Team drive #1: k=400-1001 n=333.2K-600K

[mdettweiler]

Originally Posted by em99010pepe

Max, please add the rest of the ranges to IB400.

Carlos

Well, we'll probably want to put at least some of them on G4000 when it comes back up--but, yes, you're right, most of these will be eventually loaded into IB400. By my estimate IB400 will dry its range at least by Saturday. When G4000 come back online I'll add 3K to it and 7K to IB400; that will clean up the rest of the ranges and should balance the two pretty well considering that Gary said he's going to switch his machines to G4000 when it comes back online.

[MyDogBuster]

I'll add 3K to it and 7K to IB400; that will clean up the rest of the ranges and should balance the two pretty well considering that Gary said he's going to switch his machines to G4000 when it comes back online.

Max/Gary I have no problerm switching back to G4000 if Gary wants to keep his "farm" on IB400 where most of the remaining work will be. At 4K / day + what Max was doing, it should be pretty easy to figure out what the spread between the machines should be. Let me know where you want me.

[mdettweiler]

Originally Posted by MyDogBuster

Max/Gary I have no problerm switching back to G4000 if Gary wants to keep his "farm" on IB400 where most of the remaining work will be. At 4K / day + what Max was doing, it should be pretty easy to figure out what the spread between the machines should be. Let me know where you want me.

First of all, I presume you meant .4K/day, instead of 4K? Even Lennart can't crank out 4K/day.

As for which server your machines would be most useful on: Gary, how do you think we should do this? I'm fine with whatever.

[henryzz]

why not just keep everyone on IB400 it can cope with the load and it makes it easier to administrate

[gd_barnes]

Oh, who knows. We'll figure it out a little later.

My plan was to switch to port 4000 but if Ian wants to, then perhaps I'll stay on port 400. That probably makes more sense. That way, I won't inadvertantly have pairs sent to me that Ian is ready to send results back on.

Henry, we can't all be on port 400 the rest of the way. Port 4000 still has quite a few pairs left to be processed.

OK, let's do this: Ian, switch all of your machines back to port 4000 after you get the word that we have everything working. (I'm hoping late Tues. night or early Weds. morning.) Stay on it until you dry it out. If we add all the remaining ranges to ports 400 and 443 and dry them before you dry port 4000, then we'll switch over to port 4000. If port 4000 dries first (more likely I think with the large amount of range left still not loaded in port 400/443), then you can switch back over to port 400/443.

To clarify, we won't add any more work to port 4000 for the 1st drive. It will all go into ports 400 and 443.

One further thing that Max, David, and I are in discussion about in PM's. I'm recommending that we add k=400-600 for n>600K to port 400 even before we have finished this drive. That way, we can just keep right on processing with no interruption and people can leave their cores there.

The good thing about k=400-600 for n>600K is that it is sieved far deeper and it's a much smaller k-range, so we'll process the n-ranges much more quickly for quite a while until we have another fftlen change.


Gary

[MyDogBuster]

Ian, switch all of your machines back to port 4000 after you get the word that we have everything working. (I'm hoping late Tues. night or early Weds. morning.) Stay on it until you dry it out.

Sounds like a plan.

[mdettweiler]

Originally Posted by gd_barnes

Oh, who knows. We'll figure it out a little later.

My plan was to switch to port 4000 but if Ian wants to, then perhaps I'll stay on port 400. That probably makes more sense. That way, I won't inadvertantly have pairs sent to me that Ian is ready to send results back on.

Henry, we can't all be on port 400 the rest of the way. Port 4000 still has quite a few pairs left to be processed.

OK, let's do this: Ian, switch all of your machines back to port 4000 after you get the word that we have everything working. (I'm hoping late Tues. night or early Weds. morning.) Stay on it until you dry it out. If we add all the remaining ranges to ports 400 and 443 and dry them before you dry port 4000, then we'll switch over to port 4000. If port 4000 dries first (more likely I think with the large amount of range left still not loaded in port 400/443), then you can switch back over to port 400/443.

To clarify, we won't add any more work to port 4000 for the 1st drive. It will all go into ports 400 and 443.

One further thing that Max, David, and I are in discussion about in PM's. I'm recommending that we add k=400-600 for n>600K to port 400 even before we have finished this drive. That way, we can just keep right on processing with no interruption and people can leave their cores there.

The good thing about k=400-600 for n>600K is that it is sieved far deeper and it's a much smaller k-range, so we'll process the n-ranges much more quickly for quite a while until we have another fftlen change.


Gary

Okay, that sounds good. In that light, I'll send an 8K range to IB400 right now; that will then leave 2K remaining in the 1st Drive in case C443 needs a bit more or possibly if Lennart wants to grab another range. Once we get closer to drying out IB400 after that, I'll add the last 2K and some k=400-600 n>600K work right after that.

BTW, IB400 finished the last of 560K-567K last night; I'll get that processed later today.

Edit: Oh, I almost forgot: Reserving 588K-596K for LLRnet IB400. That may have been somewhat obvious from my above message, but I figured I'd better make it official.

[henryzz]

sorry i forgot about drying the other servers
in future i think 3 servers for one drive is a bit unnecessary though unless we have some proxies

[nuggetprime]

A little boost:
reserving 596.0-596.2!

takin' one of the last available chunks!

[mdettweiler]

Originally Posted by nuggetprime

A little boost:
reserving 596.0-596.2!

takin' one of the last available chunks!

File sent via PM.

[mdettweiler]

LLRnet IB400 has completed 555.6K-567.0K, lresults emailed to Gary.