Hello, we are using MINA 2.0.7 in our servers and clients. Recently (twice, on two separate servers), we saw a rather unexpected phenomenon: MINA's scheduled write messages kept rising (over about half a day) until our server ran out of memory. This has not happened before.
There were also a number of connections stuck in CLOSE_WAIT, which usually don't occur on our servers. At the end we had more than 10 million scheduled messages with more than 500 MByte of scheduled data. Screenshot of our munin graphs when this occurred: http://xehonk.de/files/2013-12-09/mina-scheduled-write-messages.png How do I prevent the scheduled write messages from queuing up to such a large amount? Thanks in advance Henning Rohlfs MINA settings we're using: sessionConfig.setIdleTime( IdleStatus.BOTH_IDLE, 0 ); sessionConfig.setIdleTime( IdleStatus.WRITER_IDLE, 0 ); sessionConfig.setIdleTime( IdleStatus.READER_IDLE, 1800); 2013-12-04 06:29:05,498 INFO [main]: - defaultLocalAdresses: [null] 2013-12-04 06:29:05,498 INFO [main]: - minReadBufferSize: 64 2013-12-04 06:29:05,498 INFO [main]: - maxReadBufferSize: 65536 2013-12-04 06:29:05,498 INFO [main]: - readBufferSize: 2048 2013-12-04 06:29:05,498 INFO [main]: - readerIdleTimeMillis: 1800000 2013-12-04 06:29:05,499 INFO [main]: - receiveBufferSize: -1 2013-12-04 06:29:05,499 INFO [main]: - sendBufferSize: -1 2013-12-04 06:29:05,499 INFO [main]: - throughputCalcMillis: 3000 2013-12-04 06:29:05,499 INFO [main]: - writerIdleTimeMillis: 0 2013-12-04 06:29:05,499 INFO [main]: - writeTimeoutMillis: 60000
