Closed lukebakken closed 1 year ago
[~lbakken] Can you please coordinate with the engineering team if additional work needs to be scheduled to resolve this request?
_[posted via JIRA by Magnus Kessler]_
@DSomogyi @alexmoore - who can I contact to see if we should fix this?
@lukebakken We should talk to @JeetKunDoug tomorrow.
See customer ticket #13106
Reproduction steps:
riak stop
on one nodeRpbErrorResp
with theUnknown message code:
errorThis could be due to the fact that PB message handlers are de-registered from the ETS table used for lookup before all sockets are closed. What should happen is that, during shutdown, the API application cancels all in-progress Riak operations and returns an
RpbErrorResp
(with valid error code) indicating that that node is shutting down, then begins un-registering handler modules.