Github user aweisberg commented on a diff in the pull request:
https://github.com/apache/cassandra/pull/224#discussion_r187668202
--- Diff: src/java/org/apache/cassandra/dht/RangeFetchMapCalculator.java ---
@@ -158,14 +159,15 @@ static boolean isTrivial(Range<Token> range)
boolean localDCCheck = true;
while (!added)
{
- List<InetAddressAndPort> srcs = new
ArrayList<>(rangesWithSources.get(trivialRange));
+ ReplicaList replicas = new
ReplicaList(rangesWithSources.get(trivialRange));
// sort with the endpoint having the least number of
streams first:
- srcs.sort(Comparator.comparingInt(o ->
optimisedMap.get(o).size()));
- for (InetAddressAndPort src : srcs)
+ replicas.sort(Comparator.comparingInt(o ->
optimisedMap.get(o.getEndpoint()).size()));
+ Replicas.checkFull(replicas);
+ for (Replica replica : replicas)
{
- if (passFilters(src, localDCCheck))
+ if (passFilters(replica, localDCCheck))
{
- fetchMap.put(src, trivialRange);
+ fetchMap.put(replica.getEndpoint(), trivialRange);
--- End diff --
So this is something I have been running into on my end, but why unwrap
here? I have been need to fix that because I need the transientness of what I
am fetching for to determine whether the remote side is going to send me the
transient data or the full data.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
