Alrighty, time to give it another go!

[Careface]

Well, I'm back from my little "SETI holiday" which I was forced to take for a few reasons (money, mainly. Since the fianceè left me last year, money has been somewhat tight, but hey, new year, new beginning, right? :) ), and I found there was a nice little fix to the issue I'd been having with CUDA and making video lag like crazy.. annoyingly enough, the "fix" was added about a week after I quit SETI :P typical. Either way, the <exclusive_gpu_app> tag in cc_config.xml has been amazing :)

7/02/2011 4:24:45 p.m. Config: don't use GPUs while mpc-hc.exe is running

So, huzzah! Looks like I'm back in the game. A quick peek at the SETI site shows we've had a bit of trouble over the months? :(.. I hate to ask, but.. the project isn't dying, is it? :(

[Pete Church]

Far from dieing, but yes that had some serious hardware issues. But through donations etc they've gotten the hardware situation under control, until the next issue of course.

And welcome back!

[Careface]

Well, that news is as good as can be expected, I suppose :) Thanks for the welcome :D

[Saaby900t]

They I think the problem is GPU's can crunch way more data than cpu's and with the optimizing apps and such we are overloading/asking to much from the old parts of the system.

[QuietIce]

:welcome: back!!!


A quick peek at the SETI site shows we've had a bit of trouble over the months? :(.. I hate to ask, but.. the project isn't dying, is it? :( It's in, what I would call, the last stages of a major restoration. They got a couple of new servers last Oct and started transferring everything over, which, of course, involved moving all the tasks from one server to another in a domino effect. ((You know, the worst server had the lightest load, best server the worst load - two new servers go to the head of the line and all the others move down. It's been an adventure to say the least.)) Took them a month to merge, transfer, and synch the main database. They replaced the power into the server closet while they were at it, added some other basic hardware that's needed doing but took a shut-down as well, had a new network line run to the building - lots of new goodies! But it did take awhile and there are still adjustments being made as they come across certain issues.

On top of that there were the usual things that happen. For example, the data collector on top of Arecibo went down so no data was being collected for awhile. Everything's looking fairly good at this point, though. Weekly maintenance has changed to ~1200 GMT Tuesday to ~1200 GMT Thursday so make sure the caches are set right! :)

Good to have you back, man! :beer:

[Careface]

Thanks everyone for the welcomes ;D :beer: / :soda: for all!

Interesting news QIce.. Guess I came back at just the right time then :) Woot!

[zulfy26]

I just started again too. I had a random question, is there a way to request like a week of WUs so I can keep crunching when SETI servers goes down?

[eaglescouter]

Adjust your cache to hold an additional 10 days of work. This setting is located in the web account.

[zulfy26]

Project down for database maintenance D:

[Sephis]

You can adjust the cache in the BOINC manager program as well. Go to advanced/ preferences/ network usage tab and set your additional work buffer to whatever you want.

[zulfy26]

You can adjust the cache in the BOINC manager program as well. Go to advanced/ preferences/ network usage tab and set your additional work buffer to whatever you want.

Yeah I think that is only the local machine? Oh and how do you use the Advanced > Select Computer... option? I can't ever get it to connect to any session other than localhost

[Sephis]

Yeah I think that is only the local machine? Oh and how do you use the Advanced > Select Computer... option? I can't ever get it to connect to any session other than localhost

I dunno, never used it. I just use logmein to remote control my rigs from anywhere with an internet connection.

[Careface]

Yeah I think that is only the local machine? Oh and how do you use the Advanced > Select Computer... option? I can't ever get it to connect to any session other than localhost

There's a couple of ways - I just used to use the "--allow_remote_gui_rpc" flag when starting the client which lets any PC connect to it (useful if you have a a dedibox with no monitor that you only want to set up once and leave, but then add more to your farm and want to connect to it from the new rigs), or you can create a file "remote_hosts.cfg" in your BOINC data folder (not sure where that's located these days - I kinda miss when it was all in 1 folder) and put the IP addresses (not sure if it works for IP ranges, but I'd imagine probably not) on each line that you wish to allow to connect to the host. Note that while you can use host names rather than IP addresses, I always found it worked better with IP addresses.. I'm assuming because it could skip the DNS look up? Not sure. Either way!

After that, you should just be able to go Advanced -> Select Computer -> [remote host's IP]/[password] and you're set!

EDIT: Seems the data folder is in %appdata%\BOINC\ on a default win xp install.

[Pete Church]

You can adjust the cache in the BOINC manager program as well. Go to advanced/ preferences/ network usage tab and set your additional work buffer to whatever you want.

Doing this method was the only way that worked in the end. The account settings never seemed to take hold. But once i did the above in the boinc manager, i started building cache right away. So this is *my* preferred method. Maybe the others are having luck using the seti settings.

[Careface]

Doing this method was the only way that worked in the end. The account settings never seemed to take hold. But once i did the above in the boinc manager, i started building cache right away. So this is *my* preferred method. Maybe the others are having luck using the seti settings.

+1. The cache settings in BOINC manager are actually an override for the site prefs. Imo, the site prefs are more for the 'average' user, or the overly insane user (NEZ anyone?) who makes use of the very limited venues available (School, Work, Home..)

