Planned downtime on GB servers

[henryzz]

Originally Posted by mdettweiler

Gary, I've now got dumpford set as the "gateway" machine. This will require a small change to your remote-desktop-launcher.pl file; see attachment for the updated version.

This now completes the changeover of everything "important" from crunchford to dumpford! Now crunchford is just an ordinary crunching quad again. Gary, when you get the chance, I'd highly recommend doing a clean install of Ubuntu 8.04 64-bit on crunchford like you did before on the rest of the AMDs. (Well, except hueford, which as I recall is still on 7.04. You'll want to upgrade that one too. ) That way you'll be able to do sieving at full 64-bit speed, and won't have the limitation of not being able to install anything on it due to the Ubuntu 7.04 package repository being long gone. Also, it's a humongous mess from the hackish tweaks I had to do to it to get around the 7.04 limitations, so it could reall yuse a clean sweep.

how come you are suggesting he uses 1 1/2 year out of date 8.04?
i wouldnt suggest he uses 9.10 yet as it seems to be full of bugs but 9.04 will survive for longer

[mdettweiler]

Originally Posted by henryzz

how come you are suggesting he uses 1 1/2 year out of date 8.04?
i wouldnt suggest he uses 9.10 yet as it seems to be full of bugs but 9.04 will survive for longer

8.04 is the LTS (Long Term Support) version. As such, it's going to be supported for a much longer span of time than a non-LTS version would. It's designed for cases like this where it's not crucial to have the absolute latest version, but rather what is most important is long-term stability. As you can see here, 8.04 will in fact be supported as long as 9.10 will (until April 2011).

When the next LTS version comes out, then we'll start upgrading them to that, though in the meantime, 8.04 is still fully supported and should suit our purposes quite nciely.

[AMDave]

I just finished migrating another stats site to a new machine and have the procedure worked out.

When you think you have stabilised your servers, decide which server has enough HDD space & backup media, etc, let me know by email and I'll set about migrating the stats site over.

After it is migrated, integration of the PRPnet ports and double check files can proceed.

It's a lot of HDD space so that is a major consideration. I'll try to calculate the total bandwidth for the database migration and spread it over time.

Anyway, drop me an email when you have things sorted out a bit more than they are now.

FTR CentOS is preferred for performance reasons, but Ubuntu is my home turf too, so all will be easily squared away.

[mdettweiler]

Originally Posted by AMDave

I just finished migrating another stats site to a new machine and have the procedure worked out.

When you think you have stabilised your servers, decide which server has enough HDD space & backup media, etc, let me know by email and I'll set about migrating the stats site over.

After it is migrated, integration of the PRPnet ports and double check files can proceed.

It's a lot of HDD space so that is a major consideration. I'll try to calculate the total bandwidth for the database migration and spread it over time.

Anyway, drop me an email when you have things sorted out a bit more than they are now.

FTR CentOS is preferred for performance reasons, but Ubuntu is my home turf too, so all will be easily squared away.

Okay, cool. The server should have a 320GB hard drive, which I'm guessing should be plenty, right?

[AMDave]

Confirmed, and room for the planned expansion. Great.
I'll be in touch after work.

[gd_barnes]

Max,

I'm getting "host key verification failed" when I try the new launcher so I can't access any of my machines remotely. What's up?

It seems like more is needed than just changing the launcher on my laptop. Aren't there some behind-the-scenes changes needed? I see there's a known_hosts file. Does that need to be changed? I guess the question is, what all changes did you make to your machine to be able to remotely access my machines?

I need to send a factors file to Tim at CRUS for Riesel base 24 that finished late Thursday.


Thanks,
Gary

[gd_barnes]

OK, I was able to get rid of that message in the above post by first attempting to accesss dumpford but now I'm getting "permission denied" in the Windows DOS box when I type what should be the correct password. In addition to that, I'm getting "Failed to get server address. Did you type the host name correctly?" in a separate Windows pop up.

It looks like some work is still needed to get the remote access working on my laptop.


