Stacki at the Seattle Scalability Meetup

Joe Kaiser
Not a Doctor, just A System Engineer

Open Source Stack Installer
Stacki is a very fast and ultra reliable Linux server provisioning tool … at scale.
With zero prerequisites for taking systems from bare metal to a ping and prompt.

History
 Roots in Open Source
 Started life as the Rocks Project at UCSD
 Started in May ’00.
 6 month project going on ~16 years
 Roots in the HPC world

What Problem are we trying to Solve?

Problem
 OS Provisioning
 Disk Configuration
 Disk Controller Configuration
 Disk Partitioning
 Network Configuration
 Services configuration
 Application Deployment
 Life-cycle management of the cluster
Server Provisioning

Problem – Contd. …
Datacenter Provisioning
 Server Provisioning
 Heterogeneous Hardware
 Complex Network Configuration
 Bonding
 Bridging
 VLANs
 Combinations of the above

Datacenter Architecture
Frontend
Network 1 Network 2 Network 3 Network 4

Challenge
 Fast
 Correct
 Consistent
 Repeatable

From Bare Metal Up
Take complete control of the Stack
 Modified CentOS Installer
 Parallel package sharing installer
 Database to keep persistent data about the System
 Command Line to interact with Stacki
 Dynamic Kickstart File Generation

Frontend Services
Services to build backend nodes
 DHCP – MAC to IP address Mapping
 TFTP – Serve out PXE files, Installation Kernel, and RAM Disk
 Apache – Serve Kickstart files
 DNS (optional)
Services to access backend nodes
 SSH key management
 Parallel execution shell

Stacki Positioning
DevOps / Configuration Tool
DHCP /
DNS / TFTP
NetworkDiskOS
In-house
developed
deployment
tools
- Disk Array Controller Configuration
- Disk Partitioning Configuration

Download and Boot the ISO
Download the ISO from
www.stacki.com
 It’s 1.5 GB
 stacki pallet
 Subset of CentOS 6.7
Boot the ISO on the host that will be
your frontend

Adding Hosts
Method 1: Discovery
Advantages
 Prior knowledge of MAC addresses Not Required
 Automatic Sensible Hostname, IP address assignment
Disadvantages
 Automatic Sensible Hostname, IP address assignment
 Complex network configuration has to be done post-installation
Run
# insert-ethers

Adding Hosts
Method 2: Host Configuration Spreadsheet
Advantages
 Complete control of Hostname, IP address, and network assignments
 Easy to make changes
 Fits very well with existing datacenter management processes.
 Lots and lots of Error Checking
Disadvantages
 A little tedious the first time around
 Requires prior knowledge of
 MAC addresses,
 IP address assignments
 Physical location of machines (Rack & Position)

Host Configuration Spreadsheet

Backend Installation
 Save your Host Configuration spreadsheet as a CSV
 Import CSV on frontend
 # stack load hostfile file=hosts.csv
 Tell backend nodes to install on their next PXE boot
 # stack set host boot backend action=install
 PXE boot all backend nodes
 Go!

BitTorrent-Inspired Package Installation
Stacki

Advanced Networking
Advanced Network Configuration
 Bonded interfaces
 VLANs
 Bridging
 Any combo of the above
Multiple Subnets
 Build a single cluster from hosts in multiple subnets
 Manage hosts in multiple datacenters

Disk Controller Configuration
Disk Controller Support
 LSI MegaRAID controller & derivatives
 Intel MegaRAID
 Dell MegaRAID
 Cisco SAS MegaRAID
 Any controller that supports the “storcli” or “megacli” command
 HP Smart Storage Controller support
 Supports RAID 0,1,5,6,10,50,etc.
 Configure Controllers using Spreadsheets
 # stack load storage controller

Disk Controller Configuration Spreadsheet

Disk Partitioning
 Sensible Default Disk partitioning
 Support for multiple disks
 Support for file system options and mount options
 Support for Software RAID configuration
 Disk Partitioning through spreadsheets
 # stack load storage partition

