Actionable Logging For Smoother Operation and Faster Recovery Mandi Walls AOL, LLC June 23, 2008 Velocity 2008.

Actionable LoggingFor Smoother Operation and Faster Recovery

Mandi WallsAOL, LLCJune 23, 2008Velocity 2008

Actionable Logging

•What is “Actionable”•Goals of logging in production• Logging quality information• Improving log contents

Actionable

•No nonsense logging•Concise, easy to understand• Express symptoms of production issues•Anything that makes the log needs to be fixed

Why It’s Important

• Expending resources on production systems• The point of logging in production• Diagnosis of issues• The 4am Test

Logging Goals

• Diagnosis and recovery• Statistics and monitoring• Provide insight into the behavior of the application• Indicate potential issues, and areas for improvement• Not the same goals as development and QA environments!

Types of Logs

• Access log• Server log, i.e., catalina.out• Application logs• Special use logs for recording specific groups of activities

Log File Location

•Where logs are located on the system should be predictable and obvious• It may be helpful to locate logs on different disk partitions but link them back to the app•Keep older logs in an obvious place

Log File Management

• Everyone has their own method• Roll logs into files with timestamps:

– host-01.log.003 vs host-01.log.06202008• Roll all the logs at the same time for a given app to make coordination of events easier• Roll when the app needs logs rolled: hourly, daily, weekly• Don’t rely on STDOUT or server files that can’t be rolled without a hassle

Logging Quality Information

• Logs should be expressive but not overly verbose• Keys to making logs more actionable:

– Appropriate Formats– Quality Messages

Quality Information: Format

• Timestamping: what not to do1213988938:tvdata shows/617/3061213988939:tvdata shows/618/307

20/130055 err(4) lang-locale “es-us” not found

SEVERE: Error listenerStart

Quality Information: Format

• Timestamps that mean somethingJun 19, 2008 4:20:25 PM org.apache.catalina.startup.Catalina start

192.168.1.10 - - [20/Jun/2008:15:15:58 -0400] "GET /monitors HTTP/1.0" 200 230 0.049909

• Good timestamps give context for linking to external events like network outages or traffic anomalies

Quality Information: Format

• Other considerations in log file format include:– Creating a common format for multiple products and log

types– Limiting the number of log entries that write to multiple log

lines for faster parsing– Deciding how much is too much information

Quality Info: Good Messages

• Here’s some bad messages:[19/Jun/2008:11:14:03][14960.229405][-conn:thread::6] Error: $$$$$$$$$$$$$$$

[19/Jun/2008:11:58:32][32652.67698738][channels_news] Notice: My gallery : xl

[19/Jun/2008:12:03:29][32652.67010608][channels_games] Notice: 0

[19/Jun/2008:11:58:28][32652.67715090][channels_money] Error: ViewCounter: APP2 returns statusCode=400,

statusText=Invalid request• Other things to avoid: messages with only numerical error codes in them

Quality Info: Good Messages

• Here’s some messages that are reasonably helpful:[19/Jun/2008:12:03:30][32652.66764839][channels_money] Notice: INFO_FEED: moduleId(283403) failed with url=http://rss.businessweek.com/bw_rss/bwdaily

[19/Jun/2008:12:09:52][32652.68059183][channels_news] Error: processModule.inc: us.news.story: can't read "useragent": no such variable

• One that needs a little tweaking:[19/Jun/2008:00:01:48][15446.36831248][channels_games] Error: dom parse timeout doc: error "syntax error" at line 1 character 0

"t <--Error-- imeout"

Quality Info: Making Messages Useful

• Misleading severity [19/Jun/2008:11:55:40][20300.704556][-conn:thread::24] Error: [fatal]: APP1: no Published data for: app1_config3, dirpage.index

• Incorrect severity, particularly of debugging messages left in at production logging levels• Not logging anything for fatal errors

Improving Log Messages

• Log at the first point an error is encountered – don’t log a timeout to a backend as a parse error of data expected from the request• Messages include the method name and key variables to speed up fixes• Suppress anything in the log that isn’t actionable – whether debugging information or chronic issues no one will fix by changing the log level• Make checking the logs part of the QA process

Log Message Convergence

• Actively managing, parsing, pruning logs make new errors more obvious• Check the logs after every install for new messages that indicate issues or are junk that slipped through into production

Things to Avoid in Logs

• Usernames, passwords, database logins– Provide crib notes for anyone gaining unauthorized access to

the system– These are hard to avoid in some environments, particularly if

the username is part of the url– User name can be separated from display name to avoid

revealing too much in logs

How much is too much?

• A server or application log that has more than 25% of the number of access log entries is a hindrance. Even 10% may be too much in most environments• If a single application log has more entries than its corresponding access log, it’s time to have a long talk with development about removing log entries or creating multiple log files

Conclusion

• The log is the first line of information when a problem occurs• A production log should be focused on providing information to Operations staff, not for developers• When, where, and how messages are logged can help or hinder recovery after a problem

Questions and Comments