SQL Server Error Log Configuration

Comments 0

Share to social media

Most applications have a log of some sort that tracks activity. SQL Server is no different. It has a log file known as the error log for tracking what’s happening on an instance of SQL Server. Server. Each instance of SQL Server has its own set of error log files. In this article, I will discuss what the error log is, how SQL Server manages it,  and where you can find it. I’ll also show you how to configure the SQL Server error log. 

What is the error log?

The error log is a file that tracks what kind of activity is happening on an instance. The log is just a journal of events that have occurred on a SQL Server instance, in chronological order. The log doesn’t track everything, but it does track significant events and errors that occur while SQL Server is running. These errors could be informational only, warnings, as well as actual instance and application errors. You can find things like start up and shut down information, backup and restore commands, as well as custom application messages in the error log file. A DBA can also configure SQL Server to write additional logging, for example, logins and logouts. The error log is a great place to look for problems or potential problems associated with an instance of SQL Server.

The error log is not a single file but a series of files. Each time SQL Server starts up, a new error log file is created. A running instance of SQL Server writes to current log (the one created at startup) and by default has six archived error log files. If you need to keep more than six archived files, you can override the default to keep as many as you need (more on this later).

If an instance of SQL Server crashes or will not start for some reason, the error log is the place to go for troubleshooting these problems. As a DBA, you should periodically review the error log to look for potential problems. By reviewing the log, you might find some unusual things going on that might otherwise go unnoticed, like a backup job has failed or someone trying to hack the SA password.

Where can the error log files be found?

By default, the error log files are stored in the following location: Program Files\Microsoft SQL Server\MSSQL<n>.<instance name>\MSSQL\LOG\ERRORLOG, where <n> is an archived version number, and <instance name> is the name of the instance. This is only the default location. Some instances might be set up to write their error log files to a different location. If the error log files are not in the default location, then there are a number of different ways to find them. I will show you two of those methods.

The first method is to use SQL Server Configuration Manager. To find the location of the log using this method, first, open up SQL Server Configuration Manager. Next double click on the instance of SQL Server that you want to locate the error log file location. Then click on the Advanced tab. The location of the error log file directory is identified in the Dump Directory item. To see the full name of the error log file directory, click on the little down error to the right of the Dump Directory item, as shown below in Figure 1.

Figure 1: Location of error log file directory in Configuration Manager

A second method to find the location of the error log files is to use SSMS to browse one of the error log files. To do this, you must connect to the instance of SQL Server in which you want to find the error log location with SSMS. Expand the Management item and the SQL Server Logs item. Then double-click on the Current error log file. When you do this, the Log File Viewer will be displayed. To find the error log file location you can either browse thru the log file until you find it or use the Search… option to find it. When using the search option, use the string Logging SQL Server messages in file as the search criteria. The image in Figure 2 shows these steps. The log location can be found in the line that is highlighted in the log.

Figure 2: Using SSMS to find the location of error log files

 

Types of error log files and their naming conventions

Each SQL Server instance has two different types of error log files. There is an error log file for the currently running instance and then a number of archived error log files.

All error log files names start with ERRORLOG. The error log for the running instance is called just ERRORLOG, whereas all other error log files names start with ERRORLOG but have a number for the file extension. The archived error log files are ERRORLOG.1, ERRORLOG.2, ERROLOG.3, … up to the number to the configured number of archived error log files. Where ERRORLOG.1 is most recently archived error log file, ERRORLOG.2 is the second most recently archived error log file, ERRORLOG.3 is the third most recently archived error log file, etc. If an instance is configured with the default number of error log files, then the last archived error log file is named ERRORLOG.6.

When an instance is restarted, the oldest archived error log file (ERRORLOG.6 if the default number is used) is deleted, then each of the remaining error log files is renamed. ERRORLOG.5 is renamed to ERRORLOG.6, ERROLOG.4 is renamed to ERRORLOG.5, etc. until ERRORLOG.1 is renamed to ERRORLOG.2. The last current error log file (ERRORLOG) is renamed to ERRORLOG.1, and the error log file for the newly started instance is created with the name ERRORLOG.

Changing the number and size of the error log

By default, an instance of SQL Server retains six archived error log files, and the size of each log file is unlimited. An unlimited size means it will grow as big as it needs to be, or until it runs out of disk space. You may find that these default setting are fine, but they can also be overridden.

How many archived error log files do you need and what is an appropriate maximum size? Like most SQL Server questions, the answer is “it depends.” Here, I will demonstrate how the number and size of the error logs might help or hinder your use of the error log files and share my personal opinion of how many log files I like to have.

First, let me explore the number of archived log files. I like to retain at least two months’ worth of error log files. By having multiple months of error log information, I can track trends from one month to the next. Keep in mind that depending on what is logged and the activity of an instance, two months’ worth of error log files might take up a lot of disk space.

