Repository: kudu
Updated Branches:
refs/heads/master 2c5c9d06b -> 34bb7f93b
[rebalancer] location-aware rebalancer (part 1/n)
Added logic to identify and fix placement policy violations.
Also, added units tests to cover the new functionality.
Change-Id:
[rebalancer] location-aware rebalancer (part 3/n)
Some refactoring on the rebalancer code:
* Separated rebalancer-relevant information from KsckResults into
a dedicated ClusterRawInfo structure.
* Added ClusterInfo structure; ClusterBalanceInfo is aggregated
into the new ClusterInfo
[rebalancer] location-aware rebalancer (part 2/n)
Small refactoring on tools::Rebalancer: renamed
Rebalancer::ResetKsck() into Rebalancer::RefreshKsckResults()
and made the actual ksck utility run as a part of the renamed
method.
Change-Id: Ia1bdd4ee13d0522e49467b8b6e4e3a76ec560b26
Reviewed-on:
[rebalancer] location-aware rebalancer (part 4/n)
Refactored Rebalancer and Runner classes, separating common base for
a runner of the rebalancing process.
Change-Id: Id47183fc853573390b22ec714751adec93e0ea3a
Reviewed-on: http://gerrit.cloudera.org:8080/11745
Tested-by: Alexey Serbin
[rebalancer] location-aware rebalancer (part 6/n)
Added SetReplace() and CheckCompleteReplace() auxiliary fuctions.
A follow-up patch will start using those.
Change-Id: I80b560d70c4d7383ee89917a359b4bb2f41bfd31
Reviewed-on: http://gerrit.cloudera.org:8080/11747
Tested-by: Alexey Serbin
Repository: kudu
Updated Branches:
refs/heads/master 34bb7f93b -> f731ea004
[rebalancer] location-aware rebalancer (part 5/n)
Added LocationBalancingAlgo and corresponding units tests.
Change-Id: I78446fec8b8f80b7c6112bdd9d53f3dbf506
Reviewed-on: http://gerrit.cloudera.org:8080/11746
rpc-test: fix TestClientConnectionMetrics
Every now and then, this test would fail with:
rpc-test.cc:542: Failure
Expected: (dump_resp.outbound_connections(0).outbound_queue_size()) > (0),
actual: 0 vs 0
Unfortunately, the test would go on to crash (and trigger a TSAN warning)
due to the
Repository: kudu
Updated Branches:
refs/heads/master f731ea004 -> 3a77ba131
catalog_manager: fix a TSAN race
Noticed this on the flaky test dashboard for alter_table-randomized-test;
see the end of the commit message for the complete output. I also removed
an unrelated and unnecessary lock
Repository: kudu
Updated Branches:
refs/heads/master eb6f4e8a8 -> 2c5c9d06b
Improve consensus queue overflow logging
Suppose tablet server X is a leader of T tablets for which tablet server Y is a
follower. The relevant situation is when T is on the order of 100-1000. If Y
strains under its
[master] update placement logic for RF % 2 == 0
Updated the logic of the replica placement in master to properly handle
even replication factors. Added corresponding unit tests as well.
It's possible to configure Kudu to allow creation of tables with an
even replication factor. I think it's
10 matches
Mail list logo