| Change | User | Description | Committed |
|---|---|---|---|
| 19846 | swellard | Check exit status during setup | 9 years ago |
| 19845 | ptomiak | Do not remove the user on cleanup | 9 years ago |
| 19805 | ptomiak | Change HWS service name on OS X, kick off another build. | 9 years ago |
| 19788 | ptomiak | Kick off another build - put the update back in | 9 years ago |
| 19785 | ptomiak | Kick off another build in EC - comment the apt-get update | 9 years ago |
| 19781 | ptomiak | Add an update step before installing git-fusion as per Dave's request. | 9 years ago |
| 19769 | ptomiak | Back out changelist 1404458 | 9 years ago |
| 19763 | ptomiak | Reconciled a WIP file - revert. | 9 years ago |
| 19760 | ptomiak | Rename Mac OS X service to com.perforce.HelixWebServices | 9 years ago |
| 19756 | ptomiak | Update configuration file messages in installers | 9 years ago |
| 19720 | swellard | Login changes to use basic auth | 9 years ago |
| 19704 | tjuricek | Fix assertions around checks for running/stopped services in the test. | 9 years ago |
| 19701 | tjuricek | Add "--installer" option to flag that the uninstall-helix-web-services app is running via... a package uninstaller. This should fix deadlock issues on RPM-based systems. « |
9 years ago |
| 19689 | tjuricek | Set the program name and version in automation to make it clear that the connection is not... "some old p4java version" and it's not the product. « | 9 years ago |
| 19686 | ptomiak | Run the uninstall java class on ubuntu and centos as well. | 9 years ago |
| 19677 | tjuricek | Do not follow symlinks when looking for things to clean up during deploymentCleanup. On... Linux, this left the /etc/init.d/helix-ws symlink, which caused problems during the next test of the binary archive testing. « |
9 years ago |
| 19664 | ptomiak | Add an installer for Windows, modify setup script name, modify deb and rpm package behavio...ur, add a Mac OS X dmg installer. « | 9 years ago |
| 19618 | tjuricek | Looks like two more instances of the renamed service command that are missing | 9 years ago |
| 19617 | tjuricek | On windows, sc.exe stop does not consistently wait for the process to finish, so we need t...o wait until the port becomes available. « | 9 years ago |
| 19613 | ptomiak | Package remove with purge should leave just the configuration files, anything else should... be removed. Rename the service to helix-ws. Remove uninstall-helix-web-services from packages. « |
9 years ago |
| 19610 | tjuricek | Issuing a stop before uninstalling, and just verifing ports are available. It's likely... that our upgrade tests will fail since we're not managing p4d. « |
9 years ago |
| 19609 | tjuricek | Capturing all exceptions when just checking for connections to a server, and returning fal...se, since, well, we had a problem connecting to the server. On OS X, using 'unload/load' instead of 'stop/start'. The documentation isn't clear on why this is the way to do it but it works in practice. « |
9 years ago |
| 19607 | tjuricek | Add status check to log errors and flag ERROR if any p4ds are misconfigured. Run this che...ck on startup. Note: I tried shutting down in the face of p4d configuration error - there be dragons down that route. It's very, very tricky to get the rules right for automation workflows, so I'm leaving that be at the moment and just making sure we have logging and a status reporting system. In theory, admins should be using some kind of monitor that says "uh oh error" and then they go check the logs and fix things. « |
9 years ago |
| 19552 | drobins | Align references to WebApp with the actual class name of HelixWebServices | 9 years ago |
| 19535 | drobins | Refactor package names to hws | 9 years ago |
| 19445 | tjuricek | Fix the install4j-generated launcher (was using old WebApp class name), and add diagnostic... messages for issues with DeployBinaryArchive. « | 9 years ago |
| 19394 | tjuricek | Add %pre section to RPMs (for diagnostics), add reinstall test suite. The @BeforeTest /... @AfterTest annotations do not work like JUnit - they are associated to a much larger "test" abstraction, which tends to be even higher than the suite level. It's one of those oddities of TestNG. We could use @Test with dependencies, I prefer just putting the cleanup calls directly in the test currently. « |
9 years ago |
| 19387 | tjuricek | Use the %config directive on .rpm files, allow the deb deployment task to find the prerele...ase archive on Ubuntu 14, wire up RPM tasks. « | 9 years ago |
| 19383 | tjuricek | Fixes to the "unpack the prerelease archive file" along with preliminary work on the RPM u...pgrade test suite. « | 9 years ago |
