Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations IamaSherpa on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

ZENworks Imaging 7

Status
Not open for further replies.
May 31, 2006
237
US
One thing I've noticed as we're gradually using more and more of the functionality of ZEN7, is that for whatever reason images take a very long time to create from a workstation, in the neighborhood of 215 minutes (3 hours 35 minutes) for a 4GB installation (compression set to Speed, the image file ends up being 2.5GB). Conversely, it takes only 18 minutes to bring the image down using multicast. When we created these same images in Ghost, it took around 18 minutes to create and maybe 10 minutes to come down, in the same network environment. Is there some setting in ZEN that I'm overlooking? It just seems like an inordinately long period of time that it takes to create the image. Bringing it down I'm not worried about so much, because we're not currently set up to use Ghost in multicast so we gain a lot when we do multiple machines (two PCs being imaged in Ghost can take upwards of 40 minutes, add another PC and we're looking at an hour and a half!). But I'm just wondering why the image creation takes so long.

Thanks in advance.
 
Something's wrong. I can get a 4GB image file created in about 20 minutes (compression set to speed). Are you sure you're not imaging a flaky disk?

I find Zen and Ghost to be about equal time-wise.
 
I've experienced the same creation times on several different workstations, so unless they're all flaky, I'd say, no it's not a flaky disk I'm imaging from.

I kinda thought there was something wrong, which is why I mentioned it here LOL.
 
The problem is your server, not Zenworks.


When you did the Ghost stuff, you did it from a Windows box, not a netware box. The network enviroment is the same, so yea, your right about it not being an issue with the network it's self. So as you know, the man in the middle could be the issue.

Things to look at:

What is the speed and duplex the PC being imaged, does it match whats on the server? Remember, half duplex gives you error correction, full duplex does not. Neither is better that the other from a technical stand point, but full can be faster. Server and PC should match in speed and duplex (as well as the switch port they are attached to).

How much memory is in the server?
Is it a Windows or netware server?

On Netware
What are your directory cache buffers set to?
What are your packet buffers set to?
How much space is being used by purgable files?

On Windows
Fdisk the damn box and use NetWare OES or SLES OES
(I'll have to look up the correct advice on a windows box, hate running ZfD on them my self, but I still do for client that require it. It works, it's just more of a pain than it needs to be in some cases. I will update when you tell me what kind of server your using).

=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
Brent Schmidt Certified nut case [hippy]
Senior Network Engineer
Keep IT Simple
 
Well, to begin with, our Ghost images are stored on a NetWare volume. We have no Windows servers here.

To answer your other questions:

Speed: Every box in our environment is set to 100/full. We're on HP Procurve switches.

Server: Dell PowerEdge 2550, dual P4-1Ghz processors, 2GB RAM, Perc3 RAID controller, Raid 5 (3x36GB), NW6.5 sp 5, ZENworks 7
Directory cache buffers: Min = 500, Max = 2000, Dirty Cache Delay time = 0.5, Directory Cache Allocation wait time = 2.2, Directory Cache Buffer referenced delay = 5.5
Packet buffers: Min = 2000, max=10000, size = 4224

As far as I know, these are the defaults, the guy before me set up these servers and says he never changed the defaults.

Purgeable space on this volume is 18GB. I realize that's pretty excessive, I don't think we've ever purged the volumes since I've been here (Dec. 2005).

I should also specify that the server we were pulling Ghost images from was a different box. That box is identical to this one, except that it only has 1.5GB RAM and it runs GroupWise 6.5.5 with two post offices, GWIA, and MTA. Purgeable space on that volume is 20GB.

I just created and pulled down a Ghost image on the GW server yesterday, three of them, and each time it took about 12 minutes.

Anything else you want to know? Other than passwords and directory structure I'll tell you everything (-:
 
First off, get that volume purged, it does slow down performance holding onto that stuff. You should put in place a policy that has you purging the volumes on a regular interval.

What else is the server doing? Does it do more than just host the ZfD agents?

Is the server setup to use Traditional or NSS volumes? (Forgot to ask that before).

Have you tried to place the ghost images on this netware box and pull them off it like you do with the other?

=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
Brent Schmidt Certified nut case [hippy]
Senior Network Engineer
Keep IT Simple
 
I'll try to answer all your questions to the best of my ability. As far as regular purging is concerned, I'm told that our normal policy is to purge the volumes once a week. Apparently, our backups have been shaky for quite some time, due in large part to the fact that they've not been able to perform incrementals for almost a year (a fact that I only learned recently), and also because our tapes are quite old and need to be replaced. These things are beyond my control, unfortunately, although I'm being positioned to take over those roles in the near future and that will be changed. But the bottom line is, I'm told that we won't be doing any purges until the backup issues are resolved. I hope that's soon.

