http://www.couchbase.com/issues This file is an XML representation of an issue en-us 5.2.4 845 26-12-2012 http://www.couchbase.com/issues/browse/MB-7523 Couchbase Server Recent benchmarks: <a href="http://dashboard.hq.couchbase.com/litmus/dashboard/">http://dashboard.hq.couchbase.com/litmus/dashboard/</a> <br/> <br/> Summary: <br/> <br/> Rebalance-in ( 2-4 nodes), 7M items, mixed workload: <br/> <br/> ec2: 2.0.0-1976 (RTM) took 1687 sec while 2.0.1-116 took 3581 sec <br/> thor (data center physical machines): 2.0.0-1976 (RTM) took 901 sec while 2.0.1-123 took 1553 sec <br/> <br/> Rebalance-out (4-2 nodes), 7M items, mixed workload: <br/> <br/> ec2: 2.0.0-1976 (RTM) took 2854 sec while 2.0.1-116 took 5075 sec <br/> thor (data center physical machines): 2.0.0-1976 (RTM) took 1142 sec while 2.0.1-123 took 2773 sec <br/> <br/> linux MB-7523

rebalance performance regression in 2.0.1 vs 2.0.0 (apparently only under load)

Bug Major Resolved Fixed Public Aleksey Kondratenko Ronnie Sun Fri, 11 Jan 2013 12:12:15 -0600 Fri, 18 Jan 2013 17:04:27 -0600 Fri, 18 Jan 2013 17:04:27 -0600 2.0.1 ns_server 0 1 diags: <br/> <br/> thor (data center machines), build 2.0.1-123-rel: <br/> <br/> reb-1: <a href="http://172.23.96.10:8080/job/thor-parent/112/">http://172.23.96.10:8080/job/thor-parent/112/</a> <br/> <br/> reb-1-out: <a href="http://172.23.96.10:8080/job/thor-parent/110/">http://172.23.96.10:8080/job/thor-parent/110/</a> + summary of test view views: <br/> <br/> <a href="https://docs.google.com/spreadsheet/ccc?key=0AgLUessE73UXdDV1SXhUZjJ0b0RhU3gtdlUzZGloUFE#gid=0">https://docs.google.com/spreadsheet/ccc?key=0AgLUessE73UXdDV1SXhUZjJ0b0RhU3gtdlUzZGloUFE#gid=0</a> Folks, I still see no results from 2.0.0 with +A. May I insist on having some ? Alternatively we can try 2.0.1 _without_ +A Did some runs comparing 2.0.0 and latest branch-2.0.1 and I&#39;ve found 2.0.1 to be _faster_. <br/> <br/> All my data fits in page cache plus I didn&#39;t send any mutations during rebalance. Plus I&#39;ve allowed all nodes to use just one (same) CPU core. I&#39;ll rerun with binding different node to different cores. assigning this back to Ronnie as Alk is expecting results for 2.0.0 with +A Hi Aleksey, <br/> <br/> I added first results for reb-out with +A, it shows &gt;1.5x regression (see link above). I&#39;m gathering more results but apparently it takes time. <br/> <br/> Anyway, I believe KV results with +A will be even more helpful. No. Let&#39;s have _separate_ bug for rebalance with views. Given that I&#39;ve tried rebalance myself and saw no regression (in fact speedup) in 2.0.1 I think we can start assuming that problem only occurs if rebalance is performed under load. <br/> <br/> BTW, I&#39;ve also confirmed personally that erlang core performance was not regressed in 2.0.1 (i.e. we have added -fno-strict-aliasing to CFLAGS). <br/> Fix is merged as part of chain ending at: <a href="http://review.couchbase.org/24067">http://review.couchbase.org/24067</a><br/> <br/> I&#39;ve found that my original plan to try to end serial phase of vbucket move at end of backfill helps a lot. Date of First Response Mon, 14 Jan 2013 14:23:10 -0600 Rank 3329 Sprint Status Time In Status