[bind10-dev] b10-resolver segfaults on git server

[Stephen Morris]

On 13/09/2011 13:18, Jeremy C. Reed wrote:
> The first segfault above happened at:
> 
> 2011-08-18 23:57:45.396 DEBUG [b10-cmdctl.cmdctl]
> CMDCTL_COMMAND_SENT command 'shutdown' to module 'Resolver' was
> sent
> 
> 2011-08-18 23:57:45.406 DEBUG [b10-resolver.cc] CC_START_READ
> starting asynchronous read
> 
> 2011-08-18 23:57:45.424 INFO  [b10-resolver.resolver]
> RESOLVER_SHUTDOWN resolver shutdown complete
> 
> [bind10] Process b10-resolver (PID 25228) terminated, exit status =
> 11

This makes me wonder whether the shutdown is cancelling outstanding
asynchronous reads.  If not, could the read be completing after the
shutdown has deleted some object referenced by the read completion code?

The same comment applies to the second segfault:

> The second segfault happened when I sent shutdown to module Boss.
> 
> 2011-08-18 23:58:45.317 DEBUG [b10-resolver.cc] CC_GROUP_RECEIVE
> trying to receive a message
> 
> 2011-08-18 23:58:45.317 DEBUG [b10-resolver.cc] CC_GROUP_RECEIVED 
> message arrived ('{ "from": "4e132f44_1 at bind10.isc.org", "group": 
> "Resolver", "instance": "Resolver", "seq": 7, "to": "*", "type":
> "send" }', '{ "command": [ "shutdown" ] }')
> 
> 2011-08-18 23:58:45.318 DEBUG [b10-resolver.cc] CC_REPLY replying
> to message from '{ "from": "4e132f44_1 at bind10.isc.org", "group": 
> "Resolver", "instance": "Resolver", "seq": 7, "to": "*", "type":
> "send" }' with '{ "result": [ 0 ] }'
> 
> 2011-08-18 23:58:45.318 DEBUG [b10-resolver.cc] CC_START_READ
> starting asynchronous read
> 
> 2011-08-18 23:58:45.319 DEBUG [b10-resolver.cache]
> CACHE_MESSAGES_DEINIT deinitialized message cache
> 
> 2011-08-18 23:58:45.322 INFO [b10-resolver.resolver]
> RESOLVER_SHUTDOWN resolver shutdown complete

Stephen