Hey, not anyone? Can't be...
Seam cooperates with the EJB3 container, but does not have absolute control. So when you redeploy a Seam application which contains EJB's, halt and restart to get things right.
This is not a bug but a normal restriction.
I understand. However I ran into this issue many times in production, both after redeploys and full restarts of JBoss. It feels strange to redeploy as many times as it needs to get it working (I need to redeploy the app to fix it, not
restartJBoss as this doesn't seem to help).
This behaviour means that one cannot rely on automated start scripts. After a reboot of the prod server the application might run or it might not. This restriction has been bugging me for almost 3 years when went into production.
JBoss is sometimes
too goodfor applications and tries to save everything for later restarts. I have problems with wrongly zipped wars. After redeploying the server tries to use earlier deployed stuff from the temp directory (don't ask me why..)
clean upeverything after redeployment with problems; meaning deleting the TMP, DATA and WORK directories. However this only works if everything was correctly persisted in the DB.
I have seen JBoss to pickup the older version from the tmp folder even after deploying a new war. The safest way would be to remove all files under tmp folder prior before re-starting the server.