[go: up one dir, main page]

Skip to content

A Python application to filter and transfer Zeek logs to Elastic/OpenSearch+Humio. This app can also output pure JSON logs to stdout for further processing!

License

Notifications You must be signed in to change notification settings

corelight/zeek2es

Repository files navigation

zeek2es.py

This Python application translates Zeek's ASCII TSV and JSON logs into ElasticSearch's bulk load JSON format.

Table of Contents:

Introduction

Kibana

Want to see multiple Zeek logs for the same connection ID (uid) or file ID (fuid)? Here are the hits from files.log, http.log, and conn.log for a single uid:

Kibana

You can perform subnet searching on Zeek's 'addr' type:

Kibana Subnet Searching

You can create time series graphs, such as this NTP and HTTP graph:

Kibana Time Series

IP Addresses can be Geolocated with the -g command line option:

Kibana Mapping

Aggregations are simple and quick:

Kibana Aggregation

This application will "just work" when Zeek log formats change. The logic reads the field names and associated types to set up the mappings correctly in ElasticSearch.

This application will recognize gzip or uncompressed logs. This application assumes you have ElasticSearch set up on your localhost at the default port. If you do not have ElasticSearch you can output the JSON to stdout with the -s -b command line options to process with the jq application.

You can add a keyword subfield to text fields with the -k command line option. This is useful for aggregations in Kibana.

If Python is already on your system, there is nothing additional for you to copy over to your machine than Elasticsearch, Kibana, and zeek2es.py if you already have the requests library installed.

Installation

Assuming you meet the requirements, there is none. You just copy zeek2es.py to your host and run it with Python. Once Zeek logs have been imported with automatic index name generation (meaning, you did not supply the -i option) you will find your indices named "zeek_zeeklogname_date", where zeeklogname is a log name like conn and the date is in YYYY-MM-DD format. Set your Kibana index pattern to match zeek* in this case. If you named your index with the -i option, you will need to create a Kibana index pattern that matches your naming scheme.

If you are upgrading zeek2es, please see the section on upgrading zeek2es.

Elastic v8.0+

If you are using Elastic v8.0+, it has security enabled by default. This adds a requirement of a username and password, plus HTTPS.

If you want to be able to delete indices/data streams with wildcards (as examples in this readme show), edit elasticsearch.yml with the following line:

action.destructive_requires_name: false

You will also need to change the curl commands in this readme to contain -k -u elastic:<password> where the elastic user's password is set with a command like the following:

./bin/elasticsearch-reset-password -u elastic -i

You can use zeek2es.py with the --user and --passwd command line options to specify your credentials to ES. You can also supply these options via the extra command line arguments for the helper scripts.

Docker

Probably the easiest way to use this code is through Docker. All of the files are in the docker directory. First, you will want to edit the lines with CHANGEME!!! in the .env file to fit your environment.
You will also need to edit the Elastic password in docker/zeek2es/entrypoint.sh to match. It can be found after the --passwd option.
Next, you can change directory into the docker directory and type the following commands to bring up a zeek2es and Elasticsearch cluster:

docker-compose build
dockr-compose up

Now you can put logs in the VOLUME_MOUNT/data/logs directory (VOLUME_MOUNT you set in the .env file). When logs are CREATED in this directory, zeek2es will begin processing them and pushing them into Elasticsearch. You can then login to https://localhost:5601 with the username and password you set up in the .env file.
By default there is a self signed certificate, but you can change that if you edit the docker compose files. Once inside Kibana you will go to Stack Management->Data Views and create a data view for logs* with the timestamp @timestamp. Now you will be able to go to Discover and start searching your logs! Your data is persistent in the VOLUME_MOUNT/data directory you set. If you would like to remove all data, just rm -rf VOLUME_MOUNT/data, substituting the directory you set into that remove command. The next time you start your cluster it will be brand new for more data.

Upgrading zeek2es

Most upgrades should be as simple as copying the newer zeek2es.py over the old one. In some cases, the ES ingest pipeline required for the -g command line option might change during an upgrade. Therefore, it is strongly recommend you delete your ingest pipeline before you run a new version of zeek2es.py.

ES Ingest Pipeline

If you need to delete the "zeekgeoip" ES ingest pipeline used to geolocate IP addresses with the -g command line option, you can either do it graphically through Kibana's Stack Management->Ingest Pipelines or this command will do it for you:

curl -X DELETE "localhost:9200/_ingest/pipeline/zeekgeoip?pretty"

This command is strongly recommended whenever updating your copy of zeek2es.py.

