Comments - Why are almost all ROWS_EXAMINED values 0 in events_statements_* tables (performance_schema) !?

9 years, 2 months ago Andras Fabian

Thank you for this information ... But somehow I still do not understand, why then I still do not see ROWS_EXAMINED for most (or on some servers ALL!) queries in the events_statements_summary_by_digest view? I mean, there i should have a summary of queries which have long been finished.

As an example, on a server I have for example an entry with a simple "SELECT ...." (well, I have hundred similar ones ... but this is just as an example) in events_statements_summary_by_digest which says which has a COUNT_STAR=1901867 (so, it was fired quite often) ... And the time period over which this data was collected spans about 10 days (FIRST_SEEN to LAST_SEEN). Now, I would expect, that such a query has seen "the end of the query" quite often ... and as such, it should have yielded a valid ROWS_EXAMINED value quite often (and thus contribute tu a meaningful summary). But instead, ROWS_EXAMINED=0!

And this is my main problem, that on some servers all ROWS_EXAMINED are equal 0 in the summary tables. Which makes it quite "useless" for any type of analysis (whereas I think, that this value could be one of the most interesting one ... if it worked as expected/hoped).

So, I can understand, that in a "snapshot" view (like PROCESSLIST or event_statemnts_current) the ROWS_EXAMINED might be empty ... but in the summary (where a lot of already finished stuff should be summed up), it does not makes much sense to me.

So, can you tell me, why even in all the summaries, the ROWS_EXAMINED is almost always 0 (even for many, simple "SELECT ..." queries, which should have yielded ROWS_EXAMINED quite often at the end of their execution)?

 
Content reproduced on this site is the property of its respective owners, and this content is not reviewed in advance by MariaDB. The views, information and opinions expressed by this content do not necessarily represent those of MariaDB or any other party.