VMware Cloud Community
thewaablah
Contributor
Contributor

Deploying Template taking 2+ hours

ESX 3.0.2

VC 2.0.2

Lefthand iSCSI SANS

Templates are on LUN0 (Templates)

VMs are on LUN1 (Infrastructure)

eth0 (Service Console & iSCSI connection. Autonegotiate) on Netgear Gigabit Switch

iSCSI modules are connected to seperate Netgear Gigabit Switch

The Gigabit switches are connected together no more than 5 feet away

Template is 15GB

I have another template that takes maybe 30 minutes that is also 15GB.

At one point templates took 10 minutes to deploy, now taking over 2 hours. I'm only running one deployment at a time.

Reply
0 Kudos
9 Replies
waynegrow
Expert
Expert

Do you see any errors/collisions on the physical switch?

Reply
0 Kudos
thewaablah
Contributor
Contributor

No unfortunately not. I had hoped that was the issue earlier. Not too sure what else it could be.

Don't know if VLANs could be the issue.

VLAN isn't setup with the Netgear A) because it doesn't work with our Cisco switches no matter what we try and B)Trunking really isn't neccessary on eth0

Netgear switching (not trunked)

eth0 is Service Console & iSCSI

eth1 is VMotion

Cisco switching (trunked)

eth2 is PortGroups

eth3 is teamed with eth2

Reply
0 Kudos
thewaablah
Contributor
Contributor

Well I ran some fiber between the switches, that didn't speed things up. Should I separate the iSCSI and Service Console?

I deployed 3 machines at the same time last night and it took exactly 5.5 hours. Smiley Sad

null

Reply
0 Kudos
w00ten
Contributor
Contributor

Sounds like you have an almost identical problem to the one I have. If we manually access shares and do a simple copy, a 2 gig file will transfer over in minutes. But anything done between ESX hosts, within a host, or between an ESX host and the NFS storage(yeah, I know, it sucks, but it's all we got at the moment) takes absolute hours, if not days to accomplish. I THINK this is a problem with the ESX hosts themselves and not the switch everything is on. I'm yet to find a solution. Also, VMWare converter is next to useless as it consistently fails when importing to VCenter/ESX.

Reply
0 Kudos
TravisT
Enthusiast
Enthusiast

Netgear switching (not trunked)

eth0 is Service Console & iSCSI

eth1 is VMotion

Cisco switching (trunked)

eth2 is PortGroups

eth3 is teamed with eth2

VMotion and iSCSI both go over the VMKernel network

I would have it setup so that 1 vswitch is setup for service console and another vswitch is for vmkernel.

Reply
0 Kudos
dkfbp
Expert
Expert

I think I saw another post earlier with a similar problem. He got it solved by updating the lefthand to version 6.6sp1. What version are you running?

Best regards Frank Brix Pedersen blog: http://www.vfrank.org
Reply
0 Kudos
thewaablah
Contributor
Contributor

Management Console is 6.6.01.4006

Firmware is 6.6.01.6006.0

Reply
0 Kudos
shawniverson
Contributor
Contributor

Are you seeing tons of iSCSI errors in /var/log/vmkernel on ESX?

First of all, go to a single switch for now. Eliminate as many variables as possible. Make sure all interfaces are 1000/Full and that flow control is set properly for all interfaces. Turn off jumbo frames. ESX does not support jumbo frames without a hardware HBA. Attach just single interfaces from all devices for now to the SAN switch, including the Lefthand and VMware. This will help to rule out teaming issues.

Make sure that the SAN is literally screaming when a Windows machine is attached to it with the Microsoft iSCSI initiator via the same switch and heavy IO is placed on it.

Does the problem go away temporarily when you reboot the ESX? If so, is your addressing scheme anything but 255.255.255.0 or /24?

How many VMs are you running?

Check out this post:

http://www.vmware.com/community/thread.jspa?messageID=730523&#730523

Reply
0 Kudos
shawniverson
Contributor
Contributor

That is SP1.

Reply
0 Kudos