EmergencyEMERGENCY? Get 24/7 Help Now!

Percona Monitoring and Management

 | April 18, 2016 |  Posted In: MySQL

PREVIOUS POST
NEXT POST

Percona Monitoring and ManagementPercona is excited to announce the launch of Percona Monitoring and Management Beta!

Percona Monitoring and Management (PMM) is a fully open source solution for both managing MySQL platform performance and tuning query performance. It allows DBAs and application developers to optimize the performance of the Database Layer. PMM is an on-premises solution that keeps all of your performance and query data inside the confines of your environment, with no requirement for any data to cross the internet.

Assembled from a supported package of “best of breed” open source tools such as Prometheus, Grafana and Percona’s Query Analytics, PMM delivers results right out of the box.

With PMM, anyone with database maintenance responsibilities can get more visibility for actionable enhancements, realize faster issue resolution times, increase performance through focused optimization, and better manage resources. More information allows you to concentrate efforts on the areas that yield the highest value, rather than hunting and pecking for speed.

PMM monitors and provides performance data for Oracle’s MySQL Community and Enterprise Editions as well as Percona Server for MySQL and MariaDB.

Download Percona Monitoring and Management now.

 

CPU and Load
CPU and Load
QAN top 10
Top 10 Queries
QAN create table
QAN Create Table
QAN per-query metrics
QAN per-query metrics
QAN table indexes
QAN table indexes
PREVIOUS POST
NEXT POST

20 Comments

  • I’ve tried it with Percona XtraDb Cluster and Percona Server. It seems to work pretty well!

    Only have 2 issues so far: 1) you can only monitor 1 mysql instance per client host, and 2) The Query Analytics > settings (gear icon) page always throws an error.

  • Hello Bob,

    I have a cluster with MariaDB and Galera, with 4 nodes and want to monitor it with PMM Server, but there are some gaps, I did not understand very well.

    Could talk, where can I find this information?

    Questions:

    It is true, I can monitor MariaDB and Galera with the PMM Server? I saw that PMM Server uses the my.cnf and MariaBD uses server.cnf.

    With the PMM Server installed via Docker on a dedicated server with Centos7, and PMM Client in one of the four nodes of MariaDB with Centos7, I can see the screen
    Initial PMM server, and I can see the quarys in Quarry Analitycs, but Grafana nothing appears in MySQL graphics.
    But the graphs of Prometheus appear and function properly.

    If I access the prometheus in the browser, I see all the “endpoints” “Down,” only the ‘endpoint’ of Prometheus “Up.”

    I’ll need to create a symbolic link server.cnf for my.cnf? so the PMM Server can manage Galera?
    Could you tell me the syntax that defne the server.cnf file as default file?

    Having installed PMM Server package on the monitoring server and the client package on a cluster node, it is enough?
    Or you need to install some more component?
    I need to install the client on all nodes in the cluster or just one?

    Sorry so many questions but became interested in both the tool and am studying to put it into production.

    Thank you

  • Hello Bob, Sorry..ahah

    I have a cluster with MariaDB and Galera, with 4 of us and I thought of monitors it with PMM Server, but there are some gaps,
    I did not understand very well.

    Could talk, where can I find this information?

    Questions:

    It is true, I can monitor MariaDB and Galera with the PMM Server? I wonder why PMM Server uses the my.cnf and MariaBD uses server.cnf.

    With the PMM Server installed via Docker on a dedicated server with Centos7, PMM and Client in one of the 4 we bank with Centos7, I can see the screen
    Initial PMM, I view quarys in Quarry Analitycs, but Grafana nothing appears in MySQL graphics.
    But the graphs of Prometheus appear and function properly.

    If I access the prometheus in the browser, I see all the “endpoints” “Down,” only the ‘endpoint’ of Prometheus “Up.”

    I’ll need to create a symbolic link server.cnf for my.cnf? so the PMM Server can manage Galera?
    Could you tell me the syntax that defne the server.cnf file as default file?

