[GitHub] [accumulo] ctubbsii commented on issue #1222: Determine why metric reported 4 quadrillion ms

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

ctubbsii commented on issue #1222: Determine why metric reported 4 quadrillion ms
URL: https://github.com/apache/accumulo/issues/1222#issuecomment-511999573
 
 
   The metrics code casts the size of the `updates.entrySet()` (which is an `int`) to a `double`, so it can be the divisor for the total time taken in millis (which is a `long`), presumably so it can keep some precision in the division, before it casts to a long to return the result.
   
   I see two possibilities:
   
   1. The entry set was empty, and casting it to a `0` to a double resulted in something very large when placing this "very close to, but not quite zero" value in the denominator, or
   2. The system clock was updated and caused a bigger than expected time difference in millis when using `System.currentTimeMillis()`.
   
   I think option 1 is more likely. We should probably guard against the set being empty before adding a metrics.
   
   There are two methods that do this: https://github.com/apache/accumulo/blob/ebca8e13dd12fc866887ebbd773ceeacb69a2f3e/server/tserver/src/main/java/org/apache/accumulo/tserver/TabletServer.java#L1156-L1162

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


With regards,
Apache Git Services