Wrong connection used in function get_custom_query_ids
GoogleCodeExporter opened this issue · comments
What steps will reproduce the problem?
1. Use a different monitoring host then write host
2. Run mycheckpoint <parameters> -v
3. Mysql-problem: (1146, "Table 'mycheckpoint.custom_query_view' doesn't exist")
What is the expected output? Full creation of 58 tables in database
mycheckpoint_HOST
What do you see instead? Only 5 tables are created, python script stops after
that one.
What version of the mycheckpoint are you using? revision 174
What version of the MySQL are you using? 5.1.37
On what operating system? SLES 11
I think the problem is the use of the wrong connection-parameter in the
get_rows call (no connection is defined so it uses the monitor-connection. For
this query, it should use the write_conn instead). Problem was fixed when
replacing:
query = """SELECT custom_query_id, chart_name FROM %s.custom_query_view""" %
database_name
rows = get_rows(query)
with:
query = """SELECT custom_query_id, chart_name FROM %s.custom_query_view""" %
database_name
rows = get_rows(query,write_conn)
Also, I think it is strange that no parameters can be defined for the user and
password on the monitored mysql database. Now it needs to be the same one of
both servers...
Original issue reported on code.google.com by barty.de...@gmail.com
on 13 Aug 2010 at 9:53
Thanks for the report. Will look more closely a couple of days ahead.
The issue of monitored user-password is not new; I'll consider adding this as
option.
Original comment by shlomi.n...@gmail.com
on 14 Aug 2010 at 1:03
Your diagnosis is correct. The problem with using the read-connection is that:
1. If there is master-slave relationship between write host and monitored host
- the slave may not get the view created so fast
2. If there isn;t such relationship... well, the monitored host just won't get
it...
I'll add a fix in the next revision.
Original comment by shlomi.n...@gmail.com
on 16 Aug 2010 at 7:13
- Changed state: Accepted
Fixed in r188 (not yet released)
Expect a release in the next couple of weeks.
Original comment by shlomi.n...@gmail.com
on 18 Aug 2010 at 7:42
- Changed state: Fixed
[deleted comment]
With regards to:
"...Also, I think it is strange that no parameters can be defined for the user
and password on the monitored mysql database. Now it needs to be the same one
of both servers..."
I have created Issue 15: a feature request for adding these parameters. Work in
on the way.
Original comment by shlomi.n...@gmail.com
on 22 Aug 2010 at 6:15