Yep, back to doing real work. Just got a few hundred for each of the starving clients.

---

Immediately after I got my caches filled, I started getting just '0 tasks' returns and the caches are depleting again...

Immediately after I got my caches filled, I started getting just '0 tasks' returns and the caches are depleting again...

For me it was only for about 30 mins around 12 noon GMT that there were no new tasks.The cache has been re-filled since then.

Starting to get "Project has no tasks available."

Looks like the brick fell off the accelerator pedal...

---

Results returned and awaiting validation hits another record high. I wonder if it'll make to to 18 million by the 31st?

---

Grant
Darwin NT

Starting to get "Project has no tasks available."

Looks like the brick fell off the accelerator pedal...

All the hosts that were in backoff mode are now contacting the Scheduler & returning work and asking for more. Demand exceeds supply.

---

Grant
Darwin NT

And then I couldn't post to the BOINC forums anymore. Posts stay in perpetual trying to send mode. Oh well.
Ooh same here

Starting to get "Project has no tasks available."

Looks like the brick fell off the accelerator pedal...

All the hosts that were in backoff mode are now contacting the Scheduler & returning work and asking for more. Demand exceeds supply.

. . Well I swapped over to the NBN tonight and now the wheels have well and truly fallen off. I can contact other sites but BOINC is borked. One machine uploads, reports and downloads just fine. The next machine can upload OK but attempts to report get "cannot contact server, internet is OK". The other two machine cannot upload anything, total fail and almost immediate project backoff. :(

. . AAArrrggghhh!!! :(

Stephen

:(

I have no idea why but the word KISS popped into my head ;-)

Apparently the servers dried up again.

The SSP stopped updating and both of my computers stopped getting any new work. Scheduler request still work but always return 0 tasks.

The number of results in the database was at a new record when the SSP still updated: 24.186 million.

I can't say that I'm having any problems here at all really as my 2 main rigs are topping up on every 2nd or 3rd request.

Cheers.

No new WU, eventually when one drops in my backyard, obviously a resend, is crunched almost instantly because my host is programmed to automatically start any resend first.

For now the large cache is holding. Hope that will be fixed during the night.

---

This situation may sort itself out. Possibly when results and progress drop below a certain number work will get sent out again. I am just guessing I have no hard facts. I have 15 tasks mailing for my GPU and 23 currently running on my CPU when they are finished I will be out of work until the server gives me more. I think a lot of us are in this boat. As I posted what happened I got given around 20 CPU tasks. Good luck everyone

---

The server will probably limp along on what can be done remotely until further notice:
California Governor Issues ‘Stay at Home’ Order for Residents

---

We have 14 partial blc files to split currently. The 16 splitting channels usually work on 11 files. The splitters are going at full tilt right now filling up the RTS queue (YES!), but I fear not only running out of files to split, but that the rate of splitting will be lower as it is splitting fewer files.

We still have Aricebo files, so we won't run out today though.

Just wondering if they will add more tomorrow. ???

Well, the Web site is almost dead. Forums aren't completely dead- just mostly dead.
And the Scheduler is MIA again.
I'm sure we did this yesterday.

---

Grant
Darwin NT

The server will probably limp along on what can be done remotely until further notice

The only things that can't be done remotely are hardware upgrades or pressing the reset button of a crashed and unresponsive server. Servers are managed remotely even when the person doing it is physically on site because servers in server racks rarely even have any local keyboards and displays.

And even when there is a local console, no one wants to use it unless he has to because server rooms are too cold and too noisy places to work in.

What I think the Seti staff should do is to write and run a simple script that moves all the database rows of workunits waiting for assimilation that aren't waiting for any unreturned results, the result rows linked to those workunits and the workunit and result files associated with them into a separate backup storage.

That would remove about half of all the results in the database and the memory pressure they cause, which would probably resolve the server problems for a long enough time to last until the work distribution is stopped.

And than after it is stopped and the database starts shrinking, those backed up workunits could be returned back to the assimilation queue to be processed.

The server will probably limp along on what can be done remotely until further notice

The only things that can't be done remotely are hardware upgrades or pressing the reset button of a crashed and unresponsive server. Servers are managed remotely even when the person doing it is physically on site because servers in server racks rarely even have any local keyboards and displays.

And even when there is a local console, no one wants to use it unless he has to because server rooms are too cold and too noisy places to work in.

I've visited two BOINC server rooms - the Einstein ATLAS cluster 10 years ago, and SETI's last summer. Both had a 'crash cart' parked in a corner - a trolley with monitor, keyboard and sundry useful tools. Probably a mouse too, though most server work seems to be done at the command line.

In SETI's case, each equipment rack is locked with its own security code. Eric opened one door to plug in the crash cart, and remotely shut down a different server which needed a hard disk replacing. That done and tested, he wheeled another trolley to the server we needed to upgrade, adjusted the working height, and slid that server out of the rack and onto the worksurface.

I don't remember it being seriously cold, but certainly 'pleasantly cool' compared to the California summer outside. And I've still got the disposable earplugs I was advised to wear. Souvenir!

Fun fact: Eric doesn't even have University authority to enter the server CoLo by himself. We had to meet Jeff Cobb outside to go through the formalities and be signed in.

Sounds good. In saying this I do not know how to write such a thing plus is it worth writing one for 11 or 12 days?

---