Solved

Ghost 3750x

Posted on 2010-08-20
15
1,052 Views
Last Modified: 2013-12-02
We are trying to ghost approximately 20 PCs using ghost across the same VLAN on a set of stacked 3750x switches.  The problem is, during the ghost, several of the machines just drop out of the ghost session.  The switches are 3750x-48t-l.  Two switches are stacked together.  All PCs and the ghost server are all on the same vlan.  What could be causing the machines to dropout of the ghost session?  Is there anything that could be run on the switch to make the ghost more efficient?
0
Comment
Question by:hfccadmin
15 Comments
 
LVL 8

Expert Comment

by:jimmyray7
Comment Utility
Are the ones that drop on the same or different switch as the ghost server (or both)?
0
 
LVL 24

Expert Comment

by:rfc1180
Comment Utility
>We are trying to ghost approximately 20 PCs using ghost across the same VLAN on a set of stacked 3750x switches.

How many Ghost servers are you using?
What is the port speed of the PC switchports (100Mbps, 1000Mbps, etc)?
What about the ghost server(s), what is/are the port speed of the switchports?
Do you have access to the switches, are there any errors on the ports that the server and PCs connected to?

Billy
0
 

Author Comment

by:hfccadmin
Comment Utility
1 ghost server 20 machines.  PCs are all on the same stack, but not the same switch.  All switches are same version of IOS. Ports are 10/100/1000.  I have access to the switches, and there are no port errors on server or node ports.
0
 
LVL 24

Expert Comment

by:rfc1180
Comment Utility
How many of the nodes are using the ghost session at a give time?

>Ports are 10/100/1000
Understood, but what is that actual negotiated speed of the ports

Negotiated port speed of the ghost server:
Negotiated port speed of the ghost nodes:

It is very possible that you are saturating the network that is causing the issues that you are seeing.

Billy
0
 

Author Comment

by:hfccadmin
Comment Utility
We ghost this many machines, and more, all the time with no issue.  My thought was it had to have something to do with the stackable switches, since that is the only difference.  Well, actually there are two differences.  First, in the past our gig switches were 3560Gs.  The new switches are 3750s, and we have never stacked switches before.  In the past, each gets their own uplink.  As for the negotiated speeds, I was only given this report yesterday afternoon, so I did not have a chance to watch the ghost in progress.  What I was told was that the server was on the bottom stack, say 2/0/10, while the machines being ghosted were on the top stack, ports 1/0/2-1/0/21.
0
 
LVL 24

Expert Comment

by:rfc1180
Comment Utility
Something you can try to rule on the switch and StackWise is move the ghost server to the top stack, port 1/0/22 (or next available port). If it continues, then you will have to look and both switches while the issue is occurring. This might include using wireshark and spanning a port to monitoring the ghost vlan, you should also check the bandwidth utilization during the transfer and check for output drops of the interface (you already mentioned that there were no errors on the interface).

Other items to look at:

show proc cpu sort
 

Are you able to post your config?

Can post the output of this information:

show platform stack-manager all
show switch stack-ring speed
show int status
show proc cpu history

BIlly
0
 

Author Comment

by:hfccadmin
Comment Utility
Even if it doesn't resolve my issue, it's worth the question to get the commands from you guys ;-).  I'll post the info tomorrow.
0
How your wiki can always stay up-to-date

Quip doubles as a “living” wiki and a project management tool that evolves with your organization. As you finish projects in Quip, the work remains, easily accessible to all team members, new and old.
- Increase transparency
- Onboard new hires faster
- Access from mobile/offline

 

Author Comment

by:hfccadmin
Comment Utility
Another update, going on your suggestion that it could have been traffic saturation on the switch, I had all the ports hard set to 100mbps prior to the ghosting, and that seemed to resolve the issue, although it still doesn't tell us what the exact issue is.

Here's the info minus password info, and take into account all involved ports are now set to 100


switch.txt
0
 
LVL 24

Expert Comment

by:rfc1180
Comment Utility
If you had hard set the ports, did you hard set the servers as well?
That sounds like you had a duplex mismatch somewhere.

Billy
0
 
LVL 24

Expert Comment

by:rfc1180
Comment Utility
The idea is to have the ghost server set to 1Gigabits, and hard set all other ports to 100 Full (Including the servers), to ensure that you get one host chewing up all the bandwidth. You could always bond 2 or more ports on the ghost server for additional bandwidth too.

