I don't see how it could be any one persons fault. Guess they (Predictor) didn't see AMD Users rally coming and were not ready for our enthusiasm.Originally Posted by Beerknurd
I don't see how it could be any one persons fault. Guess they (Predictor) didn't see AMD Users rally coming and were not ready for our enthusiasm.Originally Posted by Beerknurd
Could be.... WE RULE!!!!!
I was uploading / working on a few things as it went down.Originally Posted by Beerknurd
We, as a team (no names mentioned)... did a major upload and download request at nearly the same time (self included).
I think we hit it with several hundred all at once. I saw the download rate die after that... and it crashed got page/swap/memory deprived all at once. The rest is history.
If we are not responsible, we definately pushed it over the edge
my 'stand in' p4 just started it's last WU.... My AMD (thanks to my fast line and proximity) got about a day's work.... so I am continuing to process... but no new work coming down yet.
I am running in GUI mode for now
I will advise as I learn more.
I will ask that more AMD WUs are created... the Intel/AMD settings now are based on a hard number... I'll see if it can be changed.
BC
I've got 5 machines all dried up![]()
I think I'll login remotely to them and switch them over to something else until Predictor gets re-animated.
I'll keep watch on the boards for signs of life.
It's 8am here..... BC, it must be stupid o'clock in morning there :shock:
No, it's only 12:49am.... "Stupid o'clock" starts on Friday and ends Monday, or was it the other way around? DUH!Originally Posted by Ototero
It is strange here... standard dinner time is from 7-9pm. But, this is also Lost-in-the-fog Angeles!!!!
Back east (Pa)... Dinner is at 5:30 or so... 6 at the latest.
The cultures are 180 degrees opposite. East is Purtian, West is 'loose'.
I enjoy many aspects of california, but miss the 'methodical' thinking of the east coast.
Predictor: I am down to only 1 machine running Predictor now... and it has about 3-4 hours work... that's it. The WU database should be up and running again by 8am PST... probably sooner if I know our PM....
This is one VERY dedicated staff.... I like the way they handle things.
They are up front about issues; quick to fix; and good at responding to questions / problems.
*** When I spot work available, I will make the announcement immediately ******
BC
Bc
What happening with Predictor at the moment i'm only getting a few hours worth but i'm requesting four days worth of work. and then there are long gaps when no work is avaible.
the good news is i've 700 points in pending. carn't wait to get those :D
I am working on the Apache code..
We have isolated the problem to a specific Unix/Linux system call
that is different on Seti (BSD) and Redhad (SysV).
I am working on writing code for the correction now.
Once done, we will be back up.
The problem is simple.... Apache creates sub-process CGI... CGI creates the handlers for our work... program CGI is not correctly handling the normal exit of the handlers. This is what I am fixing.
(note: program 'cgi' that Seti & Predictor use is NOT the 'stock' Apache version. This is the root of the evil)
*** I *WILL* fix it today and we will be running again.
BC
Thats great news that the problem has been found :D
Its really intresting to find that this program is better suited to the AMD prossesor that Intel.
Which means next month i'll be upgarding both my 2400+ and 2600+ to the 3200+ :D
it means by next month, we will be on top of the team list.
ROFLMAO..
Now, back to work for me.
FYI: For Laughs... DUAL Xeons could not keep up with our upload rate on redhat... LOL...
I'll work on the OS level next and MAYBE see about a hardware change.
(Dual Opterons?????)
BC
No offense BC, but Predictor is not stable. It seems to be down more than it's up. But I know your doing your best to get it up. I'm just being an ass.