Coda File System

a code logic question (Jan?)

From: <>
Date: Sun, 6 May 2012 10:43:43 +0000

I am a bit unsure when/how the Server"N" members of the VolumeInfo structure
were supposed to be used.

(note that it is the use of the server addresses supplied by the RPCs
which I am working to get rid of, but I want to be sure that I understand
the intentions of the previous code)

As far as I can see, for volume replicas (and ro-clones and backup volumes)
it is Server0 which contains the server address, the rest of the structure
is mostly unused (i.e. the other ServerN and the whole RepVolMap).

I can not see where a client would use other addresses than Server0
which are supposed to be present (?#1) in a replicated volume's VolumeInfo.

Do servers use them?#3 (during resolution?#2)
Do I understand correctly that the extra addresses are not being used by the
client at all?#4

As I interpret it, the extra server address fields are there as shortcuts
to avoid (sometimes?#5 when?#6) the need to do extra RPCs per replica.
Is it correct?#7

Counting question marks above shows that there are 7 questions,
All of those besides #5 and #6 are near one-bit ones  :)

Would somebody (Jan?) shed light on this?

Received on 2012-05-06 06:44:57