Details
-
Bug
-
Resolution: Fixed
-
Major
-
None
-
None
-
Security Level: Public
-
None
Description
Attaching all the tests ran against memcached buckets.
1) Basic tests - the gets fail, whereas the sets pass.
2) Failover and readd - Here there are many assertion errors with the I/O being problematic
3) Rebalance - Here there are many assertion errors with the I/O being problematic
In 2nd and 3rd cases, the probable cause may be the following -
[SDKD(WARNING) 268.72 cbsdk.sdkd.local executor.py:66] Exception in thread "Memcached IO over
{MemcachedConnection to /10.3.121.212:11210 /10.3.121.208:11210}" java.lang.AssertionError: Attempting to overwrite channel
[SDKD(WARNING) 268.72 cbsdk.sdkd.local executor.py:66] at net.spy.memcached.protocol.TCPMemcachedNodeImpl.setChannel(TCPMemcachedNodeImpl.java:496)
4) Failover - Here there are failures in the rebalance phase, bt the rebound phase breaks in SDKD. Another SDKD issue is filed to take care of the same.
We also need to investigate the SDKD code. There might be something wrong there in handling the memcached buckets. But, till now the issue looks more likely to be with the java sdk.
Attachments
Issue Links
- is duplicated by
-
JCBC-271 Adding a node to an existing cluster causes issues with a running memcached bucket
- Resolved