Details
-
Bug
-
Resolution: Fixed
-
Critical
-
3.0
-
Security Level: Public
-
Untriaged
-
Yes
-
June 30 - July 18
Description
Using cluster run with 16 vbuckets:
export COUCHBASE_NUM_VBUCKETS=16
and the following test from pyupr:
./pyupr -h 127.0.0.1:9000 -b dev -o test_open_n_consumer_producers
if you run this a few times, a bucket delete timeout will occur:
[ns_server:debug,2014-05-28T9:08:20.447,n_0@127.0.0.1:<0.3453.0>:ns_pubsub:do_subscribe_link:136]Parent process of subscription
{buckets_events,<0.3451.0>} ex
ited with reason {#Ref<0.0.0.44817>,{timeout, ['n_0@127.0.0.1']}}
[ns_server:warn,2014-05-28T9:08:20.447,n_0@127.0.0.1:<0.357.0>:ns_orchestrator:idle:622]Nodes ['n_0@127.0.0.1'] failed to delete bucket "default" within expected time.
A default bucket cannot be created after this point.
The test tries to open consumer/producer 1024 times. It works for 1024 vbuckets since it runs once per vbucket, but with 16 this is done 64 times per vbucket. But same behavior can occur if you try do to it 4096 with 1024 vbuckets.