[GitHub] [trafficserver] zwoop edited a comment on issue #6713: Cache read failures seen in the metrics 7.1.x and 8.0.7

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

zwoop edited a comment on issue #6713:
URL: https://github.com/apache/trafficserver/issues/6713#issuecomment-620787903


   Hmmm,. we're not seeing this, it's the complete inverse, about 3% read.failure's, which falls reasonably inline with our amount of cache misses. Looking at some code, and discussing on the Slack channel, there's definitely some scenarios where the above is as expected. tldr; we're pretty sure that a normal cache miss shows up as a proxy.process.cache.read.failure as well.
   
   For example, imagine a box with lots of RAM, and very (very!) high RAM cache hit ratio. As the code looks, a cache miss (on disk) shows up as a read.failure, but since RAM cache is so effective, there's very little read.success (since it's almost all from RAM cache). Therefore, the read.failures from cache misses could dwarf everything else.
   
   I'm not saying this is the case here, but it's a reasonable explanation.


----------------------------------------------------------------
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