ESX boots faster than SAN

[Photubias]

Hello,

I'm sure that I'm not the only one with this problem. However I do not seem to find any posts about it.

The problem is that, from time to time, we suffer from power losses.

However, our infrastructure is as such that all our VMs reside on a iSCSI SAN. Which boots significantly slower than the ESX servers themselves.

As a result, when ESX boots, it does not see the SAN which means that our VMs cannot start up.

Is there a script that, on ESX(i) startup, scans the storage HBA's untill a SAN is found? And then re-executes the VM autostart scripts?

Thanks in advance,

Photubias

[idle-jam]

you could use a script to rescan http://thephuck.com/virtualization/using-powercli-to-rescan-hba-and-vmfs-in-a-cluster/ or alternative make your host boot slower by having more boot options on top and also boot from network and have it timed out from pxe.

[AndreTheGiant]

Another solution is add a boot delay in your BIOS on in your ESXi:

http://www.vm-help.com/esx/esx3i/boot_delay.php

[Kahonu84]

Sounds like you have more fundemental problems to concern yourself with.

[Photubias]

You are probably referring to the power outages.

Well, they are pretty comon where I live (Belgium), we have about 3 or 4 each year, they only last about 10 minutes.

But if it happens on saturday morning (like last weekend) we are in trouble because we cannot enter the buildings to manually interfere. Not much we can do about it.

UPS systems are also not an option because of their price. We are a university college (vIT Academy even).

I was hoping for a solution like a script that directly uses vmware-cmd or something, because if we run into an outage there usually is no (Windows) client PC to run the powershell scripts from.

Slowing down the ESX bootprocess could be an option, but only if our SAN does not need to do filesystems checks. Which it does after a crash, causing it to take significantly longer to boot. I assume even a slowed down ESX boot won't help that much.

Any other (helpfull) toughts?

Thanks

[eeg3]

If they only last about 10 minutes, can you afford a smaller consumer-level UPS(s)? You're going to run into bigger issues than them booting up in the wrong order eventually, I suspect.

[ronanian]

Bump.

I have the same problem except outages can last for hours or days. I want a UPS and generator so we can just stay online but I can't sell that idea to management. My ESXi 5.0 hosts seem to want to be rebooted after the HP P4300 SAN comes up. Rescanning doesn't do the job but rebooting does.

Photubias, did you manage to solve the problem? If so, how?

[RParker]

ronanian wrote:

Bump.

I have the same problem except outages can last for hours or days. I want a UPS and generator so we can just stay online but I can't sell that idea to management. My ESXi 5.0 hosts seem to want to be rebooted after the HP P4300 SAN comes up. Rescanning doesn't do the job but rebooting does.

Photubias, did you manage to solve the problem? If so, how?

I think the problem is with your SAN then not the hosts, if a simple reboot fixes the problem than it's probably because the SAN isn't ready by the time the hosts are.  there isn't much you can do if you aren't willing to spend the money to get a PROPER power outage contingency plan.  There is only one way to do things, and that's the RIGHT way.

Like other posts are saying.. there are other infrastructure problems that will come up WAY beyond not seeing your SAN, if your SAN loses power, or hosts.. that's REALLY bad, because that means corrupt data.  Delay the boot of the hosts for say 30 minutes by editing the grub configuration, but that's not a good solution, the idea is to KEEP the SAN from going down.... If management doesn't care about then maybe you shouldn't either.. there is only so much you can do.

rebooting fixes it.. so reboot AGAIN after the hosts come up.  There is only so much that can be accomplished if you aren't willing to actually FIX the problem.

[ronanian]

I agree in entirety...I'm just looking for a kludge to automatically take care of that reboot. I work with what I have, and I try not to stress too much over what I can't have.

[NinjaHideout]

Worst case, I'd create a small VM residing on the ESX local storage, set it to autostart, and use that for "emergency remote management".

[RParker]

NinjaHideout wrote:

Worst case, I'd create a small VM residing on the ESX local storage, set it to autostart, and use that for "emergency remote management".

Still won't help if the ESX hosts he is managing can't manage the VMs because the storage isn't present UNTIL another reboot... that doesn't help in this case.

[Photubias]

Hi,

I dediced to move my vCenter VM to an ESX host with local storage. After that VM (auto-)boots it checks (Windows Powershell scripting ) the datastores and rescans and/or reboots the ESX Hosts.