Is it this bad for anybody else?

[blcjr]

I don't think Berkeley has accepted a wu from me in over 24 hours! I've now got 317 wu's waiting for upload. I keep getting messages like this in my log:


5:18pm: s@h_wrk Sending Result: SendRequest to shserver2.ssl.berkeley.edu failed: Connection Timeout


I know about the bandwith cap. Is that the cause of my woes? Or is there something else I should be looking into.

-Bazilla

 

[Engineer]

Baz,

That seems odd...as my Queue is sending...very, very slowly...but it does send. Have you restarted the Queue?

Even OK's is sending some in....slowly, but it does send. It sent over 800 between 5am and 8am this morning.....

If you need WU's, let me know and I'll let you point to my Queue and give you a few

Hmmmm

 

[Poof]

bazilla - I'm not sure what time zone you're in but set your queue to upload at the equivalent of 3am - 6am PST.

[EDIT: And if you haven't already done so, edit your settings for "Overide for pass-through operations" to "disabled"]

 

[blcjr]

Engineer...

I've still got a couple more days worth of WU's cached. My problem is not downloading them, it is uploading the results.

Poof...

I have my settings so that they will upload results as they are received. So that will include the more limited 3 hour setting you suggested. As for "override for passthrough" I did what you suggested, and disabled it. What does that do, and why would it account for the problems I am suddendly having?

-baz

 

[Russ]

Usually my Q gets backed up during the day, flushes at night, and is empty when I come in in the morning. Not so today. Was still trying to unload.

Russ, NCNE

 

[Tarca]

I have the same problems, There are always completed units in my Queue but none over 24hrs old.

If it gets much worse I might have to change to another project......

.
.
.
.
.
.
.
.
.
.
.
.
.
.

NOT

well not yet anyway.

 

[Jeff7]

My problem is that I can't receive work units; it lets me send them though - after about 10 tries. I wonder if SETI could find another host?

 

[Poof]

baz et al. I know you guys like to look at those stats change realtime... But the days of doing that are over unless they either:

1.) Take the bandwidth cap off
2.) Setup some type of mirror WU server and stats server thing
3.) Alter the client so that results take longer (ugh)

When I first started SETI, I liked seeing those stats tick upwards as results were uploaded. But after many outages (alot of them happening almost daily for months), I got used to just checking my stats once a day. Thus you get less frustrated that way. And with Setiqueue, you can at least monitor your results and total them up yourself.

By setting Setiqueue the way I mentioned (disabling "pass-through"), that means all your clients will still get fresh WUs from your queue but the results will stay in the queue until the time that you set for it to upload (and get new WUs from SETI). You can set a range for when it should try to connect, ie., to start at say 3am and continue to connect until 6am. After that, it will not try to connect again but will dish out WUs to your machines and accept results from them.

They key is to find that sweet time when most if not all of your results will be able to upload in one fell swoop! The way it stands right now, that will NOT happen during the daytime hours (at least in the U.S.)...

 

[panhead49]

poof has the answer.......unless you want to stay up untill 2-3 am......late nite TV here i come

 

[Poof]

poof has the answer.......unless you want to stay up untill 2-3 am......

I call it "time-shifting"... It's like setting the VCR (or TiVO or whatever... ) to record at a preset time during the middle of the night so you don't have to stay up to do it manually like...

late nite TV here i come

panhead49!

LOL!

 

[blcjr]

Forgive me for being daft, but I don't see the need to change anything. Say 3am-6am PST is the best time to upload. Well, with my current settings, Seti@ will try to upload during that time ... along with every other hour of the day. So if they'll upload then, they'll upload.

But it didn't work last night, that I can tell. Those hours were no better than the rest.

And now I've got 353 wu's accumulated.

I'm with Tarca. If this doesn't improve in the next couple of days, I'm switching to another project.

-bazilla

 

[IsOs]

<< I'm with Tarca. If this doesn't improve in the next couple of days, I'm switching to another project. >>



I'm sorry to hear that your problem continues. Are you running low in your cache of fresh workunits or just having STATSamania syndrome like me. I too have problem sending my completed workunits.

Is there anything we can do to help? Maybe send some of those to other public Q like TACUBE and Orangekid. That way, the more Q trying to send your workunits, the faster they will go through. Just an idea, I hope they will agree

 

[DarkMajiq]

You can try using my SETIQ (darkmajiq.dns2go.com:5001) if you want, it seems to be sending WUs in with only a very small delay.

