История изменений

Power-failure test script (located at git://git.cogentembedded.com/pub/celf/pf-test) performs unattended simulation of power failures and analyzes file system and data integrity check results. Detailed execution log is saved for further examination. Power-fail test server residing on host manages the execution of various remote tests on target with particular file system configuration set up, and controls power source to initiate power outages at random time instants during remote test execution. Test server connects to the target and retrieves logging and diagnostic information. The following steps are executed in a loop for each remote test and file system: 1. connect to the target; 2. perform mandatory file system check and recovery; 3. perform data integrity check if applicable; 4. mount current file system configuration; 5. start background test; 6. initiate power outage after random interval. Methodology The following file system workloads are proposed: • fsstress, that would stress file system metadata; • custom SQLite test scripts, that would reveal occasional file data corruption. Fsstress proved to be quite effective in detecting various file system implementation issues in the past. And SQLite is known for its extra care about power failures. Therefore, if SQLite database becomes corrupted, the most likely reason for this lies in file system or physical storage layers.

ну?

Power-failure test script (located at git://git.cogentembedded.com/pub/celf/pf-test) performs unattended simulation of power failures[/bb] and analyzes file system and data integrity check results. Detailed execution log is saved for further examination. Power-fail test server residing on host manages the execution of various remote tests on target with particular file system configuration set up, and controls power source to initiate power outages at random time instants during remote test execution. Test server connects to the target and retrieves logging and diagnostic information. The following steps are executed in a loop for each remote test and file system: 1. connect to the target; 2. perform mandatory file system check and recovery; 3. perform data integrity check if applicable; 4. mount current file system configuration; 5. start background test; 6. initiate power outage after random interval. Methodology The following file system workloads are proposed: • fsstress, that would stress file system metadata; • custom SQLite test scripts, that would reveal occasional file data corruption. Fsstress proved to be quite effective in detecting various file system implementation issues in the past. And SQLite is known for its extra care about power failures. Therefore, if SQLite database becomes corrupted, the most likely reason for this lies in file system or physical storage layers.

ну?