Kismetdb
The kismet log format is a unified common file of packets, non-packet data, devices, location, system messages, datasource records, historical trends, and almost everything else Kismet can track. It replaces the legacy logs where multiple log files were required and eliminates the need to correlate data from multiple logs to get a complete view of the capture history.
The kismet log file can be converted to many other log types using the associated log tools, including pcap, pcapng, wiglecsv, json records, KML, and more.
Log type
log_types=kismet
Kismet log journal files
The kismet log format uses sqlite3 to create a dynamic random-access file. If Kismet exits abnormally (such as running out of RAM or the power to the device failing), it may leave behind a ...-journal file.
This file is part of the sqlite3 integrity protection, and contains partial data which was not written to the database.
The journal file will be automatically merged when the log file is opened with sqlite3 or one of the log processing tools, but you can manually merge them with sqlite command line tools:
$ sqlite3 Kismet-foo-whatever.kismet 'VACUUM;'
or via the included Kismet log cleanup tool:
$ kismetdb_clean foo.kismet
If you have a journal file, it is important that you combine it with the log before transferring the kismet file.
Kismetdb ephemeral and timed logs
The kismet log format can be used as an ephemeral (non-permanent) log file, and can automatically time-limit the data. This is extremely useful if you are using Kismet as a stand-alone sensor, where the data is being collected over the REST interface.
Kismetdb logs can be automatically purged of old data. The timeout values are in seconds - 60 * 60 * 24, which is 86400, sets a 24 hour timeout on data.
kis_log_alert_timeout=86400
kis_log_device_timeout=86400
kis_log_message_timeout=86400
kis_log_packet_timeout=86400
kis_log_snapshot_timeout=86400
Data in each of the categories is removed from the kismetdb log file after the timeout. Devices which have been idle for more than the kis_log_device_timeout are removed.
A kismetdb log may be marked as ephemeral by setting:
kis_log_ephemeral_dangerous=true
An ephemeral log is removed from the filesystem upon creation; the log is not accessible on disk, and will be removed immediately upon Kismet exiting. Ephemeral logs are most useful when deploying Kismet as a permanent fixed sensor with a time-limited history; for instance retaining the past day of devices, packets, etc, but where there is no need (or desire) to keep log files.
Rotating kismetdb logs
After 2026-10-R1, the kismetdb log can start a new file after a time or a size, so finished files can be moved off the system while Kismet keeps running:
kis_log_max_seconds=86400
kis_log_max_mb=4096
kis_log_max_seconds is in seconds, up to a year (31622400), and kis_log_max_mb is in megabytes, up to 1048576; 0, or leaving it unset, is no limit. Whichever limit is reached first starts the next file. A negative, non-numeric, or too large value is logged as an error and ignored. The time counts from when the file opened, not from midnight, so 86400 makes a new file every 24 hours after Kismet started. Limits are checked once a second; the size is that of the database, including what hasn’t been written to the file yet.
No packets are lost when the file changes. Each file stands on its own: the old file gets a final update of the devices, and the new one gets every device, the datasources, the GPS position and a system snapshot right away. Alerts and messages stay in the file they were logged to, and datasource packet counts are since Kismet started, not per file.
New files keep the name the log started with and count up its sequence number, so the log template needs %i; without it, the log is not rotated and an error says so. If a new file can’t be opened, such as on a full disk, Kismet keeps logging to the current file and tries again a minute later.
Because every new file starts with all the devices Kismet is tracking, a file has to gain a quarter of kis_log_max_mb in new data before its size starts the next one; Kismet warns when the devices alone take half the limit. Raise the limit if you see that warning.
A pcapng download from the kismetdb log (/logging/kismetdb/pcap/) keeps reading the file it started on and does not delay the change. Each file which is closed and complete is announced with a Finished kismetdb log '<path>' message; that file can then be moved or compressed. After 2026-10-R1, it is also published as a LOG_FINISHED eventbus event, and a new file can be started on request through the logging API.
After 2026-10-R1, when a very busy Kismet holds up the next file for 10 seconds or more, an error says so, and repeats every 30 seconds until the new file starts:
Starting the next kismetdb log is 40 seconds overdue: writing the devices to the old file (waits for a device log pass already running) (3 attempts); still logging to '<path>'
with what it is waiting for. When the file does start, Started the new kismetdb log 52 seconds after it was due, after 4 attempts is logged. While the next file is late, the log’s kismet.logfile.rotate_overdue_since in /logging/active is the time it was due. The current file keeps growing past the limit in the meantime.
Log contents
Each kind of record in the kismetdb log can be turned off:
kis_log_devices=true
kis_log_packets=true
kis_log_messages=true
kis_log_alerts=true
kis_log_datasources=true
kis_log_datasources_rate=30
kis_log_datasources_rate is how often, in seconds, the data sources are logged. Through 2026-10-R1, kis_log_alerts=false didn’t stop alerts being logged, and the data source rate was only read by its old name, kis_log_datasource_rate; after 2026-10-R1, both options work, and the old name still works.
