“We made it a point to hire really smart, visionary people and then let them do their work.
I wanted to delegate and let people be in charge of things. My own decision-making process was to decide who got to decide. To make decisions, you have to first outline the problem, and if you hire really great people, they’re going to know more about the problem they’re dealing with than you ever will.”–Scott McNealy

I have interviewed Scott McNealy. Scott is a Silicon Valley pioneer, most famous for co-founding Sun Microsystems in 1982. We talked about Innovation, AI, Big Data, Redis, Curriki and Wayin.

RVZ

Q1. You co-Founded Sun Microsystems in 1982, and served as CEO and Chairman of the Board for 22 years. What are the main lessons learned in all these years?

Scott …

One problem that’s a lot less common these days is swapping. Most of the issues that cause swapping with MySQL have been nailed down to several different key configuration points, either in the OS or MySQL, or issues like the swap insanity issue documented by Jeremy Cole back in 2010. As such, it’s usually pretty easy to resolve these issues and keep MySQL out of swap space. Recently, however, we had tried all of the usual tricks but had an issue where MySQL was still swapping.

The server with the issue was a VM running with a single CPU socket (multiple cores), so we knew it wasn’t NUMA. Swappiness and MySQL were both configured correctly and when you checked the output of free -m it showed 4735M of memory available.

[sylvester@host~]$ free -m
              total        used        free      shared  buff/cache   available …

I’ve been posting quite a bit about Docker as I’ve been working with it a lot as of late. I thought I would share something interesting I discovered a couple weeks ago while working on setting up a Docker container-based lab environment for training purposes here at Pythian, specifically when attempting to install the MySQL-python package inside the container.

I know what you’re thinking: why is he trying to install a package in a Docker container? Doesn’t that go against the “Docker run and forget about it” philosophy? Sure, but in this case, I’m looking to add orchestration via ansible, which I don’t think is completely out of the question in order to coordinate your Docker containers for something like replication. This requires using the …

There are numerous ways to filter your replication stream in MySQL.  So many, that the documentation team has created guided “walk-through” pages for considering the various complex DB Filtering and subsequent Table Filtering paths folks have used for the “global” filtering options.  But now there are “channel specific” options that can be implemented at startup with enhanced… Read More »

In one of the projects I have been working on, one requirement was the ability to stop traffic from reaching a MySQL host which has been lagging behind its master for longer than a specific amount of time and then bring it back online once the lag has gone away. Of course, this is all automated and no human intervention is required.

In this scenario, we are using HAProxy as the load balancer, and I will walk you through how to configure an agent so we can use HAProxy httpchk to flag the host as up or down, via systemd socket and then automatically set the host as being down/up when applicable, in HAProxy.

I will be setting up a systemd service (I’m running centos7 hosts) and creating a listening socket in the MySQL host we want to monitor so haproxy can have access to replication status.

Scenario:

master: po-mysql1
slaves: po-mysql2, po-mysql3, po-mysql4
secondary slaves: …

In a previous post I had mentioned that I was doing a bit of digging into Docker in order to get a better grasp of the technology. Part of that was exploring common administrative tasks. I would venture to say that backups are probably among the most important tasks we take on with database administration, so it’s important to know how to do this for Docker MySQL instances.

There is a fair bit of documentation on how to handle this logically (mysqldump / mydumper) as this is a simple task to perform as long as you can connect to the database instance, so I wanted to approach physical backups using the very common xtrabackup tool. Additionally, we’re trying to think with containers here, so I wanted to make sure that not only would I be taking a backup of the Docker container MySQL instance, but I would do it with another Docker container running …

Percona Live Santa Clara, an annual event where open source database users, developers and enthusiasts come together, was held in April at the Santa-Clara convention centre. Pythian was well represented once more with no less than five presentations and a total of nine attendees.

This year the conference was condensed to two days of breakout sessions and one day of tutorials. Though it was shorter in length, the organizers broadened their horizons by including not only MySQL and MongoDB tracks, but this year they even put together a full PostgreSQL track. Moving from MySQL only to multiple technologies, inspired this year’s tagline: polyglot persistence conference. The increase in number of sessions allowed for a lot more options, but the condensed schedule made it much harder to choose which sessions to attend!

My observation from last year’s …

Earlier this year, I was presented with the challenge of streamlining user access to MySQL, allowing users self-serve access using their LDAP credentials, while logging all access. Of course, various MySQL forks allow for user auditing, but the solution is also needed to eventually support other data storage systems without native user auditing. This gave me the opportunity to do a trial integration of MySQL, Vault, ProxySQL, and LDAP; Vault would be used to dynamically create user accounts, and ProxySQL would be used to limit access and log activity. To evaluate the functionality and configuration of the integration, I used Docker to set up a test environment.

Below I will present the methods used to:

1. Provision the Docker environment.
2. Configure OpenLDAP. …

Introduction

In this post we will explore one approach to MySQL high availability with HAProxy, Consul and Orchestrator.
Let’s briefly go over each piece of the puzzle first:
– HAProxy is usually installed on the application servers or an intermediate connection layer, and is in charge of connecting the application to the appropriate backend (reader or writer). The most common deployment I’ve seen is to have separate ports for writes (which are routed to the master) and reads (which are load balanced over a pool of slaves).
– Orchestrator’s role is to monitor the topology and perform auto recovery as needed.
The key piece here is how we can make HAProxy aware that a topology change has happened, and the answer lies within Consul (and Consul templates).
– Consul is meant to be told the identity of the new master by Orchestrator. By leveraging Consul templates, we can then in turn propagate that …

Introduction

One of the things I like about Continuent Tungsten is how Tungsten Connector can hold traffic while a (graceful) master switch is taking place. This means the application may experience a brief spike in latency, but has not returned any errors from the database layer.

René also described a similar process using ProxySQL and mysqlrpladmin here. In this post we will try to achieve the same feat, a graceful master switchover with ProxySQL and Orchestrator.

ProxySQL considerations

ProxySQL needs to isolate our application from the changes going on at the database layer.
By design, if a query needs to be sent to a hostgroup that has no servers in ONLINE state, ProxySQL waits until either a server becomes available or a timeout expires (mysql-connect_timeout_server_max if I am not …