| Änderungsprotokolle |
|---|
Im folgenden sind die letzten Änderungen der verschiedenen Softwarepakete aufgeführt.
Letze 10 30 100 1000 10000 Einträge
| Kategorie | Datum | Version | Eintrag |
|---|---|---|---|
| cego | 11.01.2026 | 2.54.0 | Reduction of following classes done : CegoDistManager ( now CegoQueryManager ), CegoDistCursor ( now CegoQueryCursor ), CegoDbHander ( CegoDistDbHandler is obsolete ), just straight query handling is supported now, no more distributed queries Reduction of admin classes done : CegoAdminHandler, CegoAdminThread, CegoAdmAction and friends, all mediator services have been consolidated to native admin services Still some code cleanup needed, but already a huge amount of code has been removed. |
| cego | 10.01.2026 | 2.54.0 | Thoughts about redesign of cego mode started. The current design with mediator/primary/secondary with synchronous logging to secondary seems to be to complicated and fragile. A new approach might be to make secondary recovery based on complete archive logfiles, which are copied from primary to secondary asynchronously. The secondary host then must be controlled by himself ( no more mediator role ), so the complete database switch to secondary ist managed by an upper layer ( maybe not part of cego, since this also includes switch of a virtual host ) As a consequence, the cego mediator role will disappear ( just primary and secondary role are valid ) |
| cego | 09.01.2026 | 2.53.9 | Removed PIDFILE attribute from db xml. PIDFILE should be specified by database startup with pidfile option |
| cego | 03.01.2026 | 2.53.8 | Version released |
| cego | 03.01.2026 | 2.53.8 | Small patch added to CegoAdminThread::getBackupInfo. Also empty logTS token are accepted now, since there might be still no archive logs avaiable for the backup In tools/backupManager, LOGTS was set to None, if no archive logs are available |
| cego | 01.01.2026 | 2.53.7 | Version released |
| cego | 01.01.2026 | 2.53.7 | Cleanup and reduction of logging in CegoRecoveryManager |
| cego | 31.12.2025 | 2.53.7 | Small improvement added to CegoAdmin to catch non existing batchfile exception |
| cego | 30.12.2025 | 2.53.6 | Version released |
| cego | 29.12.2025 | 2.53.6 | Added point in time recovery option to restore nut just pit ( timestamp ) but also to a specified lsn. This might be useful to restore exactly to a specific log entry ( timestamp is given in seconds, where multiple operations may occur ) |
| cego | 29.12.2025 | 2.53.6 | Added shiftactive mode for cego to rescue active online redo logs to archive destination. This is useful in case of online backup recovery after a crash, so before starting recovery, the current active online redo log can be explicit shifted. With the shift, the latest LSNs are catched and a full recovery can take place. Please note : This option is just useful, if there are uncorrupted online redo logfiles available after the crash. |
| cego | 28.12.2025 | 2.53.6 | Small improvements added to backupManager tool to evaluate next backup id ( now via shell function ) |
| cego | 26.12.2025 | 2.53.5 | Version released |
| cego | 26.12.2025 | 2.53.5 | Crash consistency checks ( checkCrashAndRecovery with 20 runs and checkRandonCrash with 200 runs ) passed |
| cego | 26.12.2025 | 2.53.5 | Patch added in CegoDistManager::startTableSet, final checkpoint must be written after starting log, otherwise no logswitch will be done |
| cego | 25.12.2025 | 2.53.5 | Basic backup recovery with heavy load, concurrent online backup and recovery up to last committed lsn passed ! |
| cego | 25.12.2025 | 2.53.5 | In CegoRecoveryManager, changed order in recovery procedure. Datafile recovery must be done after file registration, but before object registration. File recovery restores bitmaps and pages on file level and object registration loads pages into bufferpool. So we first have to update data files ( via dataFileRecovery ) and then these pages are loaded to bufferpool ( via registerObjects ). |
| cego | 24.12.2025 | 2.53.5 | Basic tableset recovery from online backup works. To treat th e 0-LSN-Sync logic, several adaptions for CegoRecoveryManager were necessary. Still missing are extensive crash recovery tests |
| cego | 24.12.2025 | 2.53.5 | Set LSN for LOGREC_SYNC operations to 0.
This is useful to simplify the recovery process especially for backup recoveries. With previous logic, LSN gaps occur in the archive log history file ( caused by LSN handling in CegoTableManger::insertDataTable with previous allocated LSN ) which result in incomplete recoveries. The change to LOGREC_SYNC=0 had some impacts for logfile and recovery handling done in CegoLogThreadPool::copyLog, CegoDistManager::startTableSet and CegoLogManager::logAction The new logic also avoids unnecessary archive logs with only LOGREC_SYNC operations, since these files are skipped by copyLog now. This new logic still has to be verified, especially in combination with online backups und subsequent recoveries |
| cego | 23.12.2025 | 2.53.5 | Added patch in CegoLogManager::logAction, logfile should be flushed if LOG_FULL occurs, otherwise it seems, that log entries get lost with subsequent log file switch ( still not verified ) |