[dvd::rip] cluster mode questions

Hi,

I have been using dvd::rip (currently 0.98.1 on ubuntu edgy) for a little while now and have a few questions. I have just gotten a few used machines and have set them up as transcode nodes. My main machine, the one that hosts the daemon and the hardrive is a 2ghz core duo macbook with 2gb ram. My second fastest is a 3.2ghz p4 with 1gb ram, 3rd is a 2.6ghz p4 with 768mb ram. Then I have a 2ghz celeron with 512mb ram and a 1ghz celeron with 256mb ram. I also have a 733mhz p3 with 320mb ram. They are all connected via 100mb ethernet and samba. My first question is what would be the optimal settings for their speed indexes? This is how I have it as of now:

Macbook:        100
P4 3.2ghz:      80
P4 2.6ghz:      60
Celeron 2.0ghz: 40
Celeron 1.0ghz: 20
P3 733mhz:      20

Do the actual numbers make any difference other than to rate how fast one is in relation to the others? If I put them all in sequence like the fastest as 100, then 90, then 80, etc., would that change anything as opposed to how I have it set up now with 100, then 80, then 60, etc.? I put my slowest machines (avg. transcode 2nd pass at 8fps) as 20 but they still sometimes get jobs even when my fastest (avg. 2nd pass at 50-60fps) is idle. I have done some experimenting but have not noticed anything major. I am open to suggestions on how I should index my nodes. Also, sometimes when I am running a project on the cluster, one or more of my machines simply stop transcoding. Their state stays as "running" for a few minutes then they go to idle while whatever job they were working on freezes at about 98 percent. I get no errors. I try to stop them then restart them but they go right back to idle and can't get them to pick up any more jobs, all the machines have done it at some point. My solution so far is to just wait for them all to go idle and then cancel the project and immediately restart it, that seems to reset the broken jobs and then they get picked up.

Another question:

All of my dvds are NTSC. I always have a/v sync problems with cluster mode, even the ones with only one PSU. They seem to start out fine and then get out of sync by about 10-15 mins, getting worse throughout the movie. I have tried using extra options like -M 0, -M 2, -M 4 but none seem to make that much of a difference. My question is could any of this be caused by slower machines transcoding chunks which are then mixed with chunks made by faster machines? What about chunk size? I know that the content *should* be the same other than it takes a lot longer. When I add the extra options, can I just add it on the TRANSCODE tab before I send it to the cluster or do I need to add it to each node's profile? I've got transcode 1.0.2 on all of them. When working on a 23min tv episode with 40000 frames I've been using 3000 frame chunks, 5000 frame chunks, even 300 frame chunks (really easy on the slow machines with limited ram, they don't have to feel so slow.) At what chunk size does a 2-pass encode become useless? 300 frame chunks are easier on the slow machines and easier on the network but are they really useable?

I love the cluster mode but wish I could overcome some of these issues.
Any help would be great!!

Thanks,

ben