Hello,
Will be great to have an opinion on the following re-connection flow,
   
https://github.com/kashirin-alex/hypertable/blob/8aacd6689517c4fc29f49a2845584e8cc192f7ce/src/cc/FsBroker/Lib/Client.cc#L124

the bool Client::re_connect method is called on comm exception for FsClient 
methods that, at this time, have file/dir name param.

RangeServer has goto retry and CellStore has the reopen_fd method 
(m_filesys->open(m_filename, 0);), 
  while the fs-client will not try to re-connect to the FsBroker.

It will be Failure Tolerance improvement if the FsClient will do 
re-connects at least with the method that do not involve fd .
-- In case, to apply re-connect to methods working with fd , it will 
require fd to map to file-name and keep track for offsets.

Cases, such as FsBroker restart and RangeServer fail of FsBroker 
communications might completely be fixed.

Thank You,
Kashirin Alex


-- 
You received this message because you are subscribed to the Google Groups 
"Hypertable Development" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
Visit this group at https://groups.google.com/group/hypertable-dev.
For more options, visit https://groups.google.com/d/optout.

Reply via email to