Filtering Data

Python Filters

zeek2es provides filtering capabilities for your Zeek logs before they are stored in ElasticSearch. This functionality can be enabled with the -a or -f options. The filters are constructed from Python lambda functions, where the input is a Python dictionary representing the output. You can add a filter to only store connection logs where the service field is populated using the -f option with this lambda filter file:

lambda x: 'service' in x and len(x['service']) > 0

Or maybe you'd like to filter for connections that have at least 1,024 bytes, with at least 1 byte coming from the destination:

lambda x: 'orig_ip_bytes' in x and 'resp_ip_bytes' in x and x['orig_ip_bytes'] + x['resp_ip_bytes'] > 1024 and x['resp_ip_bytes'] > 0

Simpler lambda filters can be provided on the command line via the -a option. This filter will only store connection log entries where the originator IP address is part of the 192.0.0.0/8 network:

python zeek2es.py conn.log.gz -a "lambda x: 'id.orig_h' in x and ipaddress.ip_address(x['id.orig_h']) in ipaddress.ip_network('192.0.0.0/8')"

For power users, the -f option will allow you to define a full function (instead of Python's lambda functions) so you can write functions that span multiple lines.

Filter on Keys

In some instances you might want to pull data from one log that depends on another. An example would be finding all ssl.log rows that have a uid matching previously indexed rows from conn.log, or vice versa. You can filter by importing your conn.log files with the -o uid uid.txt command line. This will log all uids that were indexed to a file named uid.txt. Then, when you import your ssl.log files you will provide the -e uid uid.txt command line. This will only import SSL rows containing uid values that are in uid.txt, previously built from our import of conn.log.

Command Line Examples

python zeek2es.py your_zeek_log.gz -i your_es_index_name

This script can be run in parallel on all connection logs, 10 at a time, with the following command:

find /some/dir -name “conn*.log.gz” | parallel -j 10 python zeek2es.py {1} :::: -

If you would like to automatically import all conn.log files as they are created in a directory, the following fswatch command will do that for you:

fswatch -m poll_monitor --event Created -r /data/logs/zeek/ | awk '/^.*\/conn.*\.log\.gz$/' | parallel -j 5 python ~/zeek2es.py {} -g -d :::: -

If you have the jq command installed you can perform searches across all your logs for a common field like connection uid, even without ElasticSearch:

find /usr/local/var/logs -name "*.log.gz" -exec python ~/Source/zeek2es/zeek2es.py {} -s -b -z \; | jq -c '. | select(.uid=="CLbPij1vThLvQ2qDKh")'

You can use much more complex jq queries than this if you are familiar with jq.

If you want to remove all of your Zeek data from ElasticSearch, this command will do it for you:

curl -X DELETE http://localhost:9200/zeek*

Since the indices have the date appended to them, you could delete Dec 31, 2021 with the following command:

curl -X DELETE http://localhost:9200/zeek_*_2021-12-31

You could delete all conn.log entries with this command:

curl -X DELETE http://localhost:9200/zeek_conn_*

Command Line Options

$ python zeek2es.py -h
usage: zeek2es.py [-h] [-i ESINDEX] [-u ESURL] [--user USER] [--passwd PASSWD]
                  [-l LINES] [-n NAME] [-k KEYWORDS [KEYWORDS ...]]
                  [-a LAMBDAFILTER] [-f FILTERFILE]
                  [-y OUTPUTFIELDS [OUTPUTFIELDS ...]] [-d DATASTREAM]
                  [--compress] [-o fieldname filename] [-e fieldname filename]
                  [-g] [-p SPLITFIELDS [SPLITFIELDS ...]] [-j] [-r] [-t] [-s]
                  [-b] [--humio HUMIO HUMIO] [-c] [-w] [-z]
                  filename

Process Zeek ASCII logs into ElasticSearch.

positional arguments:
  filename              The Zeek log in *.log or *.gz format.  Include the full path.

optional arguments:
  -h, --help            show this help message and exit
  -i ESINDEX, --esindex ESINDEX
                        The Elasticsearch index/data stream name.
  -u ESURL, --esurl ESURL
                        The Elasticsearch URL.  Use ending slash.  Use