| 19380 | tjuricek | Adjust post-install config to handle upgrades better, and a test for debian systems. -... Our default config file has an invalid line to flag that the file has never been touched. It's removed on the first post-install config. - We guard against overwriting things the next times post-install config is run, like the keystore files. - There's a "media" setting in the install4j configuration to treat "never override" as config files, which avoid overwriting the config (finally found it), at least on debian. - Added a new test suite just to ensure the configuration file is not overwritten, only works for .debs at the moment « |
9 years ago |
| 19362 | tjuricek | Initial approach to running an upgrade, which basically gets past one stupid issue where t...he trust files already exist. This is likely not going to be how we exactly test. We'll probably want to run some before and after verifications, for example, to modify configuration and ensure it isn't touched during upgrades. But it's a starting point for further work. « |
9 years ago |
| 19360 | tjuricek | Convert to using r16.1 builds out of the main depot instead of the p16.1 line out of the r...emote builds depot. « | 9 years ago |
| 19338 | tjuricek | Create custom reporting mechanism to allow for clear "multiple suite" reports... with log...ging! This standardizes on log4j 2 as the backend (was just used on the server side). We have an "in memory appender" that we use to capture results during testing, which we associate with each method run. The whole system spits out yaml files for each suite run, which are then read in via a final task and a single html report is spit out with ... everything... in it. « |
9 years ago |
| 19282 | tjuricek |
Use the "trust fingerprints" not the trust store for the ... trusted fingerprints. |
9 years ago |
| 19281 | tjuricek | Adding non-unicode version of the default 16.1 p4d run. | 9 years ago |
| 19277 | tjuricek | Add self-signed SSL configuration variation (just with the 16.1 p4d) for testing. No bu...gs were actually uncovered, though it may be useful to debug environments. (If people don't have a properly patched JVM with the JCE extensions, this test will fail.) « |
9 years ago |
| 19276 | tjuricek | Escape all file lines to handle spaces in paths. Looks like some new files had spaces in... them. « |
9 years ago |
| 19235 | tjuricek | The server spec crud test didn't run as super (hence a failure) and we needed to wait for... buffered reading to finish before moving ahead with background processes. « | 9 years ago |
| 19229 | tjuricek | Use a direct call to String#equals instead of List.contains | 9 years ago |
| 19220 | tjuricek | Allow HWS shutdown to retry using SIGKILL on non-Windows machines. and try to move away f...rom System.exit to "fail all tests" mode in the face of system configuration issues. « |
9 years ago |
| 19205 | tjuricek | Basic infrastructure for PHP testing, and, removing the ps check after startup - seems to... be too fast, though the server should be already responding to p4 info commands. « | 9 years ago |
| 19204 | tjuricek | Add StreamReaderThread concept for process launching, and using that by default to interac...t with background processes. « | 9 years ago |
| 19191 | tjuricek | Adding a log message if a pid file exists but the process isn't running. If there was a...n invalid pid file, the cleanup process would sometimes halt because we'd continue to look for the process. « |
9 years ago |
| 19182 | tjuricek | Allow for stopP4d to be called even if it's not running, but try to fail if we've started... p4d that we can't track. « | 9 years ago |
| 19181 | tjuricek | We may be halting too quickly during the unicode upgrade, and the PID file may not be writ...ten. Try to get to a "p4 info" response before saying things are completely started. « |
9 years ago |
| 19180 | tjuricek | Add logging and exception case where p4d doesn't create it's PID file correctly. | 9 years ago |
| 19177 | tjuricek | Reorganizing how HWS and p4d are logged during tests, and adding other messages to try to... help identify what might be going wrong with p4d restarts in the CD pipeline « | 9 years ago |
| 19175 | tjuricek | Extend timeouts waiting for port to become available. | 9 years ago |
| 19174 | tjuricek | JaCoCo code coverage setup and test cleanup revisions. During test cleanup we remove th...e directory the code coverage data is written to, so we need refined control on where the output file goes (which is copied to a build directory, currently). « |
9 years ago |
| 19170 | tjuricek | Try SIGTERM, wait a bit, then go with SIGKILL if the process is still running. This may... alleviate issues with OS X regularly having p4d startup failures after upgrading to unicode. « |
9 years ago |