[Pete Church]

Yeah I think that is only the local machine? Oh and how do you use the Advanced > Select Computer... option? I can't ever get it to connect to any session other than localhost

Using the Select Computer option requires you to install a <BOINCdir>/gui_rpc_auth.cfg file with a plain text password in it. Or actually it may be in <BOINCdir/Data/gui_rpc_auth.cfg. I am now thinking its the latter directory. Then you have to have a <BOINCdir>/Data/remote_hosts.cfg file that has an IP/host name for all IP/hosts that are allowed to remotely connect to that instance of BOINC.

You put these two files on your remote machine, then you can go to your local BOINC manager and use the select computer option and type in the remote hosts IP/hosts and password (that you saved above) and then the remote BOINC mgr will look in the remote_hosts.cfg file of its instance and see if the incoming request is one of the allowed IPs and check the password against it's local gui_rpc_auth.cfg file.

If those match up, you can then manage the remote host locally from one computer.

EDIT: sorry Careface, i was reading the posts in order and didnt see that you explained some of this already.

Just FYI, there is a wiki that talks about this: Controlling BOINC remotely (http://boinc.berkeley.edu/wiki/Controlling_BOINC_remotely)
Just for completeness, another wiki that talks about other configurations that you can turn on: Client configuration (http://boinc.berkeley.edu/wiki/Client_configuration)

[Saaby900t]

I guess I missed the Window yesterday when Seti was up.:cry:

anybody Else notice that the servers are almost all off again.:-/

[eaglescouter]

I guess I missed the Window yesterday when Seti was up.:cry:

anybody Else notice that the servers are almost all off again.:-/

Project is off is an understatement.
Project forum is offline
Tech news is offline
Website is offline
SSL page is offline

Looks like they lost their pipeline connection to the world.

[zulfy26]

I was reading on their website yesterday before it went down that there would be an extended outage from Tuesday till Friday morning (PST). However, said message also said that the website and forums would remain up most of the time.

[Pete Church]

Well the Cricket Graph (http://fragment1.berkeley.edu/newcricket/grapher.cgi?target=%2Frouter-interfaces%2Finr-250%2Fgigabitethernet2_3;view=Octets;ranges=d) looks like data is now flowing in (up to seti) as of this afternoon. So maybe uploads are working now at least. I'm not home so can't see if that is the case. Won't really help much, i guess, since everything else will be down for a while it seems.

From the news it looks like they're making some progress, just have to have patience, I suppose.

[Saaby900t]

Well the Cricket Graph (http://fragment1.berkeley.edu/newcricket/grapher.cgi?target=%2Frouter-interfaces%2Finr-250%2Fgigabitethernet2_3;view=Octets;ranges=d) looks like data is now flowing in (up to seti) as of this afternoon. So maybe uploads are working now at least. I'm not home so can't see if that is the case. Won't really help much, i guess, since everything else will be down for a while it seems.

From the news it looks like they're making some progress, just have to have patience, I suppose.

What is even the problem?

[Pete Church]

I only understand a small portion of how their servers are interconnected and what they do individually. But one of the massive RAID drives took a dive and they're having to rebuild it. I also got the impression that there's some server that they're having trouble installing the OS on, or something like that.

[Saaby900t]

I only understand a small portion of how their servers are interconnected and what they do individually. But one of the massive RAID drives took a dive and they're having to rebuild it. I also got the impression that there's some server that they're having trouble installing the OS on, or something like that.

I think they figured out the OS problem. I think it is just the raid problem. I thought that they have been rebuilding the raid the last couple of days. How massive is this raid?

[QuietIce]

I think they figured out the OS problem. I think it is just the raid problem. I thought that they have been rebuilding the raid the last couple of days. How massive is this raid? If it's the same RAID array from last fall then it's a monster. IIRC, it took them a week to rebuild it from the ground up including transferring and verifying the data (their science db). I guess they're still having growing pains from the new servers and all the shuffling they did.

PS
The way SETI collects data from Arecibo is to simply record signals from whatever the 'scope happens to be pointing, so the data is pretty much random samples from all over the sky. Since they're looking for two samples from the same patch of sky it could be years before the 'scope points at any given patch twice. As such, SETI maintains results from years of processed work units for comparison as new data gets processed. The db is huge and growing bigger everyday.

[Saaby900t]

I guess I fail to see What the science database should have to do with current WU and current results being put into the system. I know they are tight for cash and such but why wouldn't you have the master database separated from the other database. And just During the weekly outage move the current results to the science database over the GB network.

[QuietIce]

I honestly don't know what challenges they may be facing or how they organize the process. If they're trying to rebuild the science db then it could be the server it's connected to is normally used for the WU process and I'm sure they wouldn't want it multi-tasking during the rebuild. That's just conjecture, though, I don't know any more than the SETI@Home forum reports - if that much, since I don't read all the posts. :shrug: I was just getting to the point where I recognized the server names and what they did - and now I have to learn them all over again!

[Careface]

I was just getting to the point where I recognized the server names and what they did - and now I have to learn them all over again!

:argue: I know that feeling lol. Just before I made this thread, I was looking at the S@H Status page and I saw names I didn't even recognise o_O They seem to have deviated from the "spacey"-type theme, such as Carolyn (unless of course I just don't get the reference :( )... and still, "thinman" just reminds me of the atomic bomb.. lol.

Not to mention there's quite a few new processes since I last checked.. ntpckr (good to see they got that up and running finally :D ), rfi, vote_monitor, feeder/db_purge.x86-64....:blah: