Dell VRTX and UPS setup

[clickrgs]

Hi all

 

I'm after a bit of advice from anyone who currently has a VRTX.

I'm looking at getting one which will be 2 blades running Windows Hyper-V hosts and several virtual servers, basically the entire school network. My plan is to put 2 APC UPS's in the rack with it and have each connected by USB to one of the hosts and then they will run the shut down sequence in case of a power failure. All this is simple enough but there doesn't seem to be a way to shut down the VRTX chassis nicely from a sequence.

 

My question is, how does anyone who owns a VRTX do it?

 

Thanks in advance

Nicki

[mowgli82]

I have a network card in the UPS which is controlled by the virtual appliance that you run as "VM" on the host from Powerchute which shuts down all the VMs then the host. This is for ESXI I imagine there must be something for Hyper-V

[clickrgs]

I have a network card in the UPS which is controlled by the virtual appliance that you run as "VM" on the host from Powerchute which shuts down all the VMs then the host. This is for ESXI I imagine there must be something for Hyper-V

 

The Hyper-V virtual servers will happily shutdown (or at least pause) nicely as part of the host shutting down. My issue is the VRTX chassis itself which has all the shared storage in, that doesn't appear to have any way of telling it to turn off so it would just crash it when the UPS ran out of battery. I did find one person online who said that as there's nothing accessing the storage it shouldn't matter it is gets crashed like that.

[mowgli82]

Arh ok, sorry I can't be much help. (. I don't like stuff that doesnt shut down cleanly!

[clickrgs]

Arh ok, sorry I can't be much help. (. I don't like stuff that doesnt shut down cleanly!

 

Don't worry about it. I was hoping someone on here might have one and could at least tell me how theirs is set up.

[sllorep]

I'm sorry but I can't help either, however I am interested in this too as we were looking at the VRTX also. Out of interest, what is the size of your school (sorry to take it briefly off topic!!)

[clickrgs]

I'm sorry but I can't help either, however I am interested in this too as we were looking at the VRTX also. Out of interest, what is the size of your school (sorry to take it briefly off topic!!)

 

Well I've been here less that 2 months so far but I think we have approx 660 students, 80 staff and 450 pc's.

[sllorep]

Ok, thanks. The reason I ask is because everything I read about it says that it is mostly for SMB or remote office locations and was wondering if it was up to the job for us to replace all of our servers into one box for 1500 students, 150 staff, 600 client pc's (currently 7 physical and 25ish virtual servers).

[clickrgs]

Ok, thanks. The reason I ask is because everything I read about it says that it is mostly for SMB or remote office locations and was wondering if it was up to the job for us to replace all of our servers into one box for 1500 students, 150 staff, 600 client pc's (currently 7 physical and 25ish virtual servers).

 

I don't see why it couldn't replace them really as it can have 4 blade servers, each of which can be very powerful. We'll only have 2 in a cluster with 13/14 virtual servers stored on a LUN on the shared storage so there's no need to replicate the virtual servers. Currently we have 9 physical servers that are approx 8 years old, 2 of which have virtual servers on (3 in total).

[Linfit]

We have VRTX with 4 blades and run 26 virtual servers on it - all our server infrastructure pretty much, running on Hyper-V. We specced some additional 10GB network connections on the and filled the thing with storage, so we have no issues at all with performance, and it replaced an EMC SAN and a set of 6 PowerEdge 710's running XenServer, so we have a saved a load of rackspace, reduced our power bill and significantly reduced the load on our air-con in the server room, which was a real issue for us. I would highly recommend them. We have approx. 600 devices, 1100 users.

[clickrgs]

We have VRTX with 4 blades and run 26 virtual servers on it - all our server infrastructure pretty much, running on Hyper-V. We specced some additional 10GB network connections on the and filled the thing with storage, so we have no issues at all with performance, and it replaced an EMC SAN and a set of 6 PowerEdge 710's running XenServer, so we have a saved a load of rackspace, reduced our power bill and significantly reduced the load on our air-con in the server room, which was a real issue for us. I would highly recommend them. We have approx. 600 devices, 1100 users.

 

Thanks for this. Our air-con can barely cope at the minute too. How do you have UPS's connected up to the VRTX or do you not bother?

[Linfit]

The two UPS's which power the VRTX have network cards - they are monitored by a separate physical server (which we use for general IT support stuff, run various consoles and admin stuff on etc) and this initiates the shutdown as required - it uses the APC software to run a script when there is 20 mins battery life left.

[clickrgs]

The two UPS's which power the VRTX have network cards - they are monitored by a separate physical server (which we use for general IT support stuff, run various consoles and admin stuff on etc) and this initiates the shutdown as required - it uses the APC software to run a script when there is 20 mins battery life left.

 

So do you shut the chassis down as well or just the blades? That's the only thing I haven't been able to work out yet, there doesn't seem to be a way to nicely shut the chassis down and then get it to come back on again when the power is restored.

[Linfit]

Just the blades. I don't think the chassis failing when the power goes off would be a big issue. If the hosts are off there will be no disc activity at that point, so shouldn't cause any issues.

[howartp]

The SAN in our VRTX (backup/DR) is no different to the SAN in our server room - there's no means of shutting down the main SAN other than pulling the plug out, so we just shut down everything else and leave the SAN pulling relatively little power.

[clickrgs]

Ok so I won't worry about the shared storage going off. The only other thing I keep being told is that I should a physical dc in case anything screws up with shared storage and I need a dc to get things back online. Is this really necessary, do you guys have one?

[Linfit]

I do have a physical DC as well as two virtual ones - I am old-fashioned that way! Its in a different comms room at the other end of the site, so its not that I don't trust the virtualisation per se, its mainly that we wanted a physically separate DC as there are some scenarios where you might lose just part of the building.

[clickrgs]

I do have a physical DC as well as two virtual ones - I am old-fashioned that way! Its in a different comms room at the other end of the site, so its not that I don't trust the virtualisation per se, its mainly that we wanted a physically separate DC as there are some scenarios where you might lose just part of the building.

 

If I did have one I would just reuse one of our current servers as the physical dc and keep some spares for it but it would have to be in the same room as the VRTX. It wouldn't be much use if the VRTX went down anyway as noone can login without the file server being accessible. I'm struggling to find a reason to need it, short of being told I have to have one by the company who may need to support it at some point.

[howartp]

I've heard the same yarn about needing physical, and when we went VMware about 8 years ago, we kept one.

 

When we upgraded VMware and hardware 3 years ago, we got rid of physical DC.

[Linfit]

Its not that I am tied to having a physical DC, its just I want the redundancy afforded by having it on the other side of the campus.

If I had another physically separated Hyper-V host, I'd be happy for it to be a virtual DC, but all our hosts are in the same cabinet.

[clickrgs]

Its not that I am tied to having a physical DC, its just I want the redundancy afforded by having it on the other side of the campus.

If I had another physically separated Hyper-V host, I'd be happy for it to be a virtual DC, but all our hosts are in the same cabinet.

 

So your physical dc is there in case the all the hosts go down in the cabinet for whatever reason? If that is the case would all your users still be able to login when only that DC was there? Ours wouldn't as their profiles are stored on the file server so nothing would work even if they managed to login.

[DavR]

Another VTRX user here, with two blades running VMware EXSi, hosting 10 x Servers.

 

We have one LAN connected UPS, and a virtual appliance running on one of the blades. When the battery runs down on the UPS the virtual appliance shuts down the virtual servers gracefully, then we let the SAN cut out when the battery dies. I'd rather there was a graceful shutdown on that too, but it doesn't seem to be an option on a lot of SAN technologies, so I assume not a big issue.

 

One thing I did have to do when I took over this kit is to of course ensure that at least one switch was powered via the UPS, as otherwise the virtual appliance doesn't get the signal from the UPS that power has been lost. We had a few unplanned shutdowns before I uncovered that!

 

Edit - just seen the second page of this thread, oops! No, we don't keep a physical DC. We do have one DC on each blade, but that's the extent of our redundancy.

Edited December 16, 2016 by DavR

[Linfit]

So your physical dc is there in case the all the hosts go down in the cabinet for whatever reason? If that is the case would all your users still be able to login when only that DC was there? Ours wouldn't as their profiles are stored on the file server so nothing would work even if they managed to login.

 

They wouldn't be able to get to their home shares, or use e-mail (both virtual servers) but profiles are local, so if they were using their normal machine, they could log on, use SIMS (which is not yet virtualised here) get on the internet, use Office and save to local storage - it would provide limited but useful functionality. I am also thinking of more permanent problems (like a fire in the main building for example) where the server room might be lost, but with a surviving physical DC we could make a start on rebuilding infrastructure more quickly.

[mjs_mjs]

If you have an external pc doing the shutdown sequence you can fire a chassis shutdown command to the CMC which should power it all down nicely (after doing each blade). That's how i've heard of m1000e chassis' being done in the past.

[DavR]

If you have an external pc doing the shutdown sequence you can fire a chassis shutdown command to the CMC which should power it all down nicely (after doing each blade). That's how i've heard of m1000e chassis' being done in the past.

Interesting, I've not played with command line to the chassis on the VRTX much, but I wonder if I could script this from my VMware Virtual Appliance.... would have to be careful with timings and sequencing etc, but could be workable!