My configuration is VMware on Red hat 64bit linux OS. Memcached 1.4.5 and Xmemcached 1.3.5. I have a program which does eager loading of data, which is about 3GB, this happens through a multithreaded application, with 10 threads I have also created a connection pool of 50 connections to memcached. There is also a listener which tells when there is disconnection or the connection is healed.
Sometimes the program works fine without errors, but at times when I see disconnections and connections happening many times. And the memcached server gives the following errors. Failed to write, and not due to blocking: Connection reset by peer Failed to read, and not due to blocking: errno: 104 Connection reset by peer rcurr=127980b8 ritem=1442bd8a rbuf=12797c00 rlbytes=550 rsize=2048 When this happens the data is not fully loaded. At times this issue is resolved after the VM is rebooted. but that is not always the case. I tried reducing the connection pool size to 1 and thread pool to 1, even then the disconnections could happen What could be the possible reasons for this.
