ChangeLog:
Logswan 2.0.4 (2019-08-16)
- Adding #include guard in compat header file
- Add an example log file and regenerate output example
- Add dependencies installation instructions for NetBSD and FreeBSD
- Add final dots for options descriptions
- Add final dot when printing results summary
- Use EXIT_SUCCESS and EXIT_FAILURE macros for return values
- Add a trailing newline when printing JSON output
OK kamil@
Update submitted by Frederic Cambus, the upstream developer of this package.
author's notes
==============
Notable changes:
- Logswan now links against geography/libmaxminddb instead of net/GeoIP
- Take back MAINTAINER, as I'm upstream for this package
upstream changelog
==================
Logswan 2.0.3 (2018-10-15)
- Use -std=c11, Logswan requires a C11 compiler for libmaxminddb
- Enable support for parsing HTTP/2.0 requests, for real this time
- Revert back to using INT64_MAX for strtonum() maxval, as maxval is long long
and using UINT64_MAX caused bandwidth parsing to always fail (Thanks James Loh
for reporting the issue)
- Move maps of countries and continents to separate files
Logswan 2.0.2 (2018-08-05)
- Use UINT64_MAX for strtonum() maxval
- Add missing headers and reorder includes
- Add support for HTTP/2.0
Logswan 2.0.1 (2018-06-27)
- Do not use -Werror by default
- Do not always call inet_pton two times per log line, this speeds things up
- Use bool types for isIPv4 and isIPv6
- Use the monotonic clock to determine runtime
- Move conditional includes for 'pledge' and 'strtonum' in compat.h
Logswan 2.0.0 (2018-03-16)
- Use type off_t for results struct member fileSize
- Reflect OpenBSD's pledge() changes
- Switch to using libmaxminddb and GeoIP2/GeoLite2 databases
- Add Antarctica to the list of continents
Changes since 1.06:
Harmonize arrays names
Remove array of months, it's currently unused and will likely remain so
Simplify internal JSON array and object names
Use OpenBSD style(9) for function prototypes and declarations
Revert back to using strtok, at least for now
Do not use EXIT_SUCCESS and EXIT_FAILURE macros anymore
Fix implicit function declaration rrror on NetBSD (Thanks Maya Rashish)
Remove now useless variables initialization and unnecessary includes
Do not add an extra new line when displaying usage or version
Package changes:
- Trimmed package description
- Update MAINTAINER + added missing trailing slash for HOMEPAGE
Upstream ChangeLog:
Logswan 1.05 (2016-02-25)
- Documentation update (notes on measuring Logswan memory usage)
- Add additional include directories for compat functions + dependencies
to avoid using relative path in includes
- Check that *lineBuffer is not NUL before attempting to parse log line
- Perform GeoIP lookup and HLL add in the same if block
- Increment IPv4 and IPv6 hits counters individually and conditionally
- Use CMake to check if the system has OpenBSD's pledge available and
link pledge conditionally using a null implementation when compiled
on non OpenBSD systems
- Adding an array of months, for the upcoming split log functionalities
Patch provided by the upstream developer Frederic Cambus
Package changes:
- follow upstream with the rule what's installed of the files
- use newly added -DGEOIPDIR in the place of older string replace function
- stop installing an example output as a configuration file
Upstream changelog:
Logswan 1.04 (2016-01-10)
- Moving global variables into main
- Using 'size_t' instead of 'int' for array indexes in for loops
- Using 'uint32_t' for all non 'uint64_t' integers
- Do not increment hits and processed lines counter for each parsed line,
compute total only once when everything is parsed
- Setting 'CMAKE_BUILD_TYPE' to 'Release' and formatting fixes
- Sanitize CMake script to build under NetBSD (Thanks Kamil Rytarowski)
- Initializing some uninitialized variables
- Renaming 'DATADIR' variables to 'GEOIPDIR'
Logswan 1.03 (2016-01-01)
- Remove header display and do not print name of processed file
- Print results to stderr instead of stdout
- Output JSON data to stdout instead of creating a new file
- Define GeoIP databases path in CMakeLists.txt
- Adding log file name in the JSON output
- Removing some hardcoded values and replacing them with constants
defined in config.h
- Breaking the loop when a match is found in the request parser
- Using enumeration constants instead of macros
- Process GeoIP continent information
- Re-ordering protocols and methods with more common occurences on top of
the list, allowing to break earlier when iterating through the array
- Adding support for reading logs from standard input
- Renaming 'definitions' files to 'config'
- Increasing countries array size, as an attempt to be future-proof
- Initial support for using pledge() on OpenBSD
- Documentation updates (HLL precision, Features list, GeoIP databases)
- Updated JSON output example
- Added a manual page
Logswan 1.02 (2015-11-02)
- Renaming 'resource' variable to 'request' in the 'logLine' struct
- Do not attempt to parse empty date tokens
- Do not attempt to parse empty request tokens (Thanks Brian Carpenter for
reporting the issue)
Upstream changelog
==================
Logswan 1.01 (2015-10-01)
- Documentation updates
- Fixing segfault when request data is empty or malformed (Thanks Jonathan
Armani for reporting and proposing a fix)
Logswan is a fast Web log analyzer using probabilistic data structures. It is
targeted at very large log files, typically APIs logs. It has constant memory
usage regardless of the log file size, and takes approximatively 4MB of RAM.
Unique visitors counting is performed using two HyperLogLog counters (one for
IPv4, and another one for IPv6), providing a relative accuracy of 0.10%.
Project design goals include : speed, memory-usage efficiency, and keeping the
code as simple as possible.
Logswan is opinionated software :
- It only supports the Common Log Format, in order to keep the parsing code
simple
- It does not split results per day, but log files can be split prior to being
processed
