adoroszlai opened a new pull request #1013:
URL: https://github.com/apache/hadoop-ozone/pull/1013


   ## What changes were proposed in this pull request?
   
   1. Skip the check for existence of chunks directory that happened on each 
`getChunkFile` call.  Read and delete operations already verify existence of 
the chunk file itself.
   2. Instead of the above check, verify the existence of the block file after 
write once per block.  This is needed to allow marking the container unhealthy 
immediately if file is lost during write, see existing unit test in 
`TestContainerStateMachine`.
   3. Perform `finishWriteChunk` in batch, avoiding unnecessary conversion of 
all chunks, operate on already converted block data.
   4. Replace "absolute path" with "path" as key in `OpenFiles` (for 
file-per-block).  This is safe because chunk dir location is constructed 
relative to the container dir (and eventually relative to HDDS volume), so the 
result is the same.
   
   https://issues.apache.org/jira/browse/HDDS-3703
   
   ## How was this patch tested?
   
   Tested locally on docker-compose cluster.
   
   https://github.com/adoroszlai/hadoop-ozone/runs/733420600


----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
[email protected]



---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to