pmm load and count graphs not working as expected

Percona Monitoring and Management (PMM) PMM 1.x
1

Hi, we have setup pmm queries analytics in order to track down and fix slow queries on our app but it seem that the load, count and latency graphs are not working as expected.

We getting the data from the mysql slow-queries.log and can confirm that the queries in the slow-queries.log are happening at various times during the day but the graph are showing the all the queries happened just at certain point in time.

What could be causing this issue?

Your help will be truly appreciated.

photoid=51572
photoid=51572794×476 53.2 KB

2

Hi Guys, I’m still experiencing this issue.

3

Hello webbox.eric welcome to the forum and thanks for your question. I wonder if I could get you to run through the information in this blog post about troubleshooting PMM :
[LIST]
[][URL=“Troubleshooting Percona Monitoring and Management (PMM) Metrics - Percona Database Performance Blog”]https://www.percona.com/blog/2018/01...t-pmm-metrics/[/URL]
[/LIST] Also could you just run through this too, to make sure everything is set up as expected:
[LIST]
[][URL=“Percona Monitoring and Management”]https://www.percona.com/doc/percona-...onf-mysql.html[/URL]
[/LIST] If you are still struggling could you provide this information:
[LIST]
[]versions of all software (PMM, database)
[]details of environment (versions etc)
[]if there are any error logs being recorded
[]configuration file for MySQL
[/LIST] Once we have that I will alert the PMM team to your question to see if they can help you through.
THANKS!

4

Hi Lorraine.pocklington,

Thanks for getting back to me on this.

I’ve gone through the suggested post and the metrics seems to work as expected. The issue is not much with the metrics but only with the Query Analytics.

We have a remote server(freebsd 10) where pmm client cannot be installed so we have setup the pmm client and server on a local ubuntu server.

The idea is to copy the slow-queries.log from our production server to this local setup in other to analyse it and fix slow queries.

The my.cnf of the mysql server on the remote freebsd machine set the slow-queries.log to /var/db/mysql/slow-queries.log so i have copied the file to the local machine in the same location and set correct file permissions.

The file is parsed anytime it is updated: [TABLE]
[TR]
[TD]2018-05-31T13:11:00.029555814Z[/TD]
[TD]qan-analyzer-mysql-6788256a-worker[/TD]
[TD]info[/TD]
[TD]Parsed /var/db/mysql/slow-queries.log 22100859-22201653: 100.0% 22201401/22201653 100794 0.0s[/TD]
[/TR]
[/TABLE]

but the graph shows show a few picks where to which it they attribute all the queries and show zero slow query elsewhere. So the issue is the distribution of queries in time.

Please give me your feedback on this

photoid=51607
photoid=516071182×168 19.6 KB

photoid=51608

photoid=51609
photoid=516091872×759 79 KB

photoid=51610
photoid=516101319×544 277 KB

photoid=51611
photoid=516111864×713 243 KB

5

Hi Lorraine,

we’re still having this issue and have attached more details about our setup as you’ve requested.

Please alert the PMM team about this question as you promised to do.

6

Sorry for the delay.
Unfortunately what you are trying to do isn’t going to work.
qan-agent reads data as it shows up in the slow log file, so for this reason you can’t just copy it from other server.
If you aren’t able to install pmm-client on that system then your best option would be to use performance schema and connect the pmm-client remotely.
There’s information here on configuring performance schema [url]Percona Monitoring and Management