Open sileht opened 1 year ago
btw, the same problem happens with MySQL
Hi @sileht 👋 FYI, I fixed this in our backend so it should be safe to upgrade to 7.48 again for the postgres integration. This was the result of a bug introduced in 7.48, which will be fixed in the agent in the 7.50 release of the postgres integration. Sorry for the inconvenience, and thank you for reporting the bug to us!
@pkatiushyn for MySQL, what tag do you see being duplicated? Can you provide more details?
The problem with mysql check appeared also after 7.48. Here is the check config:
---
instances:
- host: d01.xxxxxxx.us-west-2.rds.amazonaws.com
username: dbuser
password: dbpass
tags:
- dbclusteridentifier:cluster1
- dbinstanceidentifier:d01
options:
replication: 1
extra_status_metrics: 1
extra_innodb_metrics: 1
- host: d00.xxxxxxx.us-west-2.rds.amazonaws.com
username: dbuser
password: dbpass
tags:
- dbclusteridentifier:cluster1
- dbinstanceidentifier:d00
options:
replication: 1
extra_status_metrics: 1
extra_innodb_metrics: 1
init_config: {}
logs: []
Then looking at mysql.performance.queries
metric and grouping by dbinstanceidentifier
, I see the following groups:
dbinstanceidentifier in avg:mysql.performance.queries{dbclusteridentifier:cluster1}
d00
d00,d01
d01
This weird d00,d01
in dbinstanceidenfier
is the wrong one.
Additional environment details (Operating System, Cloud provider, etc):
Heroku
Steps to reproduce the issue:
Our Postgres server has two databases,
db_one
anddb_two
.Here is our Postgres configuration file, to monitor
db_one
:Describe the results you received:
Since 7.48.0, our
xxx.db.users.count
is alternatively tagged withdb: db_one
, thendb: db_two,db_one
, ...Fallback to 7.47.1 solves the issue.
Describe the results you expected:
Since this is a custom_queries and
dbname:
is set we expected all our db tags of all custom_queries to be onlydb: db_one
Additional information you deem important (e.g. issue happens only occasionally):
The coma inside the tag looks suspicious.