 

[blcjr]

I'm sorry to hear that your problem continues. Are you running low in your cache of fresh workunits or just having STATSamania syndrome like me. I too have problem sending my completed workunits.


I've still got 390 WU's in my cache, good for about a day and a half. That's not the problem. The problem is that each time SetiQ tries to send a wu, I get the following result:

10:31pm: s@h_wrk Sending Result: GetResponse failed: Connection reset

Now I've got over 380 completed wu's stacked up. I sure hope things get better overnight. If some of my completed wu's get through overnight, I may come up with the patience to wait things out. But if, after another day and a half, I've used up the rest of my cache, and have 700-800 wu's that Berkeley will not take, I'm going to say the heck with it and move on to something else.

[edit] Isos, it is not so much statsmania as it is frustration that a previously smooth running machine is now not running smoothly. Or worse, not running at all.

-bazilla

 

[IsOs]

Go ahead and send some of your workunits to other public Q. That will give it a faster chance of being accepted by Berkeley. Dark Majiq seems having a better connections, try sending some to him.

 

[IsOs]

FYI, workunits are currently being accepted at a very slow rate.

 

[Fardringle]

It looks like OK's is having the same problem. It's sending results back to Berkeley, but we're sending them to OK faster than his queue can dump them...


"The following 1,186 Results are Queued to be sent to the Seti@Home server" (that's from his pending results page) :disgust:

 

[IBhacknU]

Baz,

is there a timeout setting on your queue? I think there's an option for how long the queue waits to retry after a timeout. It should be set to something low, so it keeps trying, and trying, and trying.

I've had some strange nights too when nothing has gone through though, so go figure?

And don't worry, the WUs will be accepted. I mean, I sent in ones that were a week old (remember?). They took those just fine. Just increase your days of WUs you keep on hand and you'll be fine.

?

 

[conjur]

I'm flushing/filling just fine right now. But, I'm still on dial-up. I swear whenever I have a problem with the cable modem it takes the local support people a MINIMUM of three days to get it fixed!! :|

 

[Engineer]

Baz,

I don't mind if you want to try to point your Queue towards mine and flushing to see if we can get this open...

I'll give you a couple hundred WU's to start with and mine seems to be flushing fine late at night...

If you want to try...

host: federation.2y.net
port: 443

Let me know....you can then send the WU...it will create a Queue for you on mine...I'll move some WU's over to yours...then you can transmit again....

Thanks

 

[IndyJaws]

I had a pile of WU's still in my queue when I got to work this morning (appx. 7:45 AM EST), but I just checked them about 5 minutes ago and now they're all sent. Hopefully you were able to flush here recently as well.

 

[Poof]

Seti@ will try to upload during that time ... along with every other hour of the day. So if they'll upload then, they'll upload.

But it didn't work last night, that I can tell. Those hours were no better than the rest.

I'm hoping that this morning, some of your results were uploaded.

However, by allowing it to keep trying all day and all night, it not only adds more traffic to the Berkeley network, but it activates an internal retry timer within Setiqueue that might throw you off from getting a solid connection at the right time on their data servers.

The network graphs for the campus can be seen here(select the one for "SSL" under the "offsite" group, plus there are graphs from some of the border routers). When you check them out, you will clearly see the peaks and valleys of network usage and should be able to re-direct your uploads during those valley times.

I am sure that no one on the TeAm wants to see you give up and leave. Anandtech's rise through the ranks over the past couple months has been phenomenal and due to the efforts of folks like you. Most of the veteran team members here have been through worse and are still alive and kickin'. Everyone is willing to give you a hand to get those precious results uploaded and counted towards the TeAm totals.

Just give 'em a chance.

 

[blcjr]

Indy...I just got up and checked the queue. It is down to "only" 325. So I was able to upload some of the stacked wu's overnight. I just ran setiteam, and I'm showing +150 wu's since Tuesday evening. Perhaps even more alarming, I'm only showing +3850 for the whole team in over 48 hours. That's little more than 1/3 of what has been our recent daily production. So this bandwidth cap is hurting everybody, bad.

-baz

 

[blcjr]

Engineer...

How would I point my q toward yours? Isn't SetiQ hardcoded to upload results to Berkeley? I once tried to find a setting for the URL used for uploading results, and never found it.

-baz

 

[micron]

<< How would I point my q toward yours? Isn't SetiQ hardcoded to upload results to Berkeley? I once tried to find a setting for the URL used for uploading results, and never found it. >>

In the web version you click on settings then you set "Connection Type" to WinInetProxy and input whatever proxy server and port you want to use. Click submit and it should work!