SpectroServer SSDB is not marked unclean upon SpectroSERVER crash
Article ID: 239778
Updated On:
Products
Issue/Introduction
We have noticed on Spectrum 10.4.2.2 and also in other releases such as Spectrum 21.2.8 that if the SpectroSERVER SSdb does not shutdown completely, there is no prompt to restore a previously good saved SSdb.
What is the correct procedure for restarting a SSdb after a SpectroSERVER crash?
Environment
Release : 21.2.6 and higher
Component : Spectrum Core / SpectroSERVER
Cause
SS crash can cause SSdb to become corrupt if it was not shutdown gracefully
Resolution
As per Engineering reviewing the SpectroSERVER code starting in 21.2.6 and higher
During the start up, we are checking whether SSDDB is already opened or not. If it is already opened we are validating few key parameters and if they are fine then we will proceed with loading the last used SSDB
SS crash may not always ask for good saved DB.
In case if the key values from SSDB are not set correctly then only we will ask to load previously saved good DB
We still check possible errors in SSDB before trying to open.
If you would like to restore the old behavior where the SSdb would always force to reload the SSdb you can do this by adding this parameter to the $SPECROOT/SS/.vnmrc file
ssdb_cache_size_kbytes=10000
After adding this line, a stop and restart of the SpectroSERVER is required to go into effect. Afterwards if the SS is not shutdown gracefully, you will need to always perform a manual SSdb reload from a previous saved SSdb file.
Additional Information
For versions prior to 21.2.6 please follow this KB article for restoring the SSdb.
Feedback
