Server Problems?

[OberonBob]

Something wrong with the server? I am unable to access the web site or submit any blocks.

:confused:

[vaughan]

What's wrong with SoB tonight? I can ping successfully but my clients cannot submit completed blocks. Also, contrary to how I thought this application worked, the client doesn't continue processing the remaining 170+ blocks that remain. Instead the cEMs/sec slowly drops over time.

[Jammy]

Gee . . .I thought that somebody over here may know why?

Jammy



Originally posted by OberonBob
Something wrong with the server? I am unable to access the web site or submit any blocks.

:confused:

[olaright]

Same here :(

Ola

[OberonBob]

Looks like it's fixed now.

[Jammy]

So . . .the two and a half hours of work I did before I turned off SOB are lost?????

Jammy




Originally posted by OberonBob
Looks like it's fixed now.

[ceselb]

You will loose some work, yes. But 2.5h seems to be a bit much. I usually drop about 10-25% of a block, less than 5 minutes on my PIV 1.5Ghz.

Maybe you should run the sieve as backup if the client stalls (I do, and it works great).

[Jammy]

I run MathGuys handler. I do not know how to figure out how much I lost so I am just writing it off. Does the server go out often?

Jammy


Originally posted by ceselb
You will loose some work, yes. But 2.5h seems to be a bit much. I usually drop about 10-25% of a block, less than 5 minutes on my PIV 1.5Ghz.

Maybe you should run the sieve as backup if the client stalls (I do, and it works great).

[FoBoT]

my clients are all stuck, so even though the server is "fixed" , i have to reboot all these pc's :swear:

there is no holy grail of DC projects, is there :bonk:

[Stricker]

what i did to temporarily solver the problem on my end was turn off send intermediate blocks

this allowed me to continue processing overnight however i did loose a couple hours time since my comps didn't work on anything i think the time period for one machine was almost 3 hours

[Jammy]

My boxes were down from 6:20 pm yesterday evening until 4:14 am this morning.




Originally posted by Stricker
i did loose a couple hours time since my comps didn't work on anything i think the time period for one machine was almost 3 hours

[Stricker]

right i set my machines to not send intermediate results about 3 hours into the outage cause that is when i first saw that it was down

Originally posted by Jammy
My boxes were down from 6:20 pm yesterday evening until 4:14 am this morning.

[vaughan]

So let's get this straight, if you have "send intermediate blocks" on then the programme not only stalls when the server is down but cannot recover when the server comes back up? And I thought that SOB was a pretty resilient DC project.

[Stricker]

no clue about when it comes back up
:machgun: :mouserun: :sniper:

[MathGuy]

Windows client: when the server is down, *usually* it just keeps going, and recovers fine when the server comes back up. The stalling problem on the Windows client seems to happen very infrequently, but doesn't have anything to do with whether or not the server is up. For some reason unknown (to me) this server outage was different from the ones in the past - some clients repeatedly stalled when contacting the server (although they were fine when the server came back up) and others made it through the whole thing fine.

Other clients: there have been reports of some of the other clients having routine difficulty transmitting intermediate blocks (whether the server is up or not). Many people just run those with "transmit intermediate blocks" turned off for that reason.

[vaughan]

Thanks for the info MathGuy.

[Stricker]

i'm still having a slight problem my linux box clains it is reporting results to the server but i'm not getting credit for the work it is doing

it is the second fastest of my machines and i think it is doing over 100k/s but yet my stats aren't showing what they should if if the results were being turned in correctly

so does any one know why my current results aren't showing the work that my linux box is doing?

[Frodo42]

It seems that this problem has caused quite many clients to stop submitting work ... overall production has dropped to somewhere around 110 from 150, something like 25% have not yet restarted the hanged clients, hopefully we will be back op around 150 again soon ...

[Stricker]

well like i just tried to say some clients were restarted but for some reason their stats aren't showing up correctly anymore

[MathGuy]

Don't know if this is the *right* explanation for what you're seeing, but remember that the "current" rate for a given client is really the average rate over the entire Proth test that the client is currently working on. Thus, I would expect that the server outage would cause everyone's stats to appear low until they finish their current Proth test and get a new one.

[OberonBob]

I also think that is what we are seeing.

[Jammy]

Uh . . .my three clients bounced right back after the outage. I think that my total cEMs dropped about 2.5 ponts but I was able to catch back up quickly.

Uh . . .where does one find the current average rate of their client at?

Jammy



Originally posted by MathGuy
Don't know if this is the *right* explanation for what you're seeing, but remember that the "current" rate for a given client is really the average rate over the entire Proth test that the client is currently working on. Thus, I would expect that the server outage would cause everyone's stats to appear low until they finish their current Proth test and get a new one.

[jjjjL]

my understanding is that the clients run w/o stalling when our DNS goes down or some other internet connection related problem exists.

i don't understand why the stalls happen either.

I think Stricker's graphs are working fine. The problem is that the system does not graph on a block by block basis, it's over entire tests, so the downtime you suffered is sorta averaged into your graph. This means that instead of having a hole w/ 0 production followed by a return to full speed, you'll have some average between the two until your computers grab two new tests.

same idea goes for the whole network. 25% of the computers aren't still down. a few may be, but odds are when people finish the tests they were working on during the outage, the graph will climb back up to where it was.

-Louie

[Stricker]

well what ever the problem for me was when i changed where my linux client reported to, from the domain name to the ip, it fixed the stats problem

i mean go look at the my stats now i made the change around 12 this afternoon and looking at the graphs as soon as i it reported the next block my stats started jumping

[vaughan]

jjjjL I'm having this problem.
My computers that are connected to my Cable ISP are unable to send completed blocks. They also cannot reach the SOB web site. I have tried www.seventeenorbust.com and sb.pns.net without success. I am able to transmit 1 block on a computer that is on a dial-up ISP.
I am on this forum at Free-DC using Cable. I have confirmed the server host is sb.pns.net in config. I tried changing it to 216.163.34.105 but still no success. I tried shutting down everything, the PCs, the router, the switch and the Cable modem and leaving them off for 5 minutes then turning them all back on. Still no luck. I have scanned for viruses using Grisoft's AVG, Trojans using DiamondCS's TDS-3 (registered), Spyware with Spybot search & destroy, Adaware 6 (registered) and PestPatrol 4.2.0.33 (registered).
Now what?

[Nuri]

Originally posted by vaughan
Now what?
Have you tried contacting your Cable ISP? May be, for whatever the reason, they have blocked the traffic towards project server (or something like that).

[vaughan]

This is their response:

"Thanks for contacting us. You cannot ping that server, the ping traffic is filtered out by their router. I cannot load the site from a non-cable network either, I assume that the site is having problems, or is offline."

I have e-mailed Michael at SOB to see if he can get it working again. I have been able to connect until Tuesday this week. Weird!

[vaughan]

Update - its now working. Michael said he made no changes. Optus didn't have the courtesy to reply to my e-mail. I guess that's what you can expect for over $130 a month in cable access fees. :)