Ulf Wiger (TN/EAB)
Wed Aug 9 00:20:55 CEST 2006
There is a new version of gen_leader here:
(It's not in Jungerl yet - sorry.)
It has a new leader election algorithm, which is much better suited to
Erlang's semantics. The code has been tested using abstract traces (see
the paper) and QuickCheck.
The API is still the same.
[mailto:] On Behalf Of Alex Arnon
Sent: den 8 augusti 2006 21:40
Subject: gen_leader's status.
What is gen_leader's current status? Looking through my mailing
list archives, it seems that it was considered to be functioning well if
not thoroughly tested. Can anyone give me a heads-up on this?
gen_leader in fact implements a superset of what I need, so I
can actually get by with a somewhat looser scheme using pg2, I think.
- I need a group of processes (1 per node) of which only one
periodically wakes up and performs cleanup on some mnesia tables.
- It is probably not too bad if occasionally (due to node
crashes etc.) more than one would wake up, as I believe it should be
simple to avoid if nodes are reasonably stable.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the erlang-questions