GWAVA 4 NETWARE 6.5, 6, 5.1 LINUX OES, SLES9.3, SLES10.x (this info is forth-coming)
Spam training has been compromised and need to reset training to default, and start from scratch.
As GWAVA 4 is fed samples of spam/ham for the purpose of training, it updates "dictionary" files which GWAVA uses for spam filtering.
Sometimes administrators may want to restore the engine training to the default dictionaries. This is performed by doing the following:
1. Bring GWAVA down. ("gwavadn" on NetWare, "rcgwavaman stop" on Linux)
2. Rename or delete the scanner_ID folders in /GWAVA4/SERVICES/ASENGINE/CONFIGS/..
3. Bring GWAVA up. ("gwavaup" on NetWare, "rcgwavaman start" on Linux)
4. GWAVA will create a new folder with the default dictionary files. You should also notice in the GWAVA 4 Management Console under the section heuristics that the "Automatic spam system is currently using rule summation scoring method" (Server/Scanner Management->ServerX->Manage Scanners->Scanner Configuration->Antispam->Heuristics)
This article was originally published in the GWAVA knowledgebase as article ID 46.