Details
-
Bug
-
Resolution: Fixed
-
Blocker
-
2.0-developer-preview-4
-
None
-
Security Level: Public
-
None
Description
Created 10 nodes cluster. Created a view: {"reduce":{"map":"function (doc)
{\n emit(doc._id, null);\n}","reduce":"_count"}. Uploaded 100k json documents using mcsoda. Queried a view with stale=false. Removed one node from the cluster. Queried the view with stale=false again while rebalance was in progress. View returned partial result:
{"rows":[
{"key":null,"value":88693}],
"errors":[
{"from":"http://10.17.1.200:9503/_view_merge/?stale=false","reason":"{<<\"noproc\">>,<<\"{gen_server,call,[<0.2544.0>,
,infinity]}\">>}"}
]}
This state is permanent: the view keeps returning partial result with this error. There are plenty of compaction_loop_died errors in the log on the failed node:
=========================CRASH REPORT=========================
crasher:
initial call: couchbase_compaction_daemon:init/1
pid: <0.12968.0>
registered_name: couchbase_compaction_daemon
exception exit: {compaction_loop_died,
{noproc,
{gen_server,call,
[<0.2544.0>,
,infinity]}}}
in function gen_server:terminate/6
ancestors: [<0.12967.0>,ns_server_sup,ns_server_cluster_sup,<0.42.0>]
messages: []
links: [<0.12967.0>]
dictionary: []
trap_exit: true
status: running
heap_size: 233
stack_size: 24
reductions: 1074
neighbours:
Attaching logs and diag from the node where the error happens.