refactor handler Close func to use errgroup to be a bit less messy.
refactor pilosa.Server closing to actually return an underlying error if one occurs
add option to pilosa/test.Cluster and pilosa/server.Command to control close
timeout. currently is only used by the http handler, but conceivably could be
passed as a parameter to other subsystems of pilosa/server.Command
The http internal client's FragmentBlocks and Blockdata methods were being used
incorrectly, and incorrect respectively. One was not being passed a node URI by
monitorAntiEntropy, and the other was always using the defaultURI regardless of
what was passed to it. Antientropy was doubling not working because of this. I
think this crept in pretty recently, so hasn't actually affected anyone.
I exposed a SyncData method on Server so that we can invoke the anti entropy
task manually instead of trying to set up the interval so that it will run and
then sleeping and waiting for it to run. Now that this test works the way it
does, the other anti entropy test is obsolete, and I deleted it.
gossip now takes a single "MemberServer" which is implemented by server. Several
interfaces have been removed.
MemberServer contains ReceiveMessage which is a superset of the functionality of
ReceiveEvent, LocalStatus and HandleRemoteStatus are all that's left of
StatusHandler - ClusterStatus was not used and is gone. The Node() method is
actually a subset of LocalStatus() functionality. Maybe we should break up
localstatus or remove Node... not sure.
Remove BroadcastReceiver test which was a bit silly.
NodeEvent can now be unexported, and is.
since the gossip MemberSet has access to Server, it wasn't really necessary to
pass it a Node object when calling Open on it from Cluster. The end goal is to
have it be removed from Cluster entirely, and have it be Opened externally, and
this is a step toward that.
Exposing Node method on Server doesn't really expose any more than was already
there as the same info can be gotten from LocalStatus with a bit of type
casting. I figured adding the method was a little cleaner, and we could collapse
all the functionality when the dust has settled.
The Cluster.open method has been broken into two parts - one of which happens
earlier (at NewServer time), and the other will eventually just be "waiting to
make sure we've joined the cluster". Right now it's calling Memberset.Open, and
then waiting to make sure the cluster has been joined.
pilosa.Server now implements StatusHandler and EventReceiver and needs only
start a gossip memberset. A gossip member set now takes a server as an argument
explicitly and the maze of handlers and receivers and the starting sequence is
somewhat simplified.
Server now trivially implements EventHandler by passing the call along to its
Cluster object which has the actual implementation. This means that less things
will need to refer to cluster.
