[users] Possible component and proxied component restart issues

[Brian C. <bri...@er...>]

Hi,

I am seeing some behavior with OpenSAF version 4.3.0.1 that doesn't seem 
right:

I have a proxy and and a proxied component. These are different scenarios:

1. If I kill the proxy, and cause a fault on the proxied component the 
proxy is restarted, and there is a lot of messaging to get it going 
again. During that time the proxied component reports an error and the 
proxied component cleanup is called. If I don't stop health checking in 
that cleanup the proxied component will not be re-instantiated by AMF.  
Even if I stop health checks I still see one more health check after the 
clean calls the AMFresponse saying it is done, but the proxied component 
is instantiated OK.

2. If I cause a fault on the proxied component such that it issues an 
error report requesting a restart, if health checking is stopped in the 
cleanup callback it will fail with -12 (SA_AIS_ERR_NOT_EXIST).  I don't 
think that should fail.

3. If I kill the proxy and cause two faults in the proxied component the 
proxy will get two error reports when it comes up. I have consistently 
seen that if the second error report is sent to AMF while the  proxied 
component's cleanup callback is already running (on a different thread), 
then the proxied component will not be instantiated. The proxied cleanup 
callback is not called again either. The state in AMF seems to be broken 
by this situation. Getting rid of the second error report makes this 
problem go away.

I think these need to be fixed. Please advise.

thanks,
Brian Corriveau
Ericsson Vancouver