Details
-
Bug
-
Resolution: Incomplete
-
Major
-
2.0-beta
-
Security Level: Public
-
2.0.0-1717-rel
Ubuntu 10.04 64bit
6 nodes
Description
1) Created 4 buckets in a 6 node cluster and loaded some data into each (about 100-500k) items
2) Created default design docs and views for each bucket and published to production to trigger indexing.
3) Wait for cluster to return to steady state (no indexing/compaction)
4) Start deleting buckets. The "default" bucket last
- NOTE: I was also flushing swap space at time of deleting default bucket
- swapoff -a && swapon -a
4) Bucket is eventually deleted and couch files gone, but real memcached crash occurred and UI reports:
Port server memcached on node 'ns_1@10.3.3.66' exited with status 134. Restarting. Messages: Thu Sep 13 14:35:07.194686 PDT 3: TAP (Producer) eq_tapq:replication_ns_1@10.3.121.91 - Suspend for 5.00 secs
...
Abort thread
...
Thread 1 (Thread 4086):
#0 0x00007fb266edea75 in raise () from /lib/libc.so.6
#1 0x00007fb266ee25c0 in abort () from /lib/libc.so.6
#2 0x00007fb261ebc110 in Mutex::acquire (this=0x31632058) at src/mutex.cc:84
#3 0x00007fb261ed24bb in LockHolder::lock (this=<value optimized out>) at src/locks.hh:48
#4 LockHolder (this=<value optimized out>) at src/locks.hh:26
#5 TapProducer::setSuspended (this=<value optimized out>) at src/tapconnection.cc:820
#6 TapResumeCallback::callback (this=<value optimized out>) at src/tapconnection.cc:774