I didn't get good results with the netbsd daemon at all. Environment is:

Initiator: Ultra5/Sol10 118833-24
Target: Ultra1/Sol9 118558-14

On the initiator side, I saw lots of 'transport failed' and 'received status 
out of order errors'. On the target side, thousands of 'Bad "Opcode": Got 0 
expected 5' and 'iscsi_write_data_decap() failed'

I was in fact able to discover the device, see it in format and newfs the thing 
(verrrry slowly, due to overhead with error handling :-) ), and even mount it, 
but it was unusable.

Another couple of observations about the process rather than the result:
1. Might it not be better to converge iscsiadm with cfgadm instead of using 
devfsadm for adding devices?
2. One thing that solaris-aware storage devices seem to do is supply format 
with a plausible looking geometry so you don't need to make one up. I tried 
serving a 1GB extent and using the Sun 1 GB disk definition, but despite the 
fact the any considerations of geometry are totally irrelevant, solaris wasn't 
having any of it and the results were not pleasing.

Now, as far as I can remember, solaris hasn't used format.dat geometry in its 
strategy routines since about 1996, so format is looking pretty darn long in 
the tooth. Either it needs a generic geometry for SAN/iscsi devices which just 
uses a size in blocks or some such, or all that nasty geometry stuff should be 
consigned to the dustbin of history.

However I doubt anyone is in a hurry to modify such a key utility.

MrChuck

PS is this project still alive?
 
 
This message posted from opensolaris.org
_______________________________________________
opensolaris-discuss mailing list
[email protected]

Reply via email to