Server crashed -Hardware Failure.... need advice

[vshaulsk]

Half_life ....  I have been reading more about DRBD and it sounds very interesting.  I do have a question do the two servers have to be completely identical when it comes to hardware

In my case the systems have different amount of ram and completely different cpu (not only one being amd and one being intel.... the new system has dual processors vs the old one being a single)

Finally if the CPU and ram don't play a role and I can still create this cluster.... would I even need hardware raid (not talking about my raid6 this is completely separate).  Would I be able to use one SSD, one 500 gig drive and one 750 gig drive in each server..... using DRBD I would be able to create a raid1 over the network the 500 gig and 750 gig drives.  I will keep the SSD on each system for the base Operating system.

From what I understand I can combine the 500 gig and 750 gig drive under LVM (total = 1.25 TB) and with DRBD essentially have raid 1 array.   I could then create my 4 VM servers.... I would like them to always be on the new server and only switch to the old if something fails.

Hopefully I understand this correctly.. after doing some reading during lunch today

 

[vshaulsk]

So I got my new hardware (new processors and motherboard.... different brand and type then before)..... installed it and plugged my OS drives back in.

The system started right back up.  The only immediate issue has been with the network interfaces.  I did not know how to reset my original network interfaces and because they are tied to mac addresses they do not automatically work for the new network cards.

I am not sure if I did this correctly, but I set all the interfaces to not assigned (saved zentyal) and rebooted.  I then went into /etc/udev/70-persistent-net.rules and changed the eth(XX) of the new network interfaces (based on mac address) to match the number of the old interfaces.  Following this I deleted the old interfaces from the file and restarted the system. 

Now zentyal sees my new network interfaces and everything came back online.  I am not sure if this is the correct way and perhaps someone has a better explanation for me....   Also perhaps I should have made other changes when installing a new motherboard and cpu (not just simply plug in old drives and boot up)

On my second server I will try proxmox and if I am successful I will rebuild my production server.

[ichat]

hmm. im not sure if you managed your NIC's the 'default' way,  but i would have done the same thing.. this way i can just manually leave allone all existing rules and settings and only   add the new mac + driver   to that eth interface... 

as long as you dont apply vlan tagging  while the new nic doesn't support that, you should probaly be fine...   

at least i would have done the same thing - i guess.

[vshaulsk]

Ichat actually I do have Vlan taggin enabled..... I think the trick was to make sure that the original interfaces were set to nothing (basically change them to not assigned) and restart the system.  After that once I erased the old lines for the old interfaces in /etc/udev/70-persistent-net.rules (I guess I could have just commented them out).... simply changing the eth(XX) number on the new interfaces to match the originals did the trick.  It even brought back my VLans as soon as I assigned the same Vlan #. 

I guess what I have learned most from this is that unless both of my raid disks get corrupt when a motherboard, cpu or PSU go out.... I can recover my system within 24 hours by overnight parts (which is an acceptable time for me).  Also I could just keep a spare server (not as powerful) and just insert the main drive disks + /home disks which will allow me to get most my system back up within minutes.  The only thing I would not have is my raid6 array and probably the ability to turn on all the virtual machines(due to hardware limitation).


I still need to learn how to bring up my entire system from my backups which are on an FTP location..... I will probably learn that next.