surveill@home

[AMDave]

http://surveill.dei.uc.pt/surveill/

web crawler
non-cpu-intensive

team created

forum indicates various problems with the clients at the moment

[AMDave]

not getting work at the moment:
"Mon 25 Jul 2011 19:17:30 EST surveill@home Message from server: Project is temporarily shut down for maintenance"

but account creation is open

[NeoGen]

Joined in, but no tasks in sight right now

[Nflight]

Joined and I have one work unit crunching, will see how it goes! :blob3:

[NeoGen]

Been getting a few tasks here and there... but it seems I got to keep poking it constantly or otherwise boinc won't fetch more tasks on it's own... not sure why...

[AMDave]

I am observing the same behaviour.
It appears that there is less supply than demand.
This means that due to the distribution of OSs, my linux boxen are getting more work than the Win boxes.
But, yeah, I'm having to poke them all continually to get any work.

[AMDave]

It seems to be ticking-over nicely on all boxen now without intervention.
Hopefully you see the same.

One thing I do notice is that when this project is enabled on a Windows box and it is communicating with the server I frequently get the message "BOINC couldn't do Internet communication ..." yadayadayada...
It is always wrong because the projects are updating but something to do with this project is triggering the message.
It is only happening on the windows machines.
Do you see this too?

I posted the same question in their forum.

[Nflight]

Your not special it is working here too... That message of not seeing the internet connection has disappeared now, it was annoying.

[vaughan]

Yeah I was getting that message too on Windows machines. I thought it was triggered by not having enough RAM available. I see it lots on the machines running NFS and BURP.

[AMDave]

Someone else replied on the project forum confirming this as well.
He suspects that the project client app is hogging the TCP/IP stack.
I see no reason to disagree.

If it starts getting in the way of things I may choose to suspend it to get 'normal' things done.

I had been wondering if it was interfering with the Windows QoS service on my LAN, but I'm satisfied with the current answer.

[AMDave]

Your not special... I missed that the first time I read it. Feelin' the burn. :) he he he.

[AMDave]

Surveill just updated their clients (http://surveill.dei.uc.pt/surveill/forum_thread.php?id=30#163) and it carries a new feature
It appears to check how many clients are currently running on your LAN
(NB - I think they probably do this on the server by checking which clients were downloaded to hosts on the same ip address, as opposed to 'sniffing' your LAN packets, which might be a bit nefarious)

You may see this in your BOINC manager log
"Wed 17 Aug 2011 12:37:50 EST surveill@home Message from server: Not sending work - max number of probes in your network exceeded"
followed by
"Wed 17 Aug 2011 12:38:27 EST BURP Message from server: This computer has reached a limit on tasks in progress"

Smart move Surveill :)

[vaughan]

That's a shame for those running big pharms. I guess a workaround is to split the pharm over several ISPs. Luckily I have three at my disposal.

[Nflight]

I am not seeing a problem here at all. BUt, I only have two systems operating at a time!

[AMDave]

It looks like they were also trying to address the BOINC Manager internet connection popup message on Windows at the same time:
http://surveill.dei.uc.pt/surveill/forum_thread.php?id=24#125

Funny thing is that since they limited the number of concurrent cliant apps on the same network I just cut them all back to my Linux boxes.
So can someone else confirm that they are running the v1.05 client app and that the pop-up message has stopped happening?

[AMDave]

I am not seeing a problem here at all. BUt, I only have two systems operating at a time!

Just checked my boxes and it looks like you have the 'sweet spot' at 2 instances running.
I appear to be down to only 2 running at a time.
That means I can shut down the VM on the other box and try to put BURP on it too!

[EDIT -
incorrect assumption - it might be only 1 I already shut down that VM last night. I forgot
Blast.
I have zero running and all saying there are too many.
I'll need to figure this out.

Allowing work on a windows machine.
1 windows machine just got one right away.
Maybe a Linux client problem.
I'll try detaching all the windows clients and see if Linux picks one up somewhere...later
- EDIT]

[EDIT2 -
forgot about my power outage. duh
One linux box still had a WU that was not returned and showed "Aborted. Not started by deadline"
I guess that one was blocking the rest.
- EDIT2]

[AMDave]

Follow-up:
The pop-up message in windows appears to be gone.
My linux machines do get work.
It looks like only one network probe runs at a time now.

[AMDave]

Hmm. you guys are all getting much better results than me and I don't know why.
Anyone have any clues?

Are you getting multiple WUs running concurrently on the same network?

[AMDave]

It's ok. You can play your cards close, if you like :P

The client is back to running multiple instances on the same network.
I'm allowing new work on some windows machines again now the popup issue is fixed.
It was annoying the family so I had to disable it on their machines.
I'm coming back atcha.