    Having installed PMM Server package on the monitoring server and the client package on a cluster node, it is enough?
    Or you need to install some more component?
    I need to install the client on all nodes in the cluster or just one?

    Sorry so many questions but became interested in both the tool and am studying to put it into production.

    Thanks

  • Hi Victor –

    I’ll get answers to your questions put together and get them to you. Thank you for coming up with such great questions!

    Bob

    • Tanks Bob,

      I wait …

      If you want some help in testing and documentation, I can help.

      As said my idea is to place the PMM to monitor 4 nodes in production on which work and thus help develop the tool.

  • Hello Bob,

    I already have a nice draft documentation of all process of installation and configuration of Pmm-server and Pmm-Client.

    I took some screens from the mistakes and successes.

    But I still have problems with the metric in Prometheus.
    Installed with Docker, will not start the MySQL metrics, and trying to fix the problem, I do not find any of the directories that prometheus tells, not even the prometheus.yml file

    I think I’m close to getting deploy, but lacks only some detail yet at the same time looks far because not encounter the information.

    I am following trying and assembling the documentation … if you want a copy or prints, tell me how you send.

    Tanks

  • I installed the PMM docker yesterday. I am seeing gaps in the data – I am not sure what is causing those gaps. So far testing on some very lightly used databases.

  • I set up PMM and a all our our client servers started swapping. I took mysql_exporter off of all of them and they all now seem to be fine. Are there any other similar cases reported for PMM beta? Is there a suggested work around for mysql_exporter causing swap?

    • We usually see very low memory usage with mysqld_exporter, however if you’re having problems it may be that all of the options are enabled.

      A workaround to try is :
      – edit /usr/local/percona/pmm-client/exporters.yml and set -collect.info_schema.tables to false where it is true
      – service percona-prom-pm restart

      Does that help?

  • Hello Bob,

    We have freshly installed PMM and evaluating to use this for monitoring all our servers. However, the same issue with the gaps exist. Is there a workaround?

    Also, is there a way to put servers in different groups? I read Prometheus does support this. But, couldn’t find a way to do it in PMM.

    Thanks!

    • Hi Saranga!

      First off, with regard to the gaps in data, we’ve only been able to reproduce that with network latency. Are the servers on the same network?

      As for groups, I’m not sure – if it’s possible to do with Prometheus, I’m sure we can do it in PMM – we may not have built the user interface for it yet though.

      • I was also having gaps in my data – I set this up on the same server outside of the docker – just prometheus / grafana etc and it has been running for a month or so with no gaps in metrics.

        • Very interesting! I wonder if there’s something in the Docker container that is causing additional latency. It might be interesting to see if you get latency with a docker container that contains Prometheus/Grafana on it.

      • Thanks for the reply Bob, our servers are on different networks. So, may be the case.

        And, looking forward for the UI on grouping. It will be really helpful when we have hundreds of servers.

  • I’m using the docker container and only the Query Analytics works, the my Grafana graphs don’t get populated..

    Also the Docker container seems to become unresponsive after a period of time and needs to be manually restarted. I’ve tried doing a “docker attach ” but it just hangs. Doing a ctrl+c gives this

    “root@metrics:/opt/prometheus/data# docker attach f752b9d2d4a7

    ^C2016-08-17 12:58:57,885 WARN received SIGINT indicating exit request
    2016-08-17 12:58:57,885 INFO waiting for qan-api, consul, grafana, cron, nginx, prometheus, mysql to die
    2016-08-17 12:58:57,886 INFO stopped: prometheus (terminated by SIGTERM)
    2016-08-17 12:58:57,886 INFO stopped: cron (terminated by SIGTERM)
    2016-08-17 12:58:57,933 INFO stopped: nginx (exit status 0)
    2016-08-17 12:58:57,933 INFO stopped: qan-api (terminated by SIGTERM)
    2016-08-17 12:58:57,936 INFO stopped: grafana (exit status 0)
    2016-08-17 12:58:57,953 INFO stopped: consul (exit status 1)
    2016-08-17 12:58:58,029 CRIT reaped unknown pid 412)
    2016-08-17 12:59:01,032 INFO waiting for mysql to die

