Couchbase

Couchbase http://www.couchbase.com/issues This file is an XML representation of an issue en-us

5.2.4

845

26-12-2012

[MB-6974] erlang FULL_SWEEP setting needs override to 512 instead of erlang's default http://www.couchbase.com/issues/browse/MB-6974 Couchbase Server Looks like 512 is the better setting than the default... <br/> <br/>   <a href="https://docs.google.com/spreadsheet/ccc?key=0AgLUessE73UXdGd2NmhLOThCRUc0ekFkZ0FDeGdLRXc#gid=0">https://docs.google.com/spreadsheet/ccc?key=0AgLUessE73UXdGd2NmhLOThCRUc0ekFkZ0FDeGdLRXc#gid=0</a> <br/> <br/> MB-6974

erlang FULL_SWEEP setting needs override to 512 instead of erlang's default

Bug Blocker Resolved Fixed Public Aleksey Kondratenko Steve Yen Fri, 19 Oct 2012 13:03:53 -0500 Mon, 12 Nov 2012 15:09:32 -0600 Tue, 23 Oct 2012 19:42:24 -0500 2.0-beta 2.0-beta-2 ns_server 0 0 Changing title to be clearer. <br/> <br/> Also, I spoke with Damien and he's good with 512, too. You don't wait for system test results? You don't need extra perf runs for that? We can always revert later. But I'd wait at least a bit. Tommie,<br/> <br/> can you update the ticket once you have initial results from pine and plum cluster where we are running the cluster with this settings ? Current status of verification on key-value and query clusters: <br/>   <br/> Overall looks good on 23 node with key-value workload with SWEEP=512: <br/>  * load 20 Million items, size 356bytes <br/>  * access phase: --get 70% --create 10% --update 15% --delete 5% --ops 40000 <br/>  * rebalance out 3 nodes completed <br/>  * load until 44% active resident <br/>  * access phase with cache_miss ratio between 1-3% <br/>  * swap rebalance in 3 nodes, out 3 nodes completed <br/> <br/> Timeouts on 4 node cluster with views and SWEEP=10000: <br/>   * load 13 million items into 2 different buckets <br/>   * run 300/queries-per-sec against each bucket <br/>   * access phase at 70% active resident <br/>   * rebalance out 1 node <br/>         - Failed with etimeout <br/>         - usually able to retry and will succeed <br/> <br/> On 4 node cluster with views and SWEEP=512: <br/>    * verification pending... assigning back to alk to get the 512 setting changed. <br/> Recent system test verification failed due to another issue <a href="http://www.couchbase.com/issues/browse/MB-6490" title="Rebalance failed with reason "Partition 687 not in active nor passive set" in add in node rebalance"><strike>MB-6490</strike></a>….Instead of the timeouts we saw at higher gc levels. changing priority to blocker (from critical) Change is in gerrit and will be merged soon Aliaksey A. had a good point on the fix, where we need a corresponding fix for the windows? added to RN: By default we provide garbage collection more frequently than the <br/> normal default for Erlang. This keeps memory usage by the Erlang <br/> virtual machine lower, and enables better performance. Date of First Response Fri, 19 Oct 2012 16:37:44 -0500 Rank 3699 Time In Status