ESXI getting away from RDM

[Hunter69]

I am looking for suggestions to create an easy way to recover should something happen to my Sagetv virtual machine. Here is my setup:
I have two hard drives, 1 for boot and 1 for recording. Both hardrives have a physical rdm. My sage server has 2 collossus tuners passed through to it it and utilizes a hauppage hdhomerun.

So now I am looking to setup backup and recovery. Currently every night I stop the sagetv service, backup the sagetv folder to my unraid server, then restart the sagetv service. This works but can be a pain if something happens that requires a full recovery of my sagetv server. I began researching possible bakup and recovery but it seems as though the physical rdm poses a problem. I cannot perform snapshots. SO I guess I am looking for suggestions. here are my questions:
1. Can I get away from the physical RDM. I record both hd and sd but mostly HD.
2. Cost is a consideration so I am looking for a cost effective solution.


Ways I know to backup this type of setup:
1. Shutdown machine and backup the esxi folder. With RDM this would backup the esxi setting but nothing on the RDM drives (correct). SO if I created a virtual HD for the system drive, then this approach would backup the OS. I would be able to restore everything except recordings (if the recording drive goes bad), correct.
2. A different approach it to install say acronis or ghost or something similar to backup the entire machine. My thought of recovery would be: i would have to recreate the vm in esxi (including RDM's) then boot from the appropriate recovery disk and all should be well. Am I missing something?

Is there a better way? Am I correct in understanding the issues an RDM creates?

Thanks for any help and suggestions

[Skirge01]

I think passthrough also limits your ability to take snapshots, so RDM may not be your only barrier there. I looked at the RDM route because of the 2TB limit in ESXI and gave up on it when I saw all the caveats; it just wasn't worth the risk. Instead, I went with FlexRAID for my storage pool needs. My backup plan is to use EaseUS Todo Backup software and create nightly, weekly, and monthly backups of all my VM's. Restoring should be as simple as booting the VM via a recovery disc and telling it which backup to restore.

Lastly, I don't believe you need to stop the SageTV service before you do your backup. If someone knows otherwise, I'd appreciate a decent explanation as to why.

[Hunter69]

Well part of the stopping and restarting sagetv is also collossus type issue. The collossus has some weird quirks if you don't stop and restart. It is not needed daily (maybe once a week) but doesn't hurt.

So do you rdm? Do you backup the esxi folders? I can certainly run some sort of imaging software. I am just trying to come up with the easiest way to recover should the entire server crash.

Thanks for taking the time to share your knowledge

[BobPhoenix]

I use WHSv1 and True Image for my backups and never shut off service mode. I have never had a problem with my Wiz.Bin the only file I really care about backing up. Everything else I can resetup from scratch if I have to. I have - however - restored the WHOLE SageTV folder from WHSv1 backup before without any problems. I also use to just backup the wiz.bin and properties to a zip file in the past and have restored from that. That is riskier than stopping the service because it doesn't use the Shadow Volume Copy that WHSv1, True Image, and many other backup programs use. But if you do it when no one is watching anything and SageTV is not being backed up and no video conversions are being done it should be safer.

[Skirge01]

As I said, I skipped RDM completely because of everything I read about it. It's a hack to begin with and I wasn't looking forward to risking future roadblocks resulting from going down that road. I wanted stability, without workarounds, which is what I had before when I was running WHS v1 as my server for everything.

I am still new to ESXI, so I certainly won't pretend to know all there is and my understanding could be off. I currently do not back up any ESXI folders (in fact, I'm not sure exactly what you're referring to there). If you mean the vmdk's and associated files, I definitely do not. My understanding (which, again, could be wrong) is that if a VM goes south and won't boot, I can tell the VM to boot from a restore CD to restore the image of the OS, as opposed to the vmdk. If the entire vmdk dies and I can't even get it to boot a restore CD (is that even a possible scenario?), I believe I can create a brand new VM and boot from a restore CD and once again restore the OS image backup. If the server (hypervisor itself) dies, I believe I can reinstall ESXI to a new boot drive (currently a flash drive), recreate the VM's and, yet again, boot from that restore CD and restore the OS image backups.

