I did a lot of researching, and basically, there are a few primary culprits I was able to identify that will hopefully save you some time.

Check your /tmp directory

MySQL will produce this error sometimes when the temp directory is not writeable.

1. Ensure that /tmp (and/or /var/tmp) has the correct permissions (777)
2. Check the my.cnf file and search for a tmpdir=/tmp flag. Ensure the value is pointing to the correct temp directory.
3. Ensure your /tmp directory is not full

Check your my.cnf

1. If you made changes recently, revert them and restart MySQL (especially InnoDB Buffer Pool settings)
2. Restore my.cnf.back is there is one
3. If you are using InnoDB tables, ensure the skip-innodb line in my.cnf is commented out or removed.

Clear InnoDB Log Files

This step ONLY APPLIES IF THE ABOVE STEPS DID NOT WORK.

Read the MySQL Manual page on removing InnoDB log files for a safer backup and restoration procedures. Basically, the steps are:

1. Shut down MySQL
2. Remove ib_logfile* files from the MySQL data directory (move them or rename them if you want to be safe)
3. Re-start MySQL

My specific problem was that somehow the “skip-innodb” line got added back into my “my.cnf” file, so MySQL was expecting a different table format when loading data. I suspect this had something to do with my cPanel/WHM setup overwriting the file, but I’ll never know for sure.

Good Luck!

Queries are case-sensitive

Fields and queries in MongoDB are case-sensitive:

1
2
3

var test1 = db.test.find({'tags': 'jquery'}).count();
var test2 = db.test.find({'tags': 'jQuery'}).count();
test1 == test2; // Output is false - they do not query for the same information

This can cause some headaches if you don’t normalize user input ahead of time. Imagine you already have several posts tagged with ‘NoSQL’ and users enter ‘nosql’ in a tag search box. If you don’t normalize how the data is stored internally (like lowercase all tags), your user will see a much smaller set of posts than they are expecting to see. This is something you don’t have to worry or even think about with MySQL and most other relational databases.

If you can’t normalize the stored data, but still want a case-insensitive query, then you must perform a slower regular expression query:

1

db.test.find({'tags': /jquery/i}); // Note the 'i' flag for case-insensitive

Data is type-sensitive

Data stored within MongoDB knows it’s type. There is a small but significant difference between these two records:

1
2

{'count': 102}; // 'count' is stored as an int
{'count': "102"}; // 'count' is stored as a string

This is obvious to any programmer when presented like this, but what may not be obvious is that this also affects how you can query for these records:

1
2

// This returns 1 instead of 2, because it only matches the integer value
db.test.find({'count': 102}).count();

This is due to how MongoDB stores documents internally with BSON (Binary JSON), using various MongoDB data types. This means – like the point above – that you must pay attention to how you are saving data into MongoDB, because it will affect how you can query for it later.

Documents sizes are capped at 4MB each

This isn’t a big issue for most people, but it’s something to be aware of if you plan on storing large chunks of text or nesting a bunch of objects inside a single document. Nesting comments inside article documents is a particular approach that may give you pause knowing there is an upper threshold on document size.

Note that this limitation is not an issue for storing files in MongoDB because GridFS transparently divides the contents of files larger than 4MB across multiple documents.

Only one index is used per query

Simply adding more indexes doesn’t always help queries run faster. MongoDB can’t use multiple indexes together like MySQL and other RDBMS can (see “Index Merge Optimization“). This means that if a query is selecting or sorting based on multiple fields, a compound index should be created with those fields for the query to run most efficiently.

MongoDB is for high-memory and 64-bit systems only

Although MongoDB can be downloaded, compiled, and installed on 32-bit systems, it should never be run in production on them. This is because MongoDB stores all indexes in memory as well as memory-mapped files for all disk I/O for increased speed and throughput. While these two facts aren’t “gotchas” themselves – they are clearly explained on the website – it does mean that MongoDB can quickly use a lot of memory, especially as the size of your database grows and becomes significant. Since 32-bit systems have an effective 3GB memory limit of addressable memory space, they prevent you from being able to add more memory as the size of your database grows. This is also something to think about if you plan to run MongoDB on a small VPS with limited memory, even if it is 64-bit. You may be forced into a memory upgrade sooner than you are prepared for if your database is large or rapidly growing.

Subtle Differences

When starting out with MongoDB, it’s easy to draw a lot of parallels to MySQL and make a lot of assumptions based on those similarities. Collections are kind of like tables, fields are kind of like columns, a document is kind of like a row. You may find yourself going on to make more assumptions about how it’s storing data, how you can query it, what you can do with it, etc. without even knowing it. So while MongoDB has a lot of similar features and is one of the easiest NoSQL databases to transition to from a relational database, it has some very distinct differences “under the hood” that you have to be aware of and plan for ahead of time. The MongoDB documentation is well-written, and is pretty good at explaining any potential differences or “gotchas”, so make sure to read it thoroughly before making the jump – and watch that first step.

Once you fully get the NoSQL viewpoint and the implications it has on the future of data storage, you have one of those “why didn’t anyone think of this sooner?” moments. And yes, I know key/value stores like memcached existed long before this whole new “NoSQL” movement, but it’s not quite the same thing.

Object Databases

Object databases are an interesting case. They are in the “NoSQL” category and solve the same problem other NoSQL databases do – creating a storage mechanism that fits your data instead of making your data fit a predefined rigid storage mechanism. So why didn’t they catch on? What gives? It’s anyone’s guess why object databases didn’t catch on and achieve widespread adoption, but one thing is for sure – object databases really missed the boat while other NoSQL database technologies are running rings around them.

The primary problem is that most object databases are designed for a specific language, like Java or .NET. Some have more language support than others, but the problem is still the same – by integrating in directly with a language’s object model they depend on specific language implementations, which severely limits adoption. They solve the impedance mismatch problem that RDBMSes present, but fail to provide language-agnostic data persistence and access, effectively tying their use to specific technology stacks and platforms.

The Best of Both Worlds

NoSQL databases are gaining traction like no other database technology has in over 30 years. The reason is that good NoSQL databases solve both key data storage problems: impedance mismatches AND language-agnosticism. By storing the data in an intermediary format like JSON (or BSON) and allowing custom file content types, the focus is on the actual data itself instead of specific technologies or data models. The technology stack becomes irrelevant, and the walls to gaining user adoption crumble.

The bottom line is that NoSQL databases are not just the latest fad. They represent a fundamental paradigm shift in data storage and are an entirely new way of thinking about how to store and access your data in a way that makes sense for your actual usage. NoSQL databases are absolutely something you should be paying attention to and following closely – they are a strong contender to the RDBMS model and will likely become the de-facto data storage choice for most new web applications within a decade.

Return all the rows with multi-byte characters in table posts on field title:

1
2
3

SELECT *
FROM posts
WHERE LENGTH(title) <> CHAR_LENGTH(title)

This does pretty much what it looks like: returns all the rows in the posts table where the title’s character length does not match the title’s length. This works because the LENGTH function returns the number of bytes in the string, while the CHAR_LENGTH function returns the number of characters in the string. If the string contains multi-byte characters (individual characters that are made up of more than one byte) like international UTF-8 characters, the two functions will return different numbers and will not be equal, thus including that row in your results.

Using the MySQL Control Flow Functions, I quickly crafted an IF() statement in the SELECT portion of the query that looked like this:

1

IF(m.date_released IS NULL OR m.date_released = '0000-00-00', 1, 0) AS in_production

So if the release date has a NULL value or has been inserted as a blank string and formatted to ’0000-00-00′, it is given a value of 1, and 0 otherwise. We then sort descending by our new ‘in_production’ alias in the ORDER BY clause to get the NULL or empty values on top (assigned a 1 value), and then by release date and other criteria second.

The whole query looks something like this:

1
2
3

SELECT m.*, IF(m.date_released IS NULL OR m.date_released = '0000-00-00', 1, 0) AS in_production
FROM movies AS m
ORDER BY in_production DESC, m.date_released DESC

It’s an easy way to manipulate the result set in MySQL without slowing down the query or issuing multiple queries.