Server having problems?

[rshepard]

For a couple of days I've been getting the "server is restarting" msg every time each of my clients tries to connect (both from home and at work) Now it looks as if the stats are broken; also; the clients have continued to do work but it looks like nothing is being recorded at the website. :Pokes:

[PlanetScorp]

that would be an answer for my first-time-run problems decribed here (in the middle somewhere).

http://www.free-dc.org/forum/showthread.php?s=&threadid=5658

[michaelgarvie]

Firewall unblocked. Thanks for your patience!
:smoking:

[rshepard]

Thanks!
:notworthy

[PlanetScorp]

thank you, connection works fine now! :blush:

[michaelgarvie]

Web server down today due to migration from department web server to university web server.

The evolution server is running though and stats are being collected.

The new server is hopefully compatible with PHP PNG graphics and would allow graphs to be drawn from the statistics.

[rshepard]

Some of my clients are showing "timed-out" msgs when trying to get work from the server,
and it looks like the stats on the website have stopped updating.

*hunts for server-beating stick :Pokes:

[VirtualAdept]

Originally posted by rshepard
Some of my clients are showing "timed-out" msgs when trying to get work from the server

My client has been getting timeouts only (java.rmi.ConnectIOException, java.net.SocketTimeoutException: Read timed out) from the server since ~9:30 CEST this morning, too. Just when I had this remarkable species... :cry:

[michaelgarvie]

.

[rshepard]

Timed-out errors are back today, and the stats are hung up again:cry:

[michaelgarvie]

Thanks :blush:

[rshepard]

stats are stuck again :Pokes:

[michaelgarvie]

Thanks again.. There must be some Linux/Java limit the server is hitting..

[rshepard]

stuck again :bang:

[michaelgarvie]

And back with some optimisations which should mean you should see generations go by quicker.

[rshepard]

looks like it is down again :cry:

[michaelgarvie]

Power cut yesterday 7pm... :swear:

[rshepard]

stats page is stuck today :help:

[HitchHiker]

Can't get any jobs either...:bang:


Trying to get task from server //139.184.166.27/KingKong
java.rmi.ConnectIOException: error during JRMP connection establishment; nested exception is:
java.net.SocketTimeoutException: Read timed out

[dreplogle]

I don't suppose any of the moderators could suppy an ETA (Estimated Time of Access?). :trash:

[prokaryote]

:( Sorry, can't say since I don't have any info regarding U of S systems. I'm thinking that since Miguel hasn't responded yet that it's system wide over there.

[dreplogle]

I've still got two windows open processing something. Generation times are increasing. I wonder though if I'm just spinning my wheels and getting nowhere? I can't open any more windows on the other pc's I use. Anyone wondering the same thing?:taz:

[rshepard]

Islands that are running when the server goes down should continue to evolve - they just won't interact with other islands until the server comes back. You can't start up new islands without the server to hand out work, however. At least,that is how I understand it.

[prokaryote]

:( Looks like server issues again? Hope things get sorted before the weekend.

[prokaryote]

Looks like somemore issues....

Any chance that the project could get a dedicated server?

[adream]

Originally posted by prokaryote
Looks like somemore issues....

Any chance that the project could get a dedicated server?


i second that, what are the bandwidth requirements of the project server ? (not the web server)

regards

adream

:elephant: :elephant:

[michaelgarvie]

This time there's an interesting story:
It looks like some naughty stats-climber installed a client on some company's machine without their permission. This company then issued a complaint to the university of abnormal traffic between the DHEP server and their machine thinking there was a trojan involved. No one belonging to the company has owned up to being involved with DHEP!

Naughty naughty, who could this have been? :spank: :haddock: :bonk: :whip:

Server is back online now.

