Details
-
Bug
-
Resolution: Fixed
-
Blocker
-
4.0.0
-
Security Level: Public
-
Untriaged
-
Centos 64-bit
-
Unknown
Description
Build
4.0.0-3310 (today's build)
Found during system test with XDCR and tpcc.
Steps
1. bucket ram quota: 12.5 GB, indexer ram quota: 256MB, 8 node cluster, 2 index nodes, 3 query nodes, all nodes have data services enabled.
2. Create 9 tpcc buckets of total size 11gb
3. Create secondary indexes on all buckets, one at a time.
> create index WH_ID on WAREHOUSE(W_ID) using gsi WITH {"defer_build":true}
|
{u'status': u'errors', u'errors': [{u'msg': u'GSI CreateIndex() - cause: Request Timeout', u'code': 5000}], u'results': [], u'metrics': {u'elapsedTime': u'2m0.744699604s', u'executionTime': u'2m0.744503832s', u'resultSize': 0, u'resultCount': 0, u'errorCount': 1}, u'requestID': u'9abb6e72-02d2-4d0b-8a20-d4771266c6e4', u'signature': None}
|
|
Index WH_ID on .61 is @ 0% and has remained so for an hour. No dcp activity seen on the indexing side for bucket WAREHOUSE.
|
|
WAREHOUSE 172.23.105.61:8091 WH_ID Created 0%
|
Attaching logs from .61. 4 other indexes have been built successfully on the same node.
Attachments
For Gerrit Dashboard: MB-15491 | ||||||
---|---|---|---|---|---|---|
# | Subject | Branch | Project | Status | CR | V |
52807,5 | MB-15491: Drop Index time out When deleting the index for a bucket when the bucket is deleted, make the call to lifecycle mgr async. Otherwise, deadlock can happen when the indexer is trying to make call to lifecycle mgr during the while lifecycle mgr is waiting for indexer to process the next DDL. | unstable | indexing | Status: MERGED | +2 | +1 |