response time distribution and query cache

Asked by roberto spadim on 2013-08-15

http://www.percona.com/doc/percona-server/5.5/diagnostics/response_time_distribution.html

how query cache work here?
it will be placed at 0.01ms to 0.1ms range (for example) or the default query time (20 seconds for example)?

could we add a new column to this table?
query_cache ENUM('Y','N') not null default 'N'
in this case we know what kind of query was executed, the query or the query cache result...

thanks

Question information

Language:
English Edit question
Status:
Solved
For:
percona-tools Edit question
Assignee:
No assignee Edit question
Solved by:
roberto spadim
Solved:
2013-08-16
Last query:
2013-08-16
Last reply:
2013-08-16
roberto spadim (roberto-b5) said : #1

or maybe... since the today table is running...
add two more fields...

count_query_cache
total_query_cache

to know the 'no query cache' values we could execute
(count-count_query_cache)
and
(total-total_query_cache)

The time logged by response time distribution is taking the query cache into account. The time is the same as used for e.g. slow query log logging.

I am sceptical about adding new column query_cache ENUM('Y','N') not null default 'N'.

roberto spadim (roberto-b5) said : #3

what about the count_query_cache, total_query_cache?

We can consider patches but I can certainly tell that QC is not a priority for Percona Server in general.

roberto spadim (roberto-b5) said : #5

nice, well i will try to develop a patch for this (i don't know how to develop test patch, just source)
i think it's nice to filter what kind of query we are using, sometimes query cache "don't show the true" about query times

well when possible i post here again

Alexey Kopytov (akopytov) said : #6

Keep in mind that response time distribution has been dropped from Percona Server 5.6. So if the patch is merged to Percona Server 5.5, it will not be ported to 5.6 simply because there's no feature in that branch.

roberto spadim (roberto-b5) said : #7

hum, and why it was removed?

Alexey Kopytov (akopytov) said : #8

Because we didn't see much demand for that feature, so benefits of supporting that (poorly written) code were unclear.

roberto spadim (roberto-b5) said : #9

ok

Stewart Smith (stewart) said : #10

You should be able to use performance schema to get something similar too. In fact, I think this is one of the examples that has been posted to planet mysql

roberto spadim (roberto-b5) said : #11

hi!
yes, but performance schema is a gun to kill a mosquito eheheh, i prefer a gun to kill a gnu :P
i will check the source code and with time provide a patch, if anyone want, he/she can use it without problems :)
thanks guys

about planet mysql do you know what topic it was? maybe the url link?
thanks again :)