errose28 commented on PR #5061:
URL: https://github.com/apache/ozone/pull/5061#issuecomment-1638985814

   > The reason I made it a OM call because I needed current state of DAG. You 
can argue that we can get that from compaction log but I had slight doubt that 
it won't give you exact picture as in memory.
   
   I'm not currently familiar with the snapshot implementation so this is up to 
you. As a debug tool it might actually be better to return the state from 
memory like you suggested since that is what the OM is actually using.
   
   > I think most of the concerns can be resolved if I change OM request to 
return the DAG as java object and construct the image on client side or use 
compaction log instead of the way it was done.
   
   Yeah some way to pass the result as either a java/proto object or a stream 
of bytes that the client writes as an image file sounds good, assuming the 
sizes of either response will not be prohibitively large. We may need to check 
which approach is easier to implement and generates the smallest size response. 
Another option is to return the bytes as an svg instead of png. png is not 
compressed, jpg would be lossy, but svg should give all the info in the most 
compact form factor. Gzipping as 
[svgz](https://docs.fileformat.com/image/svgz/) before sending and then 
unpacking client side could help reduce size on the wire even more.
   
   > Should we revert this PR in meanwhile I make the change to address your 
concern? Or just create new one with fix?
   
   If the fix can be completed in a few days, let's just create a new one with 
the fix. I don't think we should cut a release with these changes as is, but I 
don't think that is happening from apache master too soon.
   
   


-- 
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.

To unsubscribe, e-mail: [email protected]

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