TA Cube damage report...

[Russ]

Okay, I finally got (almost) everything back to normal around here. Pproxy has been back up for a while and everyone should have no trouble fetching and flushing. My internal pproxy is another story, though, but I'll worry about that tomorrow.

Now, when you check stats you will see that there are zero for the 28th. I'm VERY sorry, but the log file got corrupted. It must have been writing to it when the quake hit. What this means is that any blocks you flushed to the pproxy will not show in your totals.

But, this does NOT mean that they didn't make it to Dnet. The only blocks that would have gotten lost would be those that were waiting to be flushed to the keyservers, and this is never more then a few hundred. So, the vast majority of your blocks for the 28th made it.

I know that some of you have some little races going and not having a whole days production included in your total messes with that. Again, very sorry.

Now, time for this boy to head home. For some reason, I feel a little worn out.

Russ, NCNE

 

[mechBgon]

Glad you're ok, Russ. I guess some people felt it here in Spokane, so it must have been quite a time over there. Rest easy.

 

[networkman]

Yeah, I think we can let it go this time, but don't try using that "earthquake excuse" again!

 

[toph99]

good to see you're ok

 

[Bakwetu]

Glad to see you're alright Russ.

I joined the pproxy two days ago, I flushed some blocks yesterday and moved up 39 steps, but now I'm gone from the list But as long as they went on to Dnet it doesn't matter. I'll be back on the list tomorrow

Does anybody know if I can see the total output of my little herd? They are on different hosts (some of my colleagues' computers which I have assimilated).

 

[dennilfloss]

Russ will have to consolidate your hosts.

HE DA MAN!

 

[RaySun2Be]

Glad you are doing OK Russ;

 

[A2KLAU]

Cheers for sorting this out Russ! I was wondering what the hell was happeneing to TA Cube last night! I thought my RC5 had gone down!

Albert.

 

[Adul]

Whoa, late post Russ. I think it is time to get the TAcube pproxy on a UPS unit.

I guess your crack rack has to start all over

 

[Kilowatt]

The Pproxy isn't as important as you being OK.
I'm glad you pulled through it with no more damage then that.
/even if you did get my goat the other day



<< I think it is time to get the TAcube pproxy on a UPS unit >>

hehe That's not a bad idea, I'm running my 4 nodes in my Rack off one 400v APC.
/me thought the last node to fire up would set the buzzers off, but it didn't

There are utilities in Linux for auto shutdowns of systems running off a few name brand Battery Backups.

 

[Bakwetu]

Thanks Dennis
Now, I have to go and change some .ini files

 

[DanC]

Two checking in...
I must be the only person around who drove from Herndon VA to Baltimore, AND Portland to Seattle all in the same day.
SeaTac airport remains closed. Overpass damage in Chehalis left me a lovely 2 1/2 hour delay through 30 miles of traffic backup on I-5.

A couple of ceiling cracks - and a slew of machines offline - but as of this morning, we're back to 100% operation.

 

[dennilfloss]

Good to see your OK. DanC.

If that can make you feel better, I just rejected time on a Fujitsu supercomputer. :Q The guy claims he could break a whole 128bit encryption in 36 hours using Unix-based encryption software but that would go against the intent of the project, which is to see that it can be done with home computers. Finishing in 36 hours would ensure second place for us. Can't have that.

 

[RaySun2Be]

Can't he just run the DNET client for 36 hours?

The output of that beast would help us get back on top.

 

[dennilfloss]

Wouldn't that be cheating? Anyway, the output of that sucker would probably result in a gigaflush. I don't think dnet would like that. Maybe Russ could hoard the keys in his pproxy and release a few at a time.

 

[Bakwetu]

Why would that be cheating? If you assimilate yet another computer, albeit a little bigger and faster than usual, I can't see any wrong with that. Go for it!!!

 

[dennilfloss]

I'm going to sleep for a few hours. Then I'll check opinions on this matter. The person is in fact an AnandTech member but there's no guarantee that a linux client could work on such a beast.

 

[Mika]

Russ: Glad to hear you are OK!

I may have some insight into the corrupt pproxy log file. When I have had a machine with the pproxy crash, similar results were observed -- ppstats would not accept the log file. I opened the file in a text editer and scrolled down to the area of the file from the time of the crash. There was 1 entry which had about 80 spaces before the actual data entry. If the file is readable, there is a good chance it can be recovered.

Mika

 

[RaySun2Be]

I don't think it would be cheating if the person had permission, and complied with the usage policies. IBM has run RC5 on a huge Beowulf cluster as a team with huge outputs.

If it is a legitimate assimilation, then I say go for it, maybe using some team proxies so DNET isn't overloaded, especially considering the flushes because of the client downgrade.

Maybe Moose is lurking and can give a DNET perspective on this.

What O/S is that beast running? There are a number of Unix versions of the dnet client, not just Linux.

 

[dennilfloss]

All I know is that it's a Unix-based machine and a Fujitsu.

 

[RaySun2Be]

Dennilfloss,
I thought you were going to go get some sleep?

 

[dennilfloss]

Hard to get away from computer...

Trying...

Trying harder...

Argh!!! NOW!!!

 

[Bakwetu]

I can see your still logged on Dennis, there's no hiding here

 

[Russ]

Yeah, a UPS is definitely getting added. Anybody have any suggestions for one that I can plug 23 systems in to? HeHe. Actually, I just need one that I can hook the server and the backup server up. Something that's Linux compatible and not too expensive.

Hardware-wise, surprisingly little damage around here. I lost the keyboard on the test bench, but it was an eight year old Dell that I picked up cheap back when keyboards actually cost money (hey, what can I say? I never throw anything away as long as I can squeeze a little more work out of it).

One CPU on crack rack #1 got pushed over the edge. It's an Athlon 500 re-wired to 750 and became a no-boot. Popped a different CPU on the board, dropped the FSB to 90Mhz, put the weakling back on, and presto, a 675Mhz CPU.

Mika,

I went through the affected logfile, and there are no blank spaces anywhere. In fact, no visible signs of corruption. But, if I put it back in the directory, stats update won't run and just comes back with the corrupted file message.

Last entry is 10:54:37, a time that will live in infamy.

Russ, NCNE

 

[Poof]

Russ - what text editor did you use? Some don't always catch carriage returns/endlines, etc.