Once more, this is all theory, based on my current level of understanding and I have not attempted most of this. If you or someone else tells me my theories are incorrect, I would certainly be an attentive listener! However, I have used WHS to restore a VM already, just as a test scenario when I first started down this path. I don't actually use WHS for backups, but I had it on hand and decided to test my theory using that; it worked. I am planning to implement my full backup plan (as I just laid it out) in the next couple of months and will fully test my theories out.

[stanger89]

It seems to me (though I haven't really looked into it seriously) that you'd want the "OS drive" in the VM on a virtual disc (I don't know why you'd want that RDM or passthrough), and then just make periodic snapshots of the VM. Then if your VM goes south you can just revert to a prior snapshot:
http://kb.vmware.com/selfservice/mic...rnalId=1015180

Now I'm not sure how much space that takes, if it's a lot, you might just want to backup your SageTV directory routinely and save the VM snapshots for major changes.

If you've got unraid, a good option for backing up SageTV is Crashplan, you can install it on your VM and unRAID and have it backup SageTV (and whatever else you want) to your unRAID. I've actually got mine backing up my SageTV install to my unRAID and Crashplan.

[BobPhoenix]

Originally Posted by stanger89

It seems to me (though I haven't really looked into it seriously) that you'd want the "OS drive" in the VM on a virtual disc (I don't know why you'd want that RDM or passthrough), and then just make periodic snapshots of the VM. Then if your VM goes south you can just revert to a prior snapshot:
http://kb.vmware.com/selfservice/mic...rnalId=1015180

Now I'm not sure how much space that takes, if it's a lot, you might just want to backup your SageTV directory routinely and save the VM snapshots for major changes.

If you've got unraid, a good option for backing up SageTV is Crashplan, you can install it on your VM and unRAID and have it backup SageTV (and whatever else you want) to your unRAID. I've actually got mine backing up my SageTV install to my unRAID and Crashplan.

Snapshots only work when you do NOT passthrough PCI devices. Don't know about RDM as my WHSv1 is using pass through for the AOC-SASLP-MV8 to control the pool drives. I use RDM for the boot WHSv1 drive. So snapshots do not work in my case at all and may not if using RDM only. That is why I do image backups of my SageTV VMs since I am passing through the tuner cards. Even if I use an ESXi backup solution and directly backup the VMDKs it turns my "thin provisioned" virtual boot drives to "thick provisioned" drives. So with my True Image I can just boot up the VM with a True Image CD and restore to a newly allocated thin drive and be backup up in at most a couple of hours.

[Hunter69]

Ok so it sounds like I am correct in my understanding and my plan. I will just image the machine to my unraid server. Then if something goes bad I will recreate the vm (if necessary) and boot from the rescue disk.

My understanding (which, again, could be wrong) is that if a VM goes south and won't boot, I can tell the VM to boot from a restore CD to restore the image of the OS, as opposed to the vmdk

Well I don't think this is true, I guess depending on the os you are using. This process is not much different than a physical machine that won't boot. Now if you don't rdm or pass through anything to your vm, you can use snapshots. Snapshots are great in that it will roll your machine back to the moment in time that the snap shot happened. I only have one vm that dies not have anything passed through or a drive rdm'd. I am planning on doing a snapshot say once a week.

[Hunter69]

Originally Posted by BobPhoenix

I use WHSv1 and True Image for my backups and never shut off service mode. I have never had a problem with my Wiz.Bin the only file I really care about backing up. Everything else I can resetup from scratch if I have to. I have - however - restored the WHOLE SageTV folder from WHSv1 backup before without any problems. I also use to just backup the wiz.bin and properties to a zip file in the past and have restored from that. That is riskier than stopping the service because it doesn't use the Shadow Volume Copy that WHSv1, True Image, and many other backup programs use. But if you do it when no one is watching anything and SageTV is not being backed up and no video conversions are being done it should be safer.

