summaryrefslogtreecommitdiff
path: root/test
Commit message (Collapse)AuthorAge
* TailorizationChris Wilson2008-04-04
| | | | | | | | | | | Import of the upstream sources from Repository: http://localhost:8000/ Kind: hg Revision: 7807b7768163f1c2537756abe5416063989cebb1 Original author: tailor@rocio.int.aidworld.org Date: 2008-03-16 19:44:36+00:00
* Add missing space in trace output.Chris Wilson2008-04-04
|
* Remove uses of obsolete TRACE macros in tests.Chris Wilson2008-04-04
|
* Allow configuration of the server port that the client will connect to Chris Wilson2008-03-28
| | | | | | | | | | | | | (bbackupd and bbackupquery). Redesign ConfigurationVerify to use classes instead of structs. Use port 22011 instead of 2201 during tests, to reduce the chances of conflicting with a running bbstored or other process. Ignore autogen_* in svn:ignore everywhere instead of individual per-file ignores.
* Fix missing space before bbstored_args in test to make them work Chris Wilson2008-03-13
| | | | | properly.
* Fix multiple connections to server (not supported on Windows) in test Chris Wilson2008-03-01
| | | | | | | that redundant locations are deleted on time. Commonise some more login code to simplify tests.
* Disable another symlink test on Win32.Chris Wilson2008-03-01
|
* Delete bbstored PID file after killing the daemon on Windows, as the Chris Wilson2008-03-01
| | | | | daemon can't clean up after itself.
* Add a sleep after every access to the read-only connection, to ensure thatChris Wilson2008-01-31
| | | | | | subsequent changes in the other connection get a new whole-second timestamp and hopefully fix test failures reported by Alex Harper on MacOS X.
* Ignore terminal type error messages from bbackupquery, may happenChris Wilson2008-01-23
| | | | | in Debian buildds.
* Send data back from test daemon to test code, and wait for it to beChris Wilson2007-12-18
| | | | | | | received in the test code, to avoid synchronisation issue where the QUIT command appears to be part of the received data. Mostly harmless, but gives a false alarm when viewing the test output.
* Pass test options to test daemons.Chris Wilson2007-12-16
|
* Report line of test where failed to kill internal daemon during SSLChris Wilson2007-12-15
| | | | | keepalive tests.
* Move comment, add whitespace.Chris Wilson2007-12-13
|
* Fix keepalive test failure: Chris Wilson2007-12-06
| | | | | | | | | * zero memory buffer before use to keep consistent behaviour. * expect a diff even if the maximum diffing time expires * really check when we have a diff and when we don't
* Re-initialise the timers when aborting test run during SSL KeepAliveChris Wilson2007-12-02
| | | | | | tests, as main() will try to clean them up and we'll get an error message if they're not initialised.
* Another compile fix for [1931]Chris Wilson2007-11-09
|
* Compile fix for [1931]Chris Wilson2007-11-09
|
* Handle bbackupd_args without leading space.Chris Wilson2007-11-07
|
* Fix calling bbackupd with NO args in keepalive tests (fixes [1928])Chris Wilson2007-11-07
|
* Pass bbackupd_args to the internal daemon (spaces are NOT supported).Chris Wilson2007-11-06
| | | | | Stop early if a keepalive test fails, for debugging.
* This is C++ not ruby (oops)Chris Wilson2007-11-06
|
* Add some debugging for a keepalive timer test failure reported byChris Wilson2007-11-06
| | | | | Matt Brown <matt@mbrown.co.uk> on 2007-11-06.
* Use lstat() instead of stat() to check whether a file exists. Thanks toChris Wilson2007-11-05
| | | | | | | | Hans-Joachim Baader for reporting this problem. (http://lists.warhead.org.uk/pipermail/boxbackup/2007-November/003958.html) Add a test that symlinks are not followed during restore.
* Add some sleeps to ensure that timestamps change enough to invalidateChris Wilson2007-11-04
| | | | | | the directory cache in the store on filesystem with low timestamp resolution (e.g. OSX).
