Upgrade to New Server Seeking Advice

Hey Tony,

Technically, if you're running CentOS 4, you should be able to just take the drive and plug it in to the new server and it will work. I mean, I had a power supply die on me the other day and had no spares, so I took the drive (which was on an older P4 1.8) and moved it into a Core 2, and it worked... it detected a bunch of different hardware upon boot but it worked.

This is what I did during an emergency.

I could have also transferred the backups from one server to the other, did a mass restore and swapped the IPs, which would also work. I think this is the best way of doing it. I mean, if you get a new OS going too, like CentOS 5.1 or something, then you'd be running latest everything!

Sometimes fresh installs + backup restores are the way to go as they give you fresh everything.

If anyone here has a quicker technique using cloning technology like G4L, do let us know.

My 0.02$ for you Tony,

 

Yes this would be possible. Mount the directories properly.

 

Hey Tony,

May I geek rant out to you here?

500 GB drives? If it was me, and it's not, I would consider staying away from them right now. I operate a DC with about 250 boxes, and 500 GB is what's failing lots as of late. I spoke to a local data recovery guru, here in Ottawa (http://www.capitaldatarecovery.com/) and he told me 500's are what he's getting the most of. He explained that when Seagate commissioned these new drives, they added an extra platter inside, but didn't upgrade the spindle they sit on and so, they're more susceptible to damage as the spindle has trouble handling the extra weight. I have had 3 drive failures in the last 2 months, all 500 GB Seagate SATA's. My 120's and 200's are all AOK though. Obviously just giving you a heads up. Maybe get 3 x a smaller caliber drive instead... just thoughts, don't have to listen to me though. Pro-action vs. reaction here.

Go to CentOS 5, 100%. It's supported for many many many years still. It's worth going to, I love it.

Software RAID1 is good... I mean, it's mirroring right, so it's not really bad. If failure happens on the array and it doesn't rebuild properly... well, you're in a pickle. The other way though, if the main drive fails, so what... you have the backups on the other drive, and the chances that they will both fail is really low... probably lower than the RAID1 setup, and also, if you set your backups to Incremental Backups, you'll have less I/O on the drive as only files that need backing up will be backed up, and not ALL files + a tar... less reading and writing is good for the servers soul.

If you think YOU were rambling, man...

Let me know if you have any other questions Tony,

 

I have to agree with this advise as well. I have a desktop box that's 3 months old and 10% of the sectors are marked bad ..i replaced it. I use tools like SpinRite and it finds and recovers new sectors all the time and marks more bad. It's a hunk of junk and it's a Segate. I also helped a guy rebuild from a cPanel restore in Jan and both his main drive and his backup drive was failing and both were 2-3 month of Segate 500's. So stay away from those. The 250GB WD drives seem to be ok ..not sure about their 500's.

As for the OS, ..RHEL5 or CentOS5 is a great way to go also I would presonally get the new OS and box setup and just pack and rsync ..and YES the incremental backups on a couple hundred gigs of userdata will typically take 10-20 minutes after the first run happens.

 

at the DC where I have my boxes I have exactally what you just mentioned. I have dual 250GB drives and the second holds the nightly, weekly and monthly backups. I then have separate boxes that I use for remote backups. This particular DC I use will slave both drives and install a new drive in 10 minutes for me and in their control center I can select any of about 9 different OS's and watch from a KVM on the IPMI cards. You can even boot from remote DVD drive across the VPN. What I normally do is restore from the still working main drive except when it crashes so bad it won't read. Then I use WHM backup restore feature to rebuild from the backup drive. You are still looking at a couple hours including the OS install and all of that. But having a backup drive I feel is a better solution than raid. The reason why I say that is because it holds a snap shot that may protect you from security problems like possible roots and compromised OS. A raid system can give you speed and hardware redundancy but I can't see where it would help in a hack restore. That why I like backups on separate drives. my 2cents

 

That's some good 0.02$ big g-eye!

Oh I'm so punny.

We do incremental local and incremental across our LAN to archives. Basically REALLY fast and not time consuming or bandwidth intensive, at all. If only cPanel would add the --delete (hint hint) option, it'd be flawless. Right now, we can't. So basically, if you have a index.html, and it's backed up, then you delete it tomorrow and add index.php, it will also be backed up so you'll have both index.html and index.php present. So if you restore, guess what shows up? Yeppo, index.html. Bad behavior. Should have a config, "Delete files that no longer exist from destination folder? (incremental only)" or something.

I fully agree about RAID1 btw; a rooted box running RAID1 will become a mirrored... rooted... box... which is bad.

Hehehehe, I'm dealing with a box that was rooted with the "aliens" rootkit now. The asses.

Regards,

 

i think i will mention that rsync --delete option to Nick see what he thinks. I didn't even notice it worked that way.

 

To Nick? No way. I guess I'm speaking to the right guy.

As per the man pages:

Code:

     --delete                delete extraneous files from dest dirs
     --delete-before         receiver deletes before transfer (default)
     --delete-during         receiver deletes during xfer, not before
     --delete-delay          find deletions during, delete after
     --delete-after          receiver deletes after transfer, not before
     --delete-excluded       also delete excluded files from dest dirs

The solution is more than ideal, and if it could --delete too... oh wow.

If you spoke of this to Nick, DannyBlaze and I would be more than happy!

Yeah, "more than happy". Sounds like a mental condition huh?

Take care,

 

i sent him an email earlier. actually next time I talk to him I will ask him about this. Until you mentioned this I had no realized the issue you are describing may be exactally what caused a restored site I did to behave the way you mentioned about the index and index. I can see where emails also could be an issue also. hummm

 

Just my bit...

I use rsnapshot for my backups. Basically it uses rsync and hard links to accomplish the task of keeping 7 dailies, 4 weeklies, and 3 monthly backups. Because of the hard links, the backups take little more space than the data drive I'm backup up. Essentially I have 4 months of backups...

Now if I could just figure out what I need to backup cpanel wise so that I can restore settings as well....