The main function of this server, apart from being our primary ZENworks server, is housing a network install of Microsoft Office 2000. When I first came here and found out that workstations were running MS Office from a network sharepoint, I almost fell through the floor. I've been working on a project in conjunction with ZEN to get Office installed locally instead of from the network. It's an archaic way of running MS Office and it also creates a lot of problems with the way people hit Access databases. I realize that at the time it was implemented, it was due to small hard drives in the boxes, but at this point there's no reason for running Office from the network. But there you go, that's the only other thing this server is doing.

The server, like all our servers, mounts NSS volumes. There are no traditional volumes anywhere in our tree.

We have in fact put Ghost images on this box and pulled them down from it in the past. We were doing this because we were not using the multicast function in Ghost, and we were pushing down multiple images simultaneously. At the time we were doing about four images per night, updating the image on 120 faculty machines. So we'd run the Ghost image on two PCs from one server and two from the other. This one was one of them. We got the normal performance out of it.

Thanks!
 
On a lark, this morning I created a Ghost image off the same machine that I created the ZEN image from, into the same directory on the same volume (VOL1:\Images) on the same server. The Ghost image creation took 8 minutes 59 seconds.
 
I misspoke earlier about the server having dual P4 1Ghz processors. I thought that looked wrong. They're P3s. Here's the info from iManage:

Intel Pentium III processor
Speed 993 MHZ, Family 6, Model 8, Stepping 10
Feature flags 0383FBFF, CPUID: GenuineIntel
L1 cache 32K Bytes, , L2 cache: 256K Bytes

Still dual processors, and the other server has identical procs.
 
Avoiding purges due to backup failures ... seen this kind of thought process way to much. I can't stress enough how important it is to purge your volumes on a regular basis. Keeping data on the volumes because backup is broken doesn't really give you what you want. Because if you leave them there, and the server crashes and the hard drive fails, you saved your self nothing. Do this to help reduce the purgable files on your volumes. Look at your backup rotation, what kind of history do you provide your users? If your backup tapes are rotated every week (tape one is always used on a monday), then you have only one week history. Purge everything that is a over a week old. Cause if a user has not called you to restore a file with in a week, there is a good chance they no longer have a need for it anymore. This will help reduce the purgable files for you, and help server performance.

Now, personally, I have not used PXE on ZfD 7 on netware, been using it on OES, and I have had the luxury using top of the line brand spanking new workstations with linux driver compiled by me for these platforms. So there is a chance the issue is actually in the drivers your working with not performing well with the hardware they are running on the PC. So try this, download the linux.1 and linux.2 files from this site:


Rename the linux.1 & 2 files you find in the SYS:TFTP directory on your server and put these ones in their place. Then try the imaging over again and see what happens. These files are using a different kernel version, so I would like to see if there is any change by using them instead of hte defaults.

=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
Brent Schmidt Certified nut case [hippy]
Senior Network Engineer
Keep IT Simple
 
I'm not using PXE, I'm using the boot CD from the .iso file. I choose Manual, then when prompted I put in the IP address of the imaging server, then type "img" at the bash prompt. From there I hit F5 to create the image, put in the pertinent information and away it goes.

Now, is there a way to update the drivers on the boot CD? Perhaps that may be the key. When I use Ghost to put the images up, I use a DOS boot disk that loads the NIC driver and the VLM, then logs in to NDS.

I'm assuming that I don't see a SYS:\TFTP directory because I'm not running PXE?
 
correct, you do not see the TFTP directory because you may not have installed PXE. You need a tool like WinISO to open and edit the iso file. Hunt down the NIC driver you are using and see if you can find an updated one, you may need to compile it, and insert it into the ISO.

=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
Brent Schmidt Certified nut case [hippy]
Senior Network Engineer
Keep IT Simple
 
I couldn't see anywhere to put those files, those files don't even exist in the ISO image. Any other thoughts?

Bear in mind that bringing images down from ZENworks is fine, it's only when I try to create the image that the PC screeches to a crawl.
 
The problem has been resolved. I was told in another forum to set the switch port speed to Auto (instead of 100/Full) and when I pushed the image up from the box it took about 12 minutes. As opposed to 215. Thanks for all your input and advice (-:
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top