* Wait for long enough after creating test files that they will actuallyChris Wilson2007-11-04
| | | | | | | be seen as old enough to upload by bbackupd. Fix comments.
* Include the time taken by bbackupquery compare in the time betweenChris Wilson2007-11-04
| | | | | | syncs, used to check that the SyncAllowScript pauses backup for the correct length of time.
* Compile fix for [1908].Chris Wilson2007-11-04
|
* Explicitly wait for sync to finish instead of assuming that 14 secondsChris Wilson2007-11-04
| | | | | is enough time.
* Move sync helper functions from test/bbackupd to Test.h to share withChris Wilson2007-11-04
| | | | | other tests, particularly test/backupstorefix.
* Test that redundant locations are really deleted after the time expires.Chris Wilson2007-10-08
|
* Convert most printf() and fprintf() calls to use logging frameworkChris Wilson2007-08-02
| | | | | instead. (refs #3)
* Remove compare again. (refs #3, merges [1651])Chris Wilson2007-07-26
|
* Make Configuration take a std::string filename instead of a char array,Chris Wilson2007-07-26
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | in C++ style. Add a function to get default config file paths at runtime, dependent on the location of the executable being run. Pass the config file name directly to Daemon::Main, instead of faking argv. No default raid file path at compile time on Windows, depends on executable location when run. Determine RaidFile path at runtime if not supplied in config file on Windows. Don't define default locations for config files at compile time on Windows, provide macros to determine them at runtime instead. Make FileHandleGuard take a std::string instead of a char array, C++ style. Determine config file location at runtime instead of hard-coding on Windows. Thanks to Paul MacKenzie, Per Thomsen, Pete Jalajas, Stuart Sanders, Dave Bamford and Gary for pushing me to do this. (fixes #12) Determine config file path at runtime. Call Daemon::Main with config file name instead of building fake argv. (refs #3, merges [1684] [1685] [1686] [1687] [1688] [1689] [1690] [1691] [1692])
* Fix regex tests for HAVE_REGEX_SUPPORT. (refs #3, merges [1682])Chris Wilson2007-07-26
|
* Fix unit tests now that bbackupd honours the hard limit and not theChris Wilson2007-07-26
| | | | | soft limit. (refs #3, merges [1662])
* Create logger object in the right place to avoid accidental triggering.Chris Wilson2007-07-26
|
* Make test/bbackupd intercept timer tests a little less strict, becauseChris Wilson2007-06-30
| | | | | | object IDs can vary depending on the order that readdir() returns files in. (refs #3)
* Gratuitous additional spaces (refs #3)Chris Wilson2007-05-09
|
* Slightly more progress information during tests (refs #3)Chris Wilson2007-05-03
|
* Read-only restore test fix (refs #3)Chris Wilson2007-05-03
|
* Fix running common tests in release mode, by not doing things thatChris Wilson2007-05-02
| | | | | crash in release mode, but assert in debug mode. (refs #3)
* Gratuitous cosmetic reformatting, to make the test easier to read (refs #3)Chris Wilson2007-04-28
|
* Call perl executable to execute external scripts, since shebang doesn'tChris Wilson2007-04-28
| | | | | work on win32. (refs #3)
* Remove misplaced hunk. (refs #3)Chris Wilson2007-04-28
|
* Close updated files during continuous-update test, so that they can beChris Wilson2007-04-28
| | | | | backed up on Win32. (refs #3)
* Use helper function rather than calling bbackupctl directly. (refs #3)Chris Wilson2007-04-28
|
* Use quiet mode in compares, to reduce test output verbosity (refs #3)Chris Wilson2007-04-28
|
* Format messages within test to make them more obvious, and improveChris Wilson2007-04-28
| | | | | readability of test output. (refs #3)
* Test that bbackupd and bbstored are still running at various points duringChris Wilson2007-04-28
| | | | | the test, and abort early if not. (refs #3)