GPS snapshots
When a GPS is configured, Kismet logs snapshots of the GPS status to the snapshots table of the kismetdb log:
GPS: The location and fix quality of each GPS, every 10 seconds. This is controlled bykis_log_gps_track.GPSSATS: The satellites each GPS has in view (constellation, number, elevation, azimuth, signal, and whether each is used in the fix), along with the satellite counts, DOP, and precision. This is logged everykis_log_gps_satellites_rateseconds, only for GPS devices which list their satellites.
kis_log_gps_track=true
kis_log_gps_satellites_rate=30
The satellites move slowly across the sky, and a satellite snapshot is larger than a location, so they are logged less often. Each GPSSATS snapshot is a few kilobytes, usually well under 1MB per hour per GPS. Setting kis_log_gps_satellites_rate=0 turns off the satellite snapshots.
Tagged data from other tools
After 2026-10-R1, other tools can save their own JSON records in the kismetdb log through the tagged data API, using an admin login or an API key with the datalog role. Each record is saved to the snapshots table with the type API:<tag>, the time, and the GPS location if there is a fix. Records can include images or other binary data as base64 strings. Records can be read back through the API with an admin login or a readonly API key, and deleted with an admin login. Deleting records does not make the log file smaller; run kismetdb_clean on the log after Kismet has closed it to reclaim the space and remove the deleted data from the file.
The largest request the API accepts, in bytes, is set by:
kis_log_tagged_data_max_size=8388608
The default is 8MB; the value is limited to between 100000 and 67108864 (64MB). Each record is stored in full in the log, so large records, such as images, will grow the log quickly.
Filtering
Kismet can filter packets and devices logged in the kismetdb log file. This can be used to exclude known devices, or include ONLY certain types of devices and packets.
Filtering is most useful when coupled with other tools, and most users won’t need to worry about it, but can be a powerful tool in specific situations.
Duplicate packets
Kismet can record duplicate packets when multiple datasources capture the same data. A packet is a duplicate when another data source delivered the same frame within the last second; after 2026-10-R1, a data source which hears the same frame again (such as a Bluetooth LE device repeating an unchanged advertisement) has seen a new transmission, which is not a duplicate. In some instances, keeping duplicate packets is desirable (such as remote captures used for device location), while in others, logging duplicates may be a waste of space.
To exclude duplicate packets, set the following in your configuration:
kis_log_duplicate_packets=false
Data packets
Kismet typically logs all types of packets. To discard data packets, retaining only management frames, set:
kis_log_data_packets=false
Pass and block filters modes
Kismet filters operate as pass or block. Events which are passed are allowed to be logged, while events which are blocked are excluded.
Default and match modes
Kismet filters have a default mode: if a record does not match any other filter, the default mode is applied.
Records can match a pass or block mode before reaching the default mode. A filter with a default of block can still allow specific records through with a matching pass element.
Group matches
MAC addresses can be specified as a single address AA:BB:CC:DD:EE:FF, or as a masked group 11:22:33:00:00:00/FF:FF:FF:00:00:00.
A masked group works like an IP netmask specification: the value after the / indicates the bytes to match on; in the case above, the match will grab any MAC address beginning with 11:22:33:....
The most common use for masked matching is to match OUIs which are the first 3 bytes of the MAC address.
Device filters
Filtering device records from the kismetdb log will not prevent Kismet from showing them, but prevents Kismet from logging them.
Devices are filtered by phyname and mac address. Mac addresses can be specified as a single MAC or as a masked group.
The default filter controls are in kismet_filter.conf.
kis_log_device_filter_default={ pass } / { block }
pass.kis_log_device_filter={ phyname,macaddress,value }
Adds a device filter for the given phyname and MAC address. For example:
kis_log_device_filter=IEEE802.11,aa:bb:cc:dd:ee:ff,pass
kis_log_device_filter=IEEE802.11,00:11:22:00:00:00/ff:ff:ff:ff:ff:ff,block
Packet filters
Filtering packets from the kismetdb log will not prevent Kismet from processing them and creating related devices, but will prevent the actual packet data from being logged.
Packets can be filtered based on the MAC address of the source, destination, network, or ‘other’ address. For Wi-Fi, the ‘other’ address is used only in WDS environments where the quad-MAC headers are used. Packet filters may also filter on ‘any’ address, which matches any of the four.
The default filter controls are in kismet_filter.conf.
kis_log_packet_filter_default={ pass } / { block }
pass.kis_log_packet_filter={ phyname,addresstype,macaddress,value }
Adds a packet filter for the given phyname, address type (source, destination, network, other, any), and MAC address. For example:
kis_log_packet_filter=IEEE802.11,source,aa:bb:cc:dd:ee,pass
kis_log_packet_filter=IEEE802.11,any,00:11:22:00:00:00/ff:ff:ff:ff:ff:ff,block
Live filter control
Packet filters can be enabled and manipulated live via the filter REST API.