Overview
IDOL index is relatively vulnerable to unexpected changes in IDOL environment. If some of the IDOL processes are killed during its operation, the index data might become corrupted. This is further emphasized in the cluster environment, where multiple separate instances, even servers are working together.To avoid the corruption of the index and possible re-indexing, system administrator has to have full control to the IDOL server over the automatic updates and other activity that might trigger the server restart.
This document gives some considerations about common situations that might cause an uncontrolled shutdown of IDOL processes. Most importantly, it describes how to shut down and restart the IDOL in a controlled manner.
Since we cannot be 100% sure that the unexpected shutdown will not happen e.g. in case of power failure the best way to avoid the re-indexing is to take backups from the index on regular basis.
Solution
Situations where the uncontrolled IDOL shutdown may occur
- Windows automatic updates. These should be handled in a way that IDOL processes are down when starting the updates. Alternatively, all updates should be installed manually so the system administrator can control the shutdown and restart processes.- Other software installations and updates.
- Severe system malfunction. These cannot be predicted easily, so use disk RAID configurations and IDOL backups to save the day when the disaster strikes.
- User ignorance, since nowadays systems can be restarted quite freely without the fear of losing data.
Steps to shut down and restart IDOL cluster safely
1. In M-Files server stop M-Files Indexing Services (MFIndexingManager.exe).2. In IDOL frontend server stop DIH/DAH server service and Daily server service e.g. DihDah19IDOLServer and Daily9IDOLServer.
-> WAIT until all corresponding processes have been shut down. Do not kill them or the index might get corrupted. The process names are:
- DihDahXXIDOLServer.exe (or equivalent via configuration)
- DailyXIDOLServer.exe (or equivalent via configuration)
- dih.exe
- dah.exe
- content.exe
3. After the frontend has been shut down, stop all backend indexing services e.g. Single10IDOLServer.
-> WAIT until all corresponding processes have been shut down. Do not kill them or the index might get corrupted. The process names are:
- SingleXXIDOLServer.exe (or equivalent via configuration)
- content.exe times the amount of engines in the corresponding server
4. After the backends have been shut down, start the cluster's services in reversed order: Backend, Frontend.
5. When the cluster is up and running (check application.log files that engines are up) you can start M-Files Indexing Services (MFIndexingManager.exe).
Steps to safely shutdown and restart IDOL single installation
1. In M-Files server stop M-Files Indexing Services (MFIndexingManager.exe).2. In IDOL server stop Single9IDOLServer service.
-> WAIT until all corresponding processes have been shut down. Do not kill them or the index might get corrupted. The process names are:
- Single9IDOLServer.exe (or equivalent via configuration)
- content.exe
3. After the processes have been shut down, start the Single9IDOLServer service.
4. When the IDOL is up and running (check application.log files that the engine is up and running) you can start M-Files Indexing Services (MFIndexingManager.exe).