    2016-08-17 12:59:04,036 INFO waiting for mysql to die
    2016-08-17 12:59:07,040 INFO waiting for mysql to die
    2016-08-17 12:59:08,041 WARN killing ‘mysql’ (8) with SIGKILL
    2016-08-17 12:59:08,042 INFO stopped: mysql (terminated by SIGKILL)
    root@metrics:/opt/prometheus/data#
    root@metrics:/opt/prometheus/data# docker attach f752b9d2d4a7
    You cannot attach to a stopped container, start it first”

    • just in case it wasn’t clear, was doing a docker attach to debug.. admittedly i don’t use docker very much so that might not be the right way to poke around the container.

  • Hello Bob,

    I just installed pmm-server:1.0.4 on one of our docker-engine:1.12.1

    I am impressed by the quality of the platform. Congratulations and thank you for this work.

    I encountered some problems running the server and agents. Here are some feedback that may interest subsequent users.

    For the pmm-server:

    Grafana behind Nginx is inaccessible.
    Nginx responds with code 500.
    Nginx logs say that : (97: Address family not supported by protocol).

    I understand that internally, Nginx resolves localhost with ipv6 and it seems that IPv6 is disabled.
    I did not know the problem live in my docker-engine or in the image you push.

    Anyway, I bypassed the problem by changing the Nginx configuration (followed by a container restart):

    « docker exec pmm-server sed -i -e ‘s; localhost, 127.0.0.1; g’ /etc/nginx/nginx.conf »

    Magic: Grafana works!

    For pmm-client now:

    We use MariaDB 10.1 on Debian Wheezy (I know … wheezy is a bit outdated).

    No problem with linux:metrics and mysql:metrics.
    But mysql:queries gave me a headache.
    The « pmm-admin add mysql queries » seems to create an « empty-quoted »  file:

    /usr/local/Percona/qan-agent/\”\”

    Checking /var/log/pmm-mysql-queries-42001.log, i see:

    Version #: Percona qan-agent-1.0.4
    Basedir #: / usr / local / Percona / qan-agent
    # Listen: 127.0.0.1:42001
    # PID: 26365
    # API: 192.168.252.181/qan-api
    # UUID: ee264dcfa6d54198640295881f34a128
    2016/09/21 18: 15: 35.727966 main.go: 160: Starting agent …
    2016/09/21 18: 15: 35.728457 main.go: 338: Agent is ready
    2016/09/21 18: 15: 35.745314 main.go: 202: API is ready
    2016/09/21 18: 15: 36.840765 main.go: 164: Agent HAS stopped
    panic: runtime error: invalid memory address or nil pointer dereference
    [0xb code signal addr = 0x1 = 0x0 pc = 0x406435]

    goroutine 1 [running]
    panic (0x8a8c00, 0xc820012070)
    /opt/local/lib/go/src/runtime/panic.go:481 + 0x3e6
    main.run (0xc820070b80, 0x0, 0x0)
    /Users/weber/gopath/src/github.com/percona/qan-agent/bin/percona-qan-age
    nt / main.go: 377 + 0x3985
    main.main ()
    /Users/weber/gopath/src/github.com/percona/qan-agent/bin/percona-qan-age
    nt / main.go: 149 + 0x14f2

    Strange. Go is not my favorite language, I admit I did not understand so much.
    The fact is, delete this « empty-quoted » file under « /usr/local/Percona/qan-agent » before starting the mysql:queries service solve the problem.

    Next steps : start my performance_schema, I’ll do that in the coming days.

    I wanted to share my experience, there is still little return on the operation of your platform.

    I hope that this return will help.

    Again thank you for your work.

    Nicolas

Leave a Reply