What tuners do you use? The stopping the service started when I got the 2 colossus tuners. The nice thing about this setup is I know when the backup fails because I use win 7 scheduled task based on a successful backup logged in the event viewer. So if the backup is not successful the sage service is not started. This happens in the middle of the night while everyone is asleep. If I don't stop the service once every one to two weeks I start getting a bad recording. The issue has improved with driver updates but never fixed altogether.

[Skirge01]

Originally Posted by Hunter69

Well I don't think this is true, I guess depending on the os you are using. This process is not much different than a physical machine that won't boot. Now if you don't rdm or pass through anything to your vm, you can use snapshots. Snapshots are great in that it will roll your machine back to the moment in time that the snap shot happened. I only have one vm that dies not have anything passed through or a drive rdm'd. I am planning on doing a snapshot say once a week.

Maybe I didn't explain myself properly. When I say "it won't boot", I'm referring to a VM where the OS installed in that VM is the problem and won't start up; not a corrupted VM that throws an error when you try to start it, before it even tries to boot the OS in the VM. I have mounted CD's or ISO's in a VM so that it would boot directly off of those, so I'm pretty sure my method will work. Does that make more sense?

[BobPhoenix]

Originally Posted by Hunter69

What tuners do you use? The stopping the service started when I got the 2 colossus tuners. The nice thing about this setup is I know when the backup fails because I use win 7 scheduled task based on a successful backup logged in the event viewer. So if the backup is not successful the sage service is not started. This happens in the middle of the night while everyone is asleep. If I don't stop the service once every one to two weeks I start getting a bad recording. The issue has improved with driver updates but never fixed altogether.

I have 2 virtualized SageTV boxes. One has USB2.0 card passed through for my 2x HD-PVRs and a AVer Media Duet tuner and a Hauppauge HVR-2250 and HVR-950Q USB tuner. The other virtualized SageTV box has 2 Duets, 1 HVR-2250 and a HDHR version 1 tuners. My colossus tuner is on a third non-virtualized SageTV server. WHSv1 lets me know when a backup fails because I backup all of my PC including my laptop to it so when a backup fails I see it right away. I agree about the Colossus driver and restarts I restart SageTV service on that PC every night but as I said it isn't virtualized.

[Fuzzy]

I'm not currently using ESXi, but I was under the impression that RDM and passthrough only prevent LIVE snapshots. If you are scheduling downtime daily, you should be able to shutdown the sage service, and perform a configuration backup, then cleanly shutdown the VM and perform a VM snapshot. Then restart the VM.

[BobPhoenix]

Originally Posted by Skirge01

Maybe I didn't explain myself properly. When I say "it won't boot", I'm referring to a VM where the OS installed in that VM is the problem and won't start up; not a corrupted VM that throws an error when you try to start it, before it even tries to boot the OS in the VM. I have mounted CD's or ISO's in a VM so that it would boot directly off of those, so I'm pretty sure my method will work. Does that make more sense?

It does work because that is how I've transferred my OS's from one ESXi server to another. I just make a backup with True Image to a common network location - my N40L unRAID standalone box where all True Image backups go. Then tell new VM on same ESXi server or another ESXi server to boot from the True Image boot CD Image I created and restore the VM from my N40L backup across the network. Usually takes less then an hour depending on how much data was backed up. Since I have True Image do daily backups I can do the same thing if a VM fails I just keep restoring back until I get one that works. And if a True Image backup doesn't work I can try with my WHS v1 backups either have worked for me.

[BobPhoenix]

Originally Posted by Fuzzy

I'm not currently using ESXi, but I was under the impression that RDM and passthrough only prevent LIVE snapshots. If you are scheduling downtime daily, you should be able to shutdown the sage service, and perform a configuration backup, then cleanly shutdown the VM and perform a VM snapshot. Then restart the VM.

That appears to be true but for me if I was doing it I would want to just backup the VMDK to a network location separate from the ESXi server anyway so I never take down the VMs except for hardware upgrades.