[adream]

wasn't me !!! :bouncy: :bouncy: :bouncy: :bouncy:

thanks for the info miguel

thats bloody naughhty aint it ;)



adream

[HitchHiker]

Is the server in trouble again? :help:
One of my machines hasn't got a task for more than a day now, it may be longer than that. Stats in ranking indicate I'm not the only one suffering. :D
My two other machines are still slowly but happily churning away though.

[rshepard]

stats are frozen again

[jfk]

I am still getting the Java time out error. Anyone else?

[michaelgarvie]

Sorry about this. Back up again. :bang:

[em99010pepe]

I like this project but why the server is always down? It's funny but the server goes down everytime I have to restart the computer.

Carlos

[jfk]

Probably what is happening is that the servers go down and your island continues to evolve without communicating to other servers or islands, then when u restart and it becomes necessary to contact the server to get new work, you become aware of the down server.

[em99010pepe]

I ran all night and beginning of morning (about 14 hours) but didn't get credit for circuits evaluated. Strange!

Is Sun java J2SE 5.0 Beta 2 quicker than Sun java J2SE 1.4.2 SDK?

[rshepard]

hung up again-- seems to always break down when the topology map is 33x33 :confused:

Also, what is up with the Marvin island?? It seems to be eating up all the empty space and not returning any info???????

[nickth]

Originally posted by rshepard
hung up again-- seems to always break down when the topology map is 33x33 :confused:

Also, what is up with the Marvin island?? It seems to be eating up all the empty space and not returning any info???????

I agree with r shepard. the server is down again :trash:

[nickth]

Originally posted by nickth
I agree with r shepard. the server is down again :trash:

Now an hour latter its back up again :bang:

Very intresting :confused:

[em99010pepe]

Originally posted by nickth
Now an hour later its back up again :bang:

Very interesting :confused:

Are you sure?

[nickth]

yes i am for the moment anyway

[michaelgarvie]

rshepard was right, the problem was Marvin. This island was just creating hundreds of spurious connections. :haddock: I this kind of thing keeps happening I'll have to write some code to handle it. Sorry about this, server is running fine now.

[jfk]

The stats show Island Not currently active, eventhough I am still chugging away. Are we down again?

[rshepard]

time to kick the server again, Miguel :D

Also, would you look at the overall rankings ( Total Circuits Evaluated)-- I seem to have disappeared from the list :cry:

Thanks

[michaelgarvie]

Server up again. :spank:

rshepard, you're in position 306. Is this not correct?

You're still in top 5 for effort.

[rshepard]

No- I was #1 for quite a while, just waiting for diGriz to run over me. I think it fell off sometime after I broke through 2,000,000,000 circuits. Some sort of overflow condition maybe???

[michaelgarvie]

Hmm.. Funny that the overflow didn't hid Bolivar. Do the stats look better now?

[rshepard]

Much Better!!

Thanks :thumbs:

[prokaryote]

:bonk: Server needs some coercive violence to get it going again. :D

[deranged128[OCAU]]

A quick dose of percussion treatment appears to be in order. :trash:

This brings me back to a question in another thread I started 'Significant Downtime" where I asked about possibly saving completed work to disk during extended server outtages and the ability to view the java output to know if significant amounts of work are waiting to be uploaded.

I'd also like to get an idea on how many servers we are communicating with for uploads. Is there are need for perhaps additional upload points with the ability to cache work on them before finally reconciling with the main server? Sort of like the old seti-q idea.

I'm keen to keep supporting this project, which I consider to have very worthwhile aims, but am finding I have to divert CPU resources to other projects due to an inability to keep the client running at full capacity. :rolleyes:

[ace_dent]

Server down again?...

[michaelgarvie]

Server going for full restart 10:37 GMT.:bang:

[prokaryote]

:bang: Blinkin Server is down again... It's not running Windoze Server 3 is it? Or maybe someone should give the little hamster in his wheel some water!:crazy:

P.S. Miguel, American_Maid_2 is trying to connect to the server (been doing so for about the last 4 days) Can get to the rest of the internet no probs. Thanks, prok.

