Formula Boinc Sprints 2018

[StefanR5R]

It would appear that ODLK server is catching up!

There are even stats updates again.

The second 18 hours:

 

[StefanR5R]

Hmm. We received 2.4 M points during the ODLK1 sprint. However, while ODLK1 and ОДЛК have similar task run times, the latter has only ~1/10 of the credits per task. That way we won't make it further up in the marathon during the next 3 days.

I must have looked at wrong tables. Credits per task and credits per time are actually very similar between ODLK1 and ОДЛК. We should pass Gridcoin and Taiwan in the ОДЛК marathon this weekend.

 

[TennesseeTony]

So....how do we get tasks to report if the <max_tasks_reported> tag doesn't help?...just let the machines sit idle and keep trying? The uploads are complete at least.

 

[StefanR5R]

No, just let the machines crunch tasks full steam and keep trying.

I have still uploads timing out, downloads timing out, scheduler requests timing out. But eventually some succeed. Hence I need to baby-sit my machines somewhat, in order to avoid too long back-off periods for uploads/ downloads/ scheduler requests.

Before I went to sleep last night, I made sure that at least my faster hosts have at least as many tasks downloaded as needed to remain busy the whole night.

 

[TennesseeTony]

After sitting in front of the TR for 10 minutes, pressing update every 20 seconds or so, and neither sending nor receiving the slightest of data, I sighed, and allowed new work for Rakesearch, and.....instantly odlk began downloading tasks. "No new tasks" for Rakesearch was selected, and, although still struggling, the TR is now back to the business at hand.

Coincidence? or cross project jealously?

 

[biodoc]

Stats have been updated on Free-DC.

 

[StefanR5R]

But the hourly stats at formula-boinc.org continue to be sporadic. Again no updates during the last 3 hours.

 

[TennesseeTony]

The TR has 2 inconclusive tasks.....the original wingman for both was anonymous host 5820 with 5 invalid's so far, so.....one of us has a problem. The back-up wingmen are two separate hosts, but both running Opteron's from 2011, one of which has a number of 'timed out' errors. This might take a while to resolve...

 

[Skillz]

Getting close to catching SUSA. Hopefully when/if it does happen the Stats will update properly before the end of the Sprint.

 

[TennesseeTony]

Thanks, Skillz and motqalden, for the fuel additive!

FB updated finally, and we're creeping up there:

1 SETI.USA 1,471,684
2 TeAm AnandTech 1,411,240
3 BOINC@AUSTRALIA 504,102
4 Czech National Team 403,425
5 Rechenkraft.net 314,882

.... Hopefully ... the Stats will update properly before the end of the Sprint.

Hmmm, I wonder what happens if they don't? I would hope Sebastian would manually update like crazy for a few minutes before accepting hours old data, but I don't assume that.

 

[crashtech]

I added a Haswell Celeron to odlk this morning, that ought to show them.

 

[StefanR5R]

That Celeron must have done this:

(from https://stats3.free-dc.org/stats.php?page=proj&proj=odl)

 

[crashtech]

Yeah, go mighty Celeron!

 

[Skillz]

Pretty sure it was the two C2D laptops I added when I joined.

 

[StefanR5R]

Pretty sure it was the two C2D laptops I added when I joined.

Ah, so that's why the "Sprint Killed ODLK's Website".

 

[StefanR5R]

Getting close to catching SUSA. Hopefully when/if it does happen the Stats will update properly before the end of the Sprint.

Hmmm, I wonder what happens if they don't? I would hope Sebastian would manually update like crazy for a few minutes before accepting hours old data, but I don't assume that.

Looking at my pending/stalled uploads and downloads, I am seeing many more downloads than uploads in progress or stuck. And this makes sense given that the ODLK server is at the admin's home, certainly behind an asymmetric connection with distinctly lower upload bandwidth than download bandwidth (from the server's POV, IOW lower download bandwidth than upload bandwidth from our POV).

It is likely that most clients do not download anymore by the end of the sprint, lowering the overall rate of networking I/O and, importantly, making room at the server's narrow upload channel for server responses.

--------

(edit) - During the past 13 hours, there was only a single stats update at formula-boinc.org. This makes me wonder whether a certain heavy hitting team which was seen to start sprinting on Saturday instead of Thursday in earlier sprints of this year is doing so again in this sprint.

 

[Orange Kid]

ODLK Day two;
In second?, I guess, as stats are pretty much non-existent.
Nothing has been updated for hours...so with my tardiness this would have been the same if I were on time.
...and WELCOME motqalden and [H]Skillz

1 SETI.USA 25 ___1,471,684
2 TeAm AnandTech 18 ___1,411,240
3 BOINC@AUSTRALIA 15 ___504,102
4 Czech National Team 12 ___403,425
5 Rechenkraft.net 10 ___314,882
6 UK BOINC Team 8 ___303,545
7 Russia Team 6 ___279,248
8 Overclock.net 4 ___273,809
9 L'Alliance Francophone 2 ___48,201
10 Planet 3DNow! 1 ___42,200

From FreeDC
xii5ku 385,156
crashtech 257,074
biodoc 121,067
motqalden 79,809
[H]Skillz 67,351
10esseeTony 59,606
OrangeKid 58,585
Ken_g6 38,092
iwajabitw 28,962
zzuupp 24,702
Tejas 19,843
Howdy2u2 14,097
GLeeM 13,846
Kiska 6,571
Rudy Toody 6,471
MarcelliusIkaP 5,948
Fardringle 682

 

[Orange Kid]

And I am running out of work as all my uploads AND downloads are not happening.

 

[Howdy]

And I am running out of work as all my uploads AND downloads are not happening.

My uploads/ downloads make it look like I'm bunkering...........................

 

[TennesseeTony]

Nooo....me Rum barrel is runnin' on fumes!

5820K is dry.
The twin Ryzen 7's are dry, a few uploads stuck.
ThunderStrike is stuck in up/down mode, no crunchy, SSL connect error, the log says.
WUSS and Devastator still running, but uploads are building up.
ThreadRipper is still running, but uploads are building up.
Quad core Xeon Skylake is still running, but uploads are building up.

 

[StefanR5R]

Nothing has been updated for hours...

 

[StefanR5R]

Until a few hours ago, I got only many HTTP timeouts and some successful transactions. Now I get this:
"Scheduler request failed: Server returned nothing"

But there are some successful requests still, like so:
"Reporting 20 completed tasks
Not requesting tasks: too many uploads in progress
Scheduler request completed"

 

[zzuupp]

Until a few hours ago, I got only many HTTP timeouts and some successful transactions. Now I get this:
"Scheduler request failed: Server returned nothing"

But there are some successful requests still, like so:
"Reporting 20 completed tasks
Not requesting tasks: too many uploads in progress
Scheduler request completed"

Yup, at lunchtime I had almost caught back up with downloads & uploads. I am falling back again.


On the bright side, it is nowhere near as bad as before. 1/2 page of tasks versus .... many many of them.

 

[ao_ika_red]

Until a few hours ago, I got only many HTTP timeouts and some successful transactions. Now I get this:
"Scheduler request failed: Server returned nothing"

This is not Sunday morning the project admin was hoped for.

 

[TennesseeTony]

This little server keeps chanting "I think I can I think I can I think I can..." I'm impressed to see it has made a comeback for the most part!