There is no easy way to keep exactly two months’ worth of error log files since a new error log is created and the oldest archived one is deleted every time an instance is restarted. The number of months of error log files depends on how often SQL Server is restarted and the number of archived error log files that are kept. Suppose an instance is recycled frequently over a short period of time due to some maintenance activity. Error log information will be cycled off based on the configured value for the number of archived error log files that should be retained.

There a few different methods to change the error log file configuration, I’ll mention two of them. The first method is to use SSMS. To use this method, first connect to the instance, expand the Management folder, right click on the SQL Server Log folder and then select the Configure item from the menu displayed, as shown in Figure 3.

Figure 3: Bringing up the error log configuration screen

When clicking on Configure, the window shown in Figure 4 is displayed.

Figure 4; Configuration Options for the Error Log

By reviewing Figure 4, you can see that my instance is currently configured to retain 6 archived error log files and the maximum size of the error log file is set to be 0, which means unlimited.

To change the number of archived error log files to retrain, first click on the checkbox labeled Limit the number of error log files before they are recycled. Doing this will enable the Maximum number of error log files option, so the number of error logs to retain can be changed. Note the number of error log files to retrain needs to be between 6 and 99.

To set a maximum limit of each error log file, enter the desired size into the Maximum size of the error log file in KB. When the error log configuration is set to limit the size of the error log file, a new error log file will be created automatically after the current log file reaches the maximum size. I personally don’t like to limit the size of the log file, but keep in mind, having an unlimited size setting means each log file will be different sizes, whereas limiting the size would make every error log file the same size.

Issues with multiple restarts

When SQL Server is restarted, the oldest archived error log file is deleted, and a new one is created. This cycling of the error log every time SQL Server restarts normally isn’t a problem, unless it deletes an archived error log file automatically that contains information that is needed for troubleshooting or trend reporting.

Since the log file contains messages and errors related to an instance, it is crucial not to lose this information until you no longer need it. The problem with having the default number of archived error log files is you might lose valuable troubleshooting information if an instance is restarted more than six times over a short period of time. This can happen for a number of different reasons. I have seen nighttime operations staff and DBAs restart an instance multiple times in an attempt to get an instance that has crashed back up and running. Suppose an instance of SQL Server is restarted multiple times without review of the error log files. The restarts might inadvertently cycle off the one critical error log file containing the error messages related to the crash. For this reason, it is important to configure SQL Server to retain enough archived error log files so they are not automatically deleted until they are no longer needed.

Keep in mind that cycling off a critical error log file is a rare situation. However, if an error log file that has valuable information about a crash is cycled off, it does hamper your troubleshooting options. Therefore, it is wise to have someone review the error log, or at least copy the error log file to a backup location after a crash to make sure you have all the log files to help with troubleshooting. At least by copying off the error log file after a crash, you can guarantee it will not be cycled off if lots of restarts are performed.

Cycling the ERRORLOG

A new error log is created every time SQL Server is restarted. Sometimes SQL Server stays up for weeks or even months before an instance is restarted. When an instance has been running for a long time, the error log will be quite large. It’s hard to review and find errors in large log files, especially when they contain error messages that could cover long periods of time.

To help keep the error log files from spanning months’ worth of messages without recycling, you can cycle the error log with a command. Cycling the error log will close the current log and create a new one. The code in Listing 1 uses the system stored procedure sp_cycle_errorlog to cycle the error log file.

Listing 1: Cycling the errorlog.

To execute the sp_cycle_errorlog system stored procedure, you need to be a member of the sysadmin fixed server role. Stored procedure sp_cycle_errorlog takes no parameters and returns a 0 if the error log is successfully recycled and a 1 if the cycling of the log has failed.

Error log file

The error log file contains messages and errors associated with a SQL Server instance. It is a valuable tool for troubleshooting problems with SQL Server as well as auditing what goes on within an instance of SQL Server. A DBA can configure SQL Server to store the error logs wherever they’d like and adjust the configuration to keep as many archived error log files as needed. As a DBA you should review this log every time SQL Server crashes and as part of your daily and weekly system checks. This review will ensure that SQL Server is functioning normally, with no unexpected errors showing up in the log.

 

If you like this article, you might also like:

Developing a backup plan

About the author

Greg Larsen

See Profile

Greg started working in the computer industry in 1982. In 1985, he got his first DBA job, and since then he has held six different DBA jobs and managed a number of different database management systems. Greg has moved on from being a full-time DBA and is now an adjunct professor at St. Martins University and does part-time consulting work. He has published numerous articles in SQL Server Magazine, and many online web sites dedicated to SQL Server. He is a former SQL Server MVP and holds a number of Microsoft Certification. Greg can be reached at gregalarsen@msn.com.

Greg Larsen's contributions