EXTREMELY slow imaging

[#1]

Hi all, long time listener, first time caller...

I've been using Casper for about a month now, we've completely migrated our imaging processes over to JSS/Casper Imaging, and for the most part everything is awesome.

Except that every once in a while I'll get a system that is slow as Christmas in taking a new image. This makes no sense to me at all since I'm on a gigabit connection to my server, and it's the ONLY system being imaged. I suppose there could be heavy network overhead, but I doubt it since it's the summer and nearly everyone is out of the office.

Has anyone else seen this behavior, and if so, did you find a way around it?

Thanks,

Pete Wann
Apple Systems Administrator
Texas Christian University

[tlarkin]

Pete

I have seen this happen for many different reasons. How are you set up to image? Unicasting, multicasting? AFP can be fickle and slow if you don't kick off the imaging all at the same time.

Also, disk I/O is always going to be your slowest bottle neck. HDs that are about to fail always image slow, or if your NIC is going bad I have seen the same thing.

---

Thomas Larkin
TIS Department
KCKPS USD500
tlarki at kckps.org
blackberry: 913-449-7589
office: 913-627-0351

[John_Wetter]

In your AFP server, add more memory, and then some more memory... I'd say 8GB is a minimum. Of course, if this is only occasional as you elude to, then it might be a server utilization issue, even if the server is doing a bunch of SUS downloads, or something is running with errors so it's logging like crazy. Also, is it new enough hardware to be all gigabit? As Tom mentioned, a really long image time or a bad image can also be a signal that the HDD is about to completely flake out.

Thanks,
John

--
John Wetter
Technology Support Administrator
Educational Technology, Media & Information Services
Hopkins Public Schools
952-988-5373

[tlarkin]

Last summer my co-workers and I reimaged all of our Macbooks with a dual booting image of OS 10.5 and XP Pro SP2. We set up 20 to 25 clients at a time and unicasted an image to them, the OS 10.5 image was around 8 or 9 gigs in size and the XP Pro image was around 5 gigs in size. Attached is a screen shot of AFP throughput in a graph. If you notice there is one graph that kind of tapers off towards the end of the imaging process. That was due to either us not launching all the clients at the same time, because AFP will try to load balance all the connections and your throughput goes down, or we had some Macbooks with poor performing disk I/O.

---

Thomas Larkin
TIS Department
KCKPS USD500
tlarki at kckps.org
blackberry: 913-449-7589
office: 913-627-0351