[michaelgarvie]

Server was shut down yesterday by Sussex IT people without warning. They say it was attacking another computer on campus with a DDOS.

[deranged128[OCAU]]

I might have to come back to this project when the server situation is resolved. :rolleyes: I'll keep watching this thread for news. :cool:

[ace_dent]

I was surprised that Island topology was not reset when the server was fixed. (Forgive me, I'm new to this project). With such an empty 'world' (only edge islands), wont migration and evolution be hindered?

Regards,
Andrew

[michaelgarvie]

Server back on-line. Looks like the attack thing was just a scare :haddock: . Thanks for your patience.

The ring topology is because the server was never actually turned off, they just disconnected the network socket and filtered it out of the univiersity firewall. I'll just be double checking for any worms and everything should be normal.

Migration should be going round the ring like on a motorway.

:cheers: Miguel

[ace_dent]

... and it's down again... :Pokes:

BTW- Is posting server outages useful?

[michaelgarvie]

And up again. We'll now be focusing on ways to stabilise the server.

[deranged128[OCAU]]

Originally posted by miguelgarvie
And up again. We'll now be focusing on ways to stabilise the server. I asked a question earlier in this thread about ways of getting more server points available. Is that possible?

[michaelgarvie]

It would be a nice extension to http://distrit.sf.net . If we get a good proposal from a coder on how they would do this, we would gladly accept it into the mainstream code. :cheers: Miguel

[ace_dent]

Down again...

Cheers.

[em99010pepe]

There's some kind of problem with the stats. Look at the numbers?

Carlos

EDIT

Problem fixed

[ace_dent]

Cannot connect...

Cheers,
Andrew

[michaelgarvie]

Up again. Power had been cut in the lab.

[ace_dent]

Hmmm... just what is going on with that pesky server?

After going down this morning, all of the crunchin' of the previous day was in vain- is there no way to recover to the last evolved state? Also, after the last reboot, the server time was actually correct- now we are back to +1:00 (well in the UK anyway).

Cheers,
Andrew.

[michaelgarvie]

Hi Ace,

Thanks for your message about the time. It's been corrected now.

The run has been restarted deliberately because it had been converged for two days and nothing was happening. A new measure has been introduced to avoid this and seems to be worked very nicely. Basically an immigrant won't be allowed into an island unless it is 0.08 better at fitness p1. This creates speciation as several islands are close to the best global best become immune to be overrun by immigrants. This avoids premature convergence on local optima. By watching the cluster it seems this is a pretty good strategy because it happens often that the global best comes from one of the top five global best, so its good to keep the diversity there.

:cheers: Miguel

[ace_dent]

A very interesting answer- thanks for the reply.

Andrew

[ace_dent]

No very important, but server time is +1:00 again.

[prokaryote]

:eek: Looks like IT shut down the server again this weekend!

Getting kind of old, these problems almost always happen on a Friday and aren't resolved till the following Monday. Is there anyway to check to see if the server is up over the weekend and give someone a call? I've got a couple of systems that I have to reboot and the project will endup losing that work since they couldn't check into the system. :rolleyes: :bang:

[michaelgarvie]

Server is up. There was power failure in the department over the weekend. They still haven't sorted out the webserver so the dhep website is down. However the DHEP server itself is up and running and collecting stats as I write.
:cheers: Miguel

[ace_dent]

Server down again?...

[HitchHiker]

Again and again, what kind of server is this?

Miguel, would you be so kind to answer ace_dent's question of 6 weeks ago: "BTW- Is posting server outages useful?" and react to the remark of prokaryote that this always seems to happen on a friday (except today of course). :help:

You seem to be a man of little words. Spend a bit more of those please.

[ace_dent]

No crunching this Christmas? :Pokes:

[stanray]

Well, seems to be lot's of questions about status and no answers to them given. Let me know when it is a go with this project again. See ya!

Stan:swear: