Interesting Resources for Technical Operations Engineers
23 Sep2013

As a leader of a technical operations team I often have to work on technical operations engineer hiring. This process involves a lot of interviews with candidates and during those interviews along with many challenging practical questions I really love to ask questions like “What are the most important resources you think an Operations Engineer should follow?”, “What books in your opinion are must-read for a techops engineer?” or “Who are your personal heroes in IT community?”. Those questions often give me a lot of information about candidates, their experience, who they are looking up to in the community, what they are interested in, and if they are actively working on improving their professional level.

Recently, one of the candidates asked me to share my lists with him and I thought this information could be valuable to other people so I have decided to share it here on my blog.

Read the rest of this entry


Join Me at Swiftype!
18 Sep2013

As you may have heard, last January I have joined Swiftype – an early stage startup focused on changing local site search for the better. It has been a blast for the past 8 months, we have done a lot of interesting things to make our infrastructure more stable and performant, immensely increased visibility into our performance metrics, developed a strong foundation for the future growth of the company. Now we are looking to expand our team with great developers and technical operations people to push our infrastructure and the product even further.

Since I have joined Swiftype, I have been mainly focused on improving the infrastructure through better automation and monitoring, and worked on our backend code. Now I am looking for a few good operations engineers to join my team to work on a few key projects like building a new multi-datacenter infrastructure, creating a new data storage for our documents data, improving high-availability of our core services and much more.

To help us improve our infrastructure we are looking both for senior operations engineers and for more junior techops people that we could help grow and develop within the company. Both positions could be either remote or we could assist you with relocation to San Francisco if you want to work in our office.

If you are interested, you can take a look at an old, but still pretty relevant post I wrote many years ago on what I believe an ops candidate should know. And, of course, if you have any questions regarding these positions in Swiftype, please email me at kovyrin@swiftype.com or use any other means for contacting me and I will try to get back to you as soon as possible. If you know someone who may be a great fit for these positions, please let them know!


Adding Custom Hive SerDe and UDF Libraries to Cloudera Hadoop 4.3
26 Jul2013

Yet another small note about Cloudera Hadoop Distribution 4.3.

This time I needed to deploy some custom JAR files to our Hive cluster so that we wouldn’t need to do “ADD JAR” commands in every Hive job (especially useful when using HiveServer API).

Here is the process of adding a custom SerDE or a UDF jar to your Cloudera Hadoop cluster:

  • First, we have built our JSON SerDe and got a json-serde-1.1.6.jar file.
  • To make this file available to Hive CLI tools, we need to copy it to /usr/lib/hive/lib on every server in the cluster (I have prepared an rpm package to do just that).
  • To make sure Hive map-reduce jobs would be able to read/write JSON tables, we needed to copy our JAR file to /usr/lib/hadoop/lib directory on all task tracker servers in the cluster (the same rpm does that).
  • And last, really important step: To make sure your TaskTracker servers know about the new jar, you need to restart your tasktracker services (we use Cloudera Manager, so that was just a few mouse clicks ;-))

And this is it for today.


MySQL Monitoring With Cacti Using Percona Monitoring Plugins (1-minute resolution)
26 Jun2013

Today, just like many times before, I needed to configure a monitoring server for MySQL using Cacti and awesome Percona Monitoring Templates. The only difference was that this time I wanted to get it to run with 1 min resolution (using ganglia and graphite, both with 10 sec resolution, for all the rest of our monitoring in Swiftype really spoiled me!). And that’s where the usual pain in the ass Cacti configuration gets really amplified by the million things you need to change to make it work. So, this is a short checklist post for those who need to configure a Cacti server with 1 minute resolution and setup Percona Monitoring Plugins on it.

Read the rest of this entry


Adding LZO Support to Cloudera Hadoop Distribution 4.3
13 Jun2013

Just a short note to myself and others who need to add LZO support for CDH 4.3.

First of all, you need to build hadoop-lzo. Since CDH 4.3 uses hadoop 2.0, most of the forks of hadoop-lzo project fail to compile against new libraries. After some digging I’ve found the original twitter hadoop-lzo branch to be the most maintained and it works perfectly with hadoop 2.0. So, download it, install pre-requisites, build it.

