What is the best way to simulate a disk I/O hang on Linux? I want to see how long a critical Oracle process can tolerate such a hang. If I delete SD disks by `echo 1 > /sys/block/sd*/device/delete', that causes the database to crash immediately. But I want to see if there's such I/O hang tolerance without causing an immediate I/O or disk error. If I send a STOP signal to a critical process such as DB writer (`kill -STOP <pid>'), the database runs normally except when a session needs the functionality of the stopped process (writing from memory to disk), this session hangs forever. But that's a test of hanging the Oracle process itself, not the same as hanging just I/O. Someone suggested SCSI fault injection test tool (https://github.com/dwalkes/scsi_fault_injection_test_tool). I haven't got it to work but it looks like an overkill for me. Yong -- redhat-list mailing list unsubscribe mailto:redhat-list-request@xxxxxxxxxx?subject=unsubscribe https://www.redhat.com/mailman/listinfo/redhat-list
