[ 
https://issues.apache.org/jira/browse/CASSANDRA-17062?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17792509#comment-17792509
 ] 

Stefan Miklosovic edited comment on CASSANDRA-17062 at 12/3/23 11:00 AM:
-------------------------------------------------------------------------

I could not resist and took a look into CASSANDRA-14572 already and what you do 
and we do here is clashing. I think that CASSANDRA-14572 is taking more 
holistic approach and what we could do is that we could refactor the metrics 
(weighted and unweighted) into one coherent hierarchy / structure as it is done 
in this patch but it would be somewhat integrated / build on top of 
CASSANDRA-14572. I think that the refactorization of cache metrics (abstracting 
the common parts and extend what differs) is valuable in itself regardless of 
what 14572 is doing.

14572 takes a different approach to metrics - they are exposed as tables where 
each table groups some metric type (counter, histogram, gauge ...). While this 
probably "just works", the problem I see is that these metrics are not viewed 
all at once as this patch does it. For example, when it come to auth caches, I 
do not want to have them being split into tables and then, when I want to have 
overall visibility into that cache, I would basically need to query multiple 
tables. I prefer to do one query and instantly know what all metrics for 
metrics of auth caches are. Maybe there are similar "domains" where instant 
overview would be welcome instead of going to individual tables. I think that 
there is a lot of situations when an operator just want to seem them all all at 
once for some logical component.

EDIT:

on the other hand, I see there are also tables "system_views.metrics_cql" and 
"system_views.metrics_storage" so it is possible to just group one domain into 
one logical table. I can imagine that metrics in this ticket would be exposed 
via "system_views.metrics_auth_caches" instead of them being split into 
multiple vtables.


was (Author: smiklosovic):
I could not resist and took a look into CASSANDRA-14572 already and what you do 
and we do here is clashing. I think that CASSANDRA-14572 is taking more 
holistic approach and what we could do is that we could refactor the metrics 
(weighted and unweighted) into one coherent hierarchy / structure as it is done 
in this patch but it would be somewhat integrated / build on top of 
CASSANDRA-14572. I think that the refactorization of cache metrics (abstracting 
the common parts and extend what differs) is valuable in itself regardless of 
what 14572 is doing.

14572 takes a different approach to metrics - they are exposed as tables where 
each table groups some metric type (counter, histogram, gauge ...). While this 
probably "just works", the problem I see is that these metrics are not viewed 
all at once as this patch does it. For example, when it come to auth caches, I 
do not want to have them being split into tables and then, when I want to have 
overall visibility into that cache, I would basically need to query multiple 
tables. I prefer to do one query and instantly know what all metrics for 
metrics of auth caches are. Maybe there are similar "domains" where instant 
overview would be welcome instead of going to individual tables. I think that 
there is a lot of situations when an operator just want to seem them all all at 
once for some logical component.

> Expose Auth Caches metrics to virtual table
> -------------------------------------------
>
>                 Key: CASSANDRA-17062
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-17062
>             Project: Cassandra
>          Issue Type: Improvement
>          Components: Feature/Virtual Tables, Observability/Metrics, 
> Tool/nodetool
>            Reporter: Aleksei Zotov
>            Assignee: Stefan Miklosovic
>            Priority: Normal
>             Fix For: 5.x
>
>          Time Spent: 10m
>  Remaining Estimate: 0h
>
> Unlike to other caches (row, key, counter), Auth Caches lack some monitoring 
> capabilities. Here are a few particular changes to get this inequity fixed:
>  # Add auth caches to _system_views.caches_ VT
>  # Expose auth caches metrics via JMX
>  # Add auth caches details to _nodetool info_
>  



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to