[GitHub] [pulsar] gfredericks commented on issue #6847: Reader skips over large block of messages after calling `seek`

[GitBox <gi...@apache.org>]

gfredericks commented on issue #6847:
URL: https://github.com/apache/pulsar/issues/6847#issuecomment-625189109


   The backlog quota is set to `consumer_backlog_eviction`, so I believe this is the explanation for what I'm seeing; it's possible that the resolution is just "remove the backlog quota", but I don't yet understand the implications of this to do it confidently.
   
   Specifically:
   
   If we intend to keep a very long history for the topic, is there any reason to have a backlog quota? I think we originally set it because we were worried about slow subscribers using a lot of resources, but do subscribers actually use much additional resources when there is already a long retention policy? Ideally a subscriber just consists of a few pointers into the data and the list of unacked message ids and perhaps some amount of data caching, but not an amount of resource usage linear in the messages between it and the end of the topic.


----------------------------------------------------------------
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:
users@infra.apache.org