I have built it for us as an RPM, you can check out the spec file here (it depends on some other packages from that repo, but you should get the idea and should be able to modify the script to build on vanilla Redhat linux w/o additional packages). Another option would be to take a look at Cloudera’s GPL Extras repository and their lzo packages and documentation.

After you have built and installed your LZO libraries, you should be able to use them with HBase without any additional configuration. To test HBase support for LZO compression you could use the following command:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
$ hbase org.apache.hadoop.hbase.util.CompressionTest file:///tmp/testfile lzo
13/06/13 04:43:14 WARN conf.Configuration: hadoop.native.lib is deprecated. Instead, use io.native.lib.available
13/06/13 04:43:14 INFO util.ChecksumType: Checksum using org.apache.hadoop.util.PureJavaCrc32
13/06/13 04:43:14 INFO util.ChecksumType: Checksum can use org.apache.hadoop.util.PureJavaCrc32C
13/06/13 04:43:14 DEBUG util.FSUtils: Creating file=file:/tmp/testfile with permission=rwxrwxrwx
13/06/13 04:43:15 ERROR metrics.SchemaMetrics: Inconsistent configuration. Previous configuration for using table name in metrics: true, new configuration: false
13/06/13 04:43:15 WARN metrics.SchemaConfigured: Could not determine table and column family of the HFile path file:/tmp/testfile. Expecting at least 5 path components.
13/06/13 04:43:15 INFO lzo.GPLNativeCodeLoader: Loaded native gpl library
13/06/13 04:43:15 INFO lzo.LzoCodec: Successfully loaded & initialized native-lzo library [hadoop-lzo rev 64cec2e0439bd92a0a6bf3af28f5015a6836fc32]
13/06/13 04:43:15 INFO compress.CodecPool: Got brand-new compressor [.lzo_deflate]
13/06/13 04:43:15 DEBUG hfile.HFileWriterV2: Initialized with CacheConfig:disabled
13/06/13 04:43:15 WARN metrics.SchemaConfigured: Could not determine table and column family of the HFile path file:/tmp/testfile. Expecting at least 5 path components.
13/06/13 04:43:15 INFO compress.CodecPool: Got brand-new decompressor [.lzo_deflate]
SUCCESS

You’re looking for that last line to say SUCCESS. If it fails, it means you did something wrong and it will tell you what that is.

Now, if you want to use LZO for map-reduce jobs, you need to make a few changes in your /etc/hadoop/conf/core-site.xml config file. If you manage your configuration yourself, just add the following to your configuration file:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
<property>
  <name>io.compression.codecs</name>
  <value>
    org.apache.hadoop.io.compress.DefaultCodec,
    org.apache.hadoop.io.compress.GzipCodec,
    org.apache.hadoop.io.compress.BZip2Codec,
    org.apache.hadoop.io.compress.DeflateCodec,
    org.apache.hadoop.io.compress.SnappyCodec,
    org.apache.hadoop.io.compress.Lz4Codec,
    com.hadoop.compression.lzo.LzoCodec,
    com.hadoop.compression.lzo.LzopCodec
  </value>
</property>

<property>
  <name>io.compression.codec.lzo.class</name>
  <value>com.hadoop.compression.lzo.LzoCodec</value>
</property>

If you’re managing your configuration with Cloudera Manager, you need to do the following:

  1. Go to your map-reduce service
  2. Click “Configuration” and select “View and Edit
  3. In the list on the left select “Gateway (Default)” and “Compression
  4. Add two items to the list of compression codecs: com.hadoop.compression.lzo.LzoCodec and com.hadoop.compression.lzo.LzoCodec
  5. Open “Service Wide” => “Advanced” in the list on the left
  6. Add the following configuration to your “MapReduce Service Configuration Safety Valve for mapred-site.xml” section:
    1
    2
    3
    4
    <property>
      <name>io.compression.codec.lzo.class</name>
      <value>com.hadoop.compression.lzo.LzoCodec</value>
    </property>
  7. Click “Save Changes
  8. Restart your map-reduce cluster with updated configuration

Now you should be able to use LZO in your map-reduce, hive and pig jobs.