Gary

[mdettweiler]

Delete the known_hosts file. That should fix the problem. Note that after doing that, you'll be asked "are you sure you want to connect" when connecting to one of your machines the first time, but after that it should remember your choice. (That's what the known_hosts file is for, but the problem is, it's got crunchford's "fingerprint" stored in there for nplb-gb1.no-ip.org. Now dumpford's there and it's expecting crunchford, so it think it's getting some kind of impostor.)

[gd_barnes]

Originally Posted by gd_barnes

OK, I was able to get rid of that message in the above post by first attempting to accesss dumpford but now I'm getting "permission denied" in the Windows DOS box when I type what should be the correct password. In addition to that, I'm getting "Failed to get server address. Did you type the host name correctly?" in a separate Windows pop up.

It looks like some work is still needed to get the remote access working on my laptop.


Gary

Originally Posted by mdettweiler

Delete the known_hosts file. That should fix the problem. Note that after doing that, you'll be asked "are you sure you want to connect" when connecting to one of your machines the first time, but after that it should remember your choice. (That's what the known_hosts file is for, but the problem is, it's got crunchford's "fingerprint" stored in there for nplb-gb1.no-ip.org. Now dumpford's there and it's expecting crunchford, so it think it's getting some kind of impostor.)

Perhaps you missed my 2nd problem posting. That doesn't fix the problem with the bad password and "failed to get server address" problems. I had already gotten around the 1st problem by 1st attempting to connect to dumpford.

Since this is not an easy fix, I need a favor from you if possible. Please connect to "buttford", go to the Desktop/Prime/prime2/sr2sieve and Desktop/Prime/prime3/sr2sieve directories and find the factors.txt files. Please copy them off and post them in the thread at CRUS where I'm coordinating with Tim (Mini-Geek) on sieving for Riesel base 24. It finished late Weds. and I told him I'd get the files to him by Thurs.

Note that it's possible that I could be slightly off on the above directories. If so, I think it should be self-explanatory on where to go. I have a prime, prime2, prime3, and prime4 subdirectories under Prime. You'll see a Riesel base 24 sieve file in the two directories that contain the needed factors.txt file. You'll just need to make sure they are Riesel base 24 factors.


Thanks,
Gary

[mdettweiler]

Originally Posted by gd_barnes

Perhaps you missed my 2nd problem posting. That doesn't fix the problem with the bad password and "failed to get server address" problems. I had already gotten around the 1st problem by 1st attempting to connect to dumpford.

Since this is not an easy fix, I need a favor from you if possible. Please connect to "buttford", go to the Desktop/Prime/prime2/sr2sieve and Desktop/Prime/prime3/sr2sieve directories and find the factors.txt files. Please copy them off and post them in the thread at CRUS where I'm coordinating with Tim (Mini-Geek) on sieving for Riesel base 24. It finished late Weds. and I told him I'd get the files to him by Thurs.

Note that it's possible that I could be slightly off on the above directories. If so, I think it should be self-explanatory on where to go. I have a prime, prime2, prime3, and prime4 subdirectories under Prime. You'll see a Riesel base 24 sieve file in the two directories that contain the needed factors.txt file. You'll just need to make sure they are Riesel base 24 factors.


Thanks,
Gary

Oh, poop! I realize what I did now. I accidentally gave you a modified version of *my* copy of the script (which I never use anyway, but I have one nonetheless)--i.e., it has my username in there instead of yours. Thus it was asking for my password. It should still work as long as you enter my password. However, I've attached the corrected one that works with your username, and should behave as it did before.

Right now I don't have time to grab the factors, though I could do it later today if need be. Since you should be able to get on one way or another now, if you can get it to work before I get home from various events that I'm doing today (should be late afternoon or so), post about it here. Otherwise, I'll do it when I get home.

[gd_barnes]

The bummer thing is that I caught that, corrected it, and I'm still having the same problem. I'll try your modified version now and see if that helps but I'm thinking it won't.