classic?</blockquote>
Seti driver.

The advantage of BOINC is it handles it's own caching.

---

Grant
Darwin NT

Grant. Yes there is with the cli version that was available. Not an official seti distribution as I recall but you could cache up to 2 weeks WUs I think. Here is the URL of the site that has it if you are looking for it. http://members.home.nl/marcel.zuiderveld/

---

How much downtime has there been that is purely Cogent's fault? Based upon this thread, it lokos like only the France Telecom issue was their fault.

---

How much downtime has there been that is purely Cogent's fault? Based upon this thread, it looks like only the France Telecom issue was their fault. I've said it before, but no one that requries uptime should be any less than dual-homed.

---

<blockquote>How much downtime has there been that is purely Cogent's fault? Based upon this thread, it looks like only the France Telecom issue was their fault. I've said it before, but no one that requries uptime should be any less than dual-homed.</blockquote>

Well I'm not connected by France Telecom and if you look at above messages and traceroutes you could see a lot of people from several countries are not connected by France Telecom but did experience the same problems...To mention a few providers that possibly were affected: KPN, Tiscali, BBNed (but that's probably also France Telecom), Telia, Tleglobe. KPN means the majority of Dutch boinc users...

---

Hey!

Is this about not bein able to update the projects? (Im not very computery so i dont really understand sorry)

I just wanted to know that it wasnt just me whos not been gettin credit and such like for the work completed, i could be completly wrong but im confused.

thanks, Fi
(england)

---

randomthoughts:
I glide upon the evening and dance upon the dawn with Sirius asmy guide and Canopus asmy friend

Some fear the dark but with friends like Orion,Cygnus and Andromeda one needs no fear

To me nature was mute, until i saw the stars

Hey!
Is this about not bein able to update the projects? (Im not very computery so i dont really understand sorry)
I just wanted to know that it wasnt just me whos not been gettin credit and such like for the work completed, i could be completly wrong but im confused.
thanks, Fi
(england)

This thread is dedicated to people who can't (or couldn't) access the server at Berkeley to download WU's.

Your problem is a problem of reporting the results.

You have one WU that was reported too late. There is a dealine of 14 days for reporting a WU. That WU was reported 23 May 2005 at 16:04:15 UTC and had a deadline of 22 May 2005 13:26:32 UTC. For this reason, no credit has been given.

See also your results.

Hi
oh sorry my mistake
Thanks
Fi

---

randomthoughts:
I glide upon the evening and dance upon the dawn with Sirius asmy guide and Canopus asmy friend

Some fear the dark but with friends like Orion,Cygnus and Andromeda one needs no fear

To me nature was mute, until i saw the stars