Disk Partition Configuration Spreadsheet

Software Footprint
Controlling Packages and Configuration
 Pallets
 Carts
 Boxes
 Distributions
 Appliances

Pallets
Software Entity
 Contains RPMS
 Contains Configuration in the form of XML
 Used for installation and configuration of an Application
 Can be applied during Frontend installation or after the fact.
 Each pallet functionally equal to a YUM repo with extra configuration
Example: Cloudera Pallet
 Contains RPMS required to install the Cloudera Distribution of Hadoop
 Contains scripts to configure and starts CDH

Example: Stacki with Cloudera Pallet
Check namenodes are
empty
Format/start HDFS
Create all directories
Create all metastores
Start services (Hbase, Hive,
Oozie, Sqoop, Impala, etc)
Deploy client configuration Configure database
Setup/assign monitors
(activity, services, and host)
Test database connections
Validate/resolve hostnamesConsistent host timezones
No bad kernel versions
running
(CDH) version consistency
Java version consistency
Daemons versions
consistency
Mgmt Agents versions
consistency
Host specification/SSH
ports
MUCH MORE …
DHCP Server/Client setup TFTP/PXE configuration
Server OS installation
Node OS Install
RAID configuration
Boot configuration
System/data disk
partitioning
Monitoring system setup
and config
Lights Out/IPMI setup
User accounts added and
synced
SSH keys on all hosts
Network node configuration
Config Mgmt install and
configuration
Route configurationOS upgrades/updates
Site specific software and
configuration
Host specification/SSH
ports
Security
Firewall setupCluster Mgmt utility Database install and config
Multiple network configPackage installation MUCH MORE …
App Config
Site Config
HW Install
Without Stacki
Stacki w/ Hadoop Pallet

Carts
Site Specific Pallets
 Contains site-specific RPM
 Contains site-specific configurations
 Structurally and Functionally equivalent to a Pallet
Example: Client Cart
 Contains RPMS to install DevOps tools
 Contains custom post-install scripts to configure DevOps tools
 Contains custom post-install scripts to run DevOps tools to bring system up to requisite
configuration.

Boxes
Logical Entity
 Loose collection of Pallets and Carts
 One-to-Many mapping to Backend Hosts
OS Pallet
Cloudera Pallet
Stacki Pallet
Pallets Carts
PayPal Cart
Ansible Cart
Boxes
RedHat Pallet
Stacki Pallet
PayPal Cart
Application
RedHat Pallet
Cloudera Pallet
Ansible Cart
OS Pallet
Stacki Pallet
PayPal Cart
Default

Boxes
OS Pallet
Stacki Pallet
PayPal Cart
Default
Stacki Pallet
PayPal Cart
Application
RedHat Pallet
Cloudera Pallet
Ansible Cart

Multiple Distributions
Default Distribution
 Based on stripped down CentOS 6.7 or 7.2
 Used to build backend nodes
Multiple Distributions on Frontend
◦ E.g., RHEL 6.x based distribution, CentOS 6.7, etc.
Backend Nodes  Distribution Mapping
 Any Node can be mapped to any distribution

In Conclusion
Production Ready
 Deploy large scale Big Data & OpenStack clusters very fast.
 Deploy test systems to evaluate multiple applications with very short turn-around times
 Deploy several small datacenters-in-a-rack that’s shipped out to customer sites.

Try it Out!
 Website
www.stacki.com
 Source Code
github.com/stackiq/stacki
 Google Groups
groups.google.com/forum/#!forum/stacki

Stacki at the Seattle Scalability Meetup

More Related Content

What's hot (20)

Viewers also liked (13)

Similar to Stacki at the Seattle Scalability Meetup (20)

More from StackIQ (20)

Recently uploaded (20)

Stacki at the Seattle Scalability Meetup

Editor's Notes