CPDN upload server outages

Milo has been given some extra storage space for FAMOUS models. While he transfers data he needs to shut down several upload servers. The CPDN Server status page already shows an upload server outage.

It is impossible to predict which files from which models will be affected at any time. If your model's file upload is stuck in the Transfers tab and cannot upload you could suspend Boinc network activity in the Boinc manager Activity menu for a few hours. Or you could take no action.

CPDN upload server outages

Milo has finished transferring FAMOUS models so all the servers are now up and running.

Pre-warning on extended outages at Seti@Home:

Matt Lebofsky wrote:

Starting next week our servers will be turned off for several days a week - initially, Tuesday through Thursday. This will allow us to devote more time to scientific development.

In addition to gaining developer time, we will take advantage of public server outages to run the science servers at full bore, getting all of the database resources. So science production, as well as science development, will increase.

- Matt

Source thread.

GPUGrid fell off the web last night, and hasn't reappeared today - on previous form, I doubt it'll restart until tomorrow morning.

Refugees have started to appear at other projects that BOINC flags as NVidia-capable.

Did I miss an announcement about AQUA being down?

edit: Yes, I apparently did. It's back up.

SETI@Home successfully came up Friday Morning pacific time and is dishing out work again.

Seti@home down:

Projects are down due to a database machine crash.
The machine that serves the BOINC database crashed this evening. It will be recovering overnight. 3 Jul 2010 4:17:58 UTC

---

Collatz appears to be down both for me and with http://downforeveryoneorjustme.com so I don't know what else to say.

Regarding Collatz

Slicker has lamented the reliability issues of his ISP over the past two months or so. Things had been running reliably at Collatz for the first eight months of the project. Since then, things have gotten progressively worse.

According to Slicker who runs the project, the problem is all on Comcast and they are not inclined to either acknowledge or resolve the problem. I guess I'm not sure that this is a 100% ISP issue and wonder if the increased popularity of Collatz along with its increased load are presenting one of those complex/compound problems for which collaborative problem solving is essential.

In any event, the problems at Collatz are encouraging folks back to MW (if they have the GPU for it) or alternatively to Dnetc.

The outage of the moment is now (1:15PM PDT) around 4 hours.

Collatz came up yesterday and is now offline again. The joys of an unreliable ISP.

Right -- though I do wonder -- long intermittent outages over an extended time period are pretty rare. This has now been going on not for days or weeks, but rather months. Somehow I think the issue might include ISP specific problems but that might not be the whole story here.

Collatz came up yesterday and is now offline again. The joys of an unreliable ISP.

I see that Collatz remains offline -- about 12 hours so far. If it is the ISP, one would think their techs could WALK to the site to confirm that there is a problem.

CPDN upload server outage

Milo is moving large quantities of uploaded FAMOUS data from several upload servers to a FAMOUS storage server called Kraken.

At the moment upload server uploader.oerc is down. Some FAMOUS zip files will not be able to upload and will remain in the Transfers tab. Zip files for some other model types are also affected, for example the final zip files _1 and _3 from HadAM3P models. Other upload servers will also be affected until Milo has completed the data transfers.

Here is the Server status page.

You could suspend Boinc network activity if you do not need it for other projects. Or you could take no action.

For those people saying it can't happen, the World Community Grid seems to be down. Forums and upload server are AWOL.

They were back some 4 hours later.

I got some Collatz work nits this morning, but nah... it's gone again.

Collatz is back as of now. Forum post says it something with his cable connection and his cable modem. Comcast claims the modem is ok when they visit but he says when he tests it some how it is showing signs of an upsteam signal to noise problems. He has given a temp credit bump for work turned in while he works on a fix. this is posted in his forums.

For those people saying it can't happen, the World Community Grid seems to be down. Forums and upload server are AWOL.

They were back some 4 hours later.

Was announced and stagged, happening to the best in class.

http://www.worldcommunitygrid.org/forums/wcg/viewthread_thread,29345

If all else fails, visit Facebook and Twitter. If Facebook app is down, check Twitter if downtime is unscheduled.

---

Coelum Non Animum Mutant, Qui Trans Mare Currunt

MW is offline due to a database crash. This is just a few days after having a a crashed validator and work generator over the weekend.

SETI is in their new maintenance cycle -- 2 days prior to maintenance folks scramble to get work, then a 3 full day maintenance, then a 2 day traffic jam as folks upload results, then the cycle begins again. I hope they are able to move past that as a 'regular' cycle.

---

Collatz went offline again -- about 5 hours ago -- I'm sure glad I don't have to cope with Slicker's ISP.

OK - Collatz is back online -- the same connectivity issues persists though so I guess we can expect this problem to surface perhaps on a daily basis unfortunately.

MW is still offline but appears to be getting closer to coming back to life.

SETI, after their weekly self-imposed outage to perform essential maintenance encountered, wait for it ..... a new maintenance problem and so their three day outage is extended. SETI seems to be something of a reverse BP - where their efforts can't START a flow these days. With their current self-imposed three day outage along with pre-outage and post-outage traffic issues, it is rather discouraging for 'SETI-only' BOINCsters. Those of us in multiproject land simply reduce the share of CPU and GPU processing that SETI sees - which probably helps SETI by reducing the clearly too large load they have from their large user base.