hlgp commented on issue #1837:
URL: https://github.com/apache/accumulo/issues/1837#issuecomment-744631533
> Based on the description, it seems that this issue is about how
table-specific properties can override system-properties. However that is the
design of per-table properties and works as expected.
No, that is not what the issue is saying. I tried to clarify it in the
discussion but it seems further clarity is needed.
Given a set of Tables T{T1, T2, ..., TN} and a set of configs C{C1,C2, ...,
CN}, with some number of system-level properties S with C1, C2, ... CN set,
for a given table T1 with C1,..., CN set, T1C1 can and does take precedence
over SC1. This is good, right, and proper. The issue is (was) that there
exists a set of properties with tablenames which can be set at both the system
level and table level for each table, C.Tname. When I say "Individual
table-specific properties are able to be overridden in *different* table
contexts" I mean C.T1 could be overridden in T2 ( *different* tables meaning
T1 != T2 ) in such a way that T1C.T1 and SC.T1 are ignored in favor of T2C.T1.
C.T1 should be specific to T1 and no table T not equal to T1 should be able to
override it. It sounds like #1201 addressed this so that SC.T1 would take
precedence over all TC.T1 which fixes the specific issue with the
HostRegexBalancer, but that violates the original idea that Tc should override
Sc for
all c in C.
----------------------------------------------------------------
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]
