Everyone in ECCp Needs to Take a Dump

[XL]

I see alot of zeros on the stats page where alot of people aren't dumping. (I hope it's not constapation.):Q Everyone who hasn't dumped needs to do it. Might even put us in to a comfortable 2nd place.

 

[Orange Kid]

I think someone broke the outlet mine won't flush

 

[Rendus]

Just dumped what I had - Around 150 or so.

 

[RC]

To the best of my knowledge the ND server is currently down. I would recommend waiting to perform any dumps until the ND server is back up. I don't think anyone can send points currently.

 

[Rendus]

It does look like they're down - I started my eccp clients with -xmit a few minutes before I posted, and the DPs haven't been sent yet.

 

[Ken g6]

That would explain the lack of a stats update.

 

[kmmatney]

I'm all backed up myself. Hopefully some of my permanently connected clients haven't froze on me..

 

[Sorcerer]

I had to turn off 'perm connection' on all of mine. They were trying so hard to send, that they were only doing like 4k per sec. As soon as I told 'em to hold their DP's, they started crunching numbers again.

Might want to check all your systems!

 

[DnetMHZ]

<< I had to turn off 'perm connection' on all of mine. They were trying so hard to send, that they were only doing like 4k per sec >>



had that same problem on a few of mine.. hopefully this situation gets fixed pretty soon.
DnetMHZ

 

[RC]

Kinda odd - This thread was created just before the ND server went down, and now the thread topic "Everyone in ECCp Needs to Take a Dump" reflects the fact that noone has been able to send DPs for the last 15 hours.

 

[DnetMHZ]

<twlilight zone music>

very....interesting..

</music>

 

[MereMortal]

Ok, I have a question for those people whose client hangs when it can't connect to send DPs. Which client are you using?

The reason I ask is that I have never noticed this when I've had my clients set on permanent connection. I'm using the optimized client w/o all the proxy server stuff and other goodies (eccp109-127_win_k7). Obviously it could be me not paying very good attention, but it might be nice to know for sure.

 

[RC]

Answer: Statsaholic Windows client
In the Oct. 28th version of the Statsaholic Windows client, Virus changed the number of network attempts from 48 to 4, this may reduce the problem. I haven't tried the 10/28 version.

So the NotreDame Windows client does not exhibit this problem? I think the ND version still uses 48 attempts. Interesting.

[Update] Ok, I tried the 10/28 version, it hangs for 12 minutes and then resumes number crunching. So if you're using an older version of the Statsaholic Windows client the stall time would be at least 144 minutes.[/Update]

 

[XL]

<< Kinda odd - This thread was created just before the ND server went down, and now the thread topic "Everyone in ECCp Needs to Take a Dump" reflects the fact that noone has been able to send DPs for the last 15 hours. >>


RC

All part of my sinister master plan to save money Bwahahahahaha

 

[RC]

XL - Sorry to inform you but my clients did not freeze even though the ND server went down Bwahahahahaha

125k by Turkey Day 12:00 noon stats report and you buy a P4?

Ok

We are only at 101K currently, so it won't be easy but perhaps we can pull it off.

 

[XL]

Nooooooooooooooooooo!!!!!!!!!!!!!!!!!!!!!

 

[Logix]

Attempting to send...Failed

How long is this gonna last. And how are some people getting through?

 

[Rendus]

The eccp109 project news page says the cgi-bin access was disabled due to high load. Basically, someone probably megadumped (and I still don't understand the point of megadumps), prompting the admins of nd.edu's web server to disable Chris' cgi access for the DP receive script. He's trying to set it up on a box he owns now.

 

[RC]

Noone is able to send DPs currently to the best of my knowledge
Latest news on the ECCp server outtage

The Tue Nov 20 13:31:12 2001 stats report was based on DPs sent yesterday before the server went down.

 

[daweasel]

Rendus I'm sure what happened was there was a little server problem, which cause every client set to perm connection to basically DOS attack the ND server.

It's happened to Virus's proxy before, so I'm sure thats the problem.

As he said on his page, the problem will be there atleast until tomorrow.

 

[Rendus]

That could be the case as well, although that's something a kill/restart would fix temporarially. Dunno. Without more info, we can only guess

And I STILL don't see the point behind megadumps

 

[daweasel]

A kill /restart of the server? I agree, but if he would have had to done it before they cut his access.

I agree about mega-dumps, but at this point, the megadumps aren't big enough to hurt the server.

 

[virus]

The previous StatsAholic clients were based on 126 code with the 127 improvements added to it. I may have missed porting over a 127 improvement back then. The new beta StatsAholic client was built from the 127-4 source. If you are comparing the older statsaholic clients to the latest stock 127-4 or newer client. Then there may have been improvements to how it connects which would be causing your differences.

I really think there is something wrong with cgi script and it's max DP limit code. It seems that instead of it blocking a single IP, it's blocking all users. When I use my test script to send a DP to the ND server. It sends back a 'Internal Error (20)' message. Which if I remember right is the same one sent when the max DP limit is reached.

As for how much traffic the clients create when they don't get there way. Since ND's server went down. I have been averaging 10 connection at all times. At this moment I have 20 active connections from clients. Usually I block access to my script when this happens. So that my server can get something done. This time I have left my script accessable and disabled all background tasks. I deleted my web server's access log this morning and already it's up to 24MB.

 

[Logix]

<<I agree about mega-dumps, but at this point, the megadumps aren't big enough to hurt the server.>>

Well, there's going to be one HECK of a megadump if this thing drags out a couple days!

 

[RC]

One side benefit of the ND server outtage is that I was forced to create some scripts to pull the dplists from the clients on my diskless nodes. Didn't want to risk losing days worth of points. So in the future if I want to save some points for a dump it will be easy to do. Backing out the changes I made today would not be something I could do in just a few minutes time, so I may just start dumping once a day rather than continuously sending points.