Billy
0
 

Author Comment

by:hfccadmin
Comment Utility
The ports were hard set on speed, but not duplex.  If all the machines originally connected at gig, which they did, it couldn't have been a duplex mismatch as gig is full no matter what., but those are some great suggestions and I appreciate all your help, especially leaving the server at gig.  Great idea.  Did you examine the rest of the data, and did you see anything there?
0
 
LVL 24

Accepted Solution

by:
rfc1180 earned 500 total points
Comment Utility
>The ports were hard set on speed
ahh, I did not look at the txt file as you said hard setting the port to 100Mbps fixed the issue. I just assumed you changed had hard-set the speed and duplex. I have looked at the config, and there does seem to be an issue:

         Stack Port Status             Neighbors    

Switch#  Port 1     Port 2           Port 1   Port 2

--------------------------------------------------------

  1        Ok        Down               2      None

  2       Down        Ok              None       1

Appears that you have a stack connection down between the switches;

I was expecting to see:

This might be part of the problem.

         Stack Port Status             Neighbors    
Switch#  Port 1     Port 2           Port 1   Port 2
--------------------------------------------------------
  1        Ok         Ok                2        2
  2        Ok         Ok                1        1

are you using one cable or two for the stack?


every link is 16Gbit/s, so both together give 32Gbit/s.

Here You can read about it:
http://www.cisco.com/en/US/products/...801b096a.shtml

....under section titled "Bidirectional Flow":

"To efficiently load balance the traffic, packets are allocated between two logical counter-rotating paths. Each counter-rotating path supports 16-Gbps of traffic for a total of 32 Gbps. The egress queues
calculate path usage to ensure that the traffic load is equally partitioned.

[...]

When a break is detected in a cable, the traffic immediately loops back across the single remaining 16-Gbps path to continue forwarding."

So if You use only one cable, for the StackWise You'll operate in "broken" mode.

Please note, switch configuration guide mentions correct two units connection model - here, in
figure 5-1:
http://www.cisco.com/en/US/products/...html#wp1167130

And to quick check it without going to CLI:
http://www.cisco.com/en/US/products/...html#wp1232305

Quote:
"After adding or removing stack members, make sure that the switch stack is operating at full bandwidth (32 Gbps). Press the Mode  button on a stack member until the Stack mode LED is on. The
last two port LEDs on the switch should be green. Depending on the switch model, the last two ports are either 10/100/1000 ports or small form-factor pluggable (SFP) module ports. If one or both of
the last two port LEDs are not green, the stack is not operating at full bandwidth."

Cisco ships StackWise cable with every 3750, so if You have two units You should have two cables to properly connect them together.


Billy
0
 
LVL 15

Expert Comment

by:deepdraw
Comment Utility
To make the ghost more efficient you can try pim sparse multicasting.
http://www.experts-exchange.com/Hardware/Networking_Hardware/Routers/Q_21526368.html
http://www.symantec.com/connect/forums/gss-25-ghost-casting-over-procurve-network-switches-vlans
 
looks like these people have managed to configure this.
 
Greg
0
 

Author Comment

by:hfccadmin
Comment Utility
billy,
Once again, thanks.  I thought the second stack cable was only needed in a three switch or more stack to give multiple paths back in the event of a failure.  I saw something on that and completely missed it.
0
 

Author Closing Comment

by:hfccadmin
Comment Utility
Excellent advice and help
0

Featured Post

How your wiki can always stay up-to-date

Quip doubles as a “living” wiki and a project management tool that evolves with your organization. As you finish projects in Quip, the work remains, easily accessible to all team members, new and old.
- Increase transparency
- Onboard new hires faster
- Access from mobile/offline

Join & Write a Comment

The Delta outage: 650 cancelled flights, more than 1200 delayed flights, thousands of frustrated customers, tens of millions of dollars in damages – plus untold reputational damage to one of the world’s most trusted airlines. All due to a catastroph…
PRTG Network Monitor lets you monitor your bandwidth usage, so you know who is using up your bandwidth, and what they're using it for.
This tutorial will walk an individual through the process of configuring basic necessities in order to use the 2010 version of Data Protection Manager. These include storage, agents, and protection jobs. Launch Data Protection Manager from the deskt…
This tutorial will walk an individual through the process of installing of Data Protection Manager on a server running Windows Server 2012 R2, including the prerequisites. Microsoft .Net 3.5 is required. To install this feature, go to Server Manager…

763 members asked questions and received personalized solutions in the past 7 days.

Join the community of 500,000 technology professionals and ask your questions.

Join & Ask a Question

Need Help in Real-Time?

Connect with top rated Experts

7 Experts available now in Live!

Get 1:1 Help Now