Open bmildren opened 4 years ago
Hi @bmildren, This is a duplicate of #2661 , fixed already in 2.1
@JavierJF : can you confirm please?
ahh cool stuff - I did search before I created this one, clearly not very well though! 😄 Do we know roughly when 2.1 will be released? (or whether the fix will be backported to 2.0?) - just looking to plan around when/how to push out a fix for this 🙂 👍
ProxySQL 2.10 / Ubuntu 18.04
Description:
Any query that includes a value that uses scientific notation for a double data type (or equivalent) isn't abstracted by the regex used to canonicolise the query and will therefore likely be treated as a unique fingerprint (given the supplied value ofc) leading to a bloated query_digest.
How to repeat:
A query with an insert containing a decimal value like this:
insert into test_table(col1, col2, col3, col4, col5) values ('1', 1, 1, 1, 0.0) on duplicate key update col5 = values(col5);
produces this output in the stats_mysql_query_digest table:
insert into test_table(col1, col2, col3, col4, col5) values (?, ?, ?, ?, ?.?) on duplicate key update value = values(value)
A query with an insert containing scientific notation like this:
insert into test_table(col1, col2, col3, col4, col5) values ('1', 1, 1, 1, 1.1234567e+8) on duplicate key update col5 = values(col5);
produces this output in the stats_mysql_query_digest table:
insert into test_table(col1, col2, col3, col4, col5) values (?, ?, ?, ?, ?.1234567e + ?) on duplicate key update value = values(value)