All jobs that are marked with a ! and errored status (different from failed) have been killed by travis due to exceeding the max runtime of 50 minutes:
The job exceeded the maximum time limit for jobs, and has been terminated.
In this case, some parts of big suites like main, rpl and binlog may never get to attended. The solution would be to reconsider the current test suite matrix split inside .travis.yml:
- GCC_VERSION=4.8 TYPE=RelWithDebInfo MYSQL_TEST_SUITES=rpl
- GCC_VERSION=5 TYPE=RelWithDebInfo MYSQL_TEST_SUITES=main,archive,optimizer_unfixed_bugs,parts,sys_vars,unit,vcol,innodb,innodb_gis,innodb_zip,innodb_fts
- GCC_VERSION=6 TYPE=RelWithDebInfo MYSQL_TEST_SUITES=binlog,binlog_encryption,encryption,rocksdb
- GCC_VERSION=6 TYPE=RelWithDebInfo MYSQL_TEST_SUITES=csv,federated,funcs_1,funcs_2,gcol,handler,heap,json,maria,perfschema,plugins,multi_source,roles
Maybe split big suites like main, innodb. binlog into separe jobs for a start and see the change? I can try that out on my own Travis setup first: