feat(lb) add req_dc_rr policy #114
Conversation
kikito
force-pushed
the
feat/req-dc-rr
branch
2 times, most recently
from
21b3209
to
15fd153
Compare
| self.remote_tried = 0 | ||
| if ngx and ngx.ctx and ngx.ctx.cassandra_coordinator then | ||
| self.cassandra_coordinator = ngx.ctx.cassandra_coordinator | ||
| end |
There was a problem hiding this comment.
Instead, I think we could do:
self.ctx = ngx and ngx.ctx
self.initial_cassandra_coordinator = self.ctx and self.ctx.cassandra_coordinator or nil And in next_peer, simply:
if i == 1 and state.initial_cassandra_coordinator then
end
-- ...
if state.cassandra_coordinator == peer then
end
if state.ctx then
state.ctx.cassandra_coordinator = peer
endThis way we avoid all of the global ngx accesses and the metatable ones for ctx
t/14-lb_req_dc_rr.t
Outdated
| ngx.say() | ||
| for i, peer in lb:iter() do | ||
| ngx.say("1.", i, ' ', peer.host) | ||
| ngx.log(ngx.INFO, i, ' ', peer.host) |
There was a problem hiding this comment.
It seems like this log is unecessary
| 2.5 10.0.0.3 | ||
| 2.6 10.0.0.1 | ||
| --- no_error_log | ||
| [error] |
There was a problem hiding this comment.
Shouldn't we have a test that calls lb:iter() once, then breaks, then calls it again, breaks again, then calls it again, and all 3 calls it should return the same peer? I don't see such a test, which I believe reflects the basic desired behavior for this new policy? The first test somewhat reflects that behavior, but does not break after each invocation, instead, it serves the last iteration's last returned peer as the first peer of the new iteration.
There was a problem hiding this comment.
Added the extra test, thanks for the review!
From #114 Signed-off-by: Thibault Charbonnier <thibaultcha@me.com>
|
Thanks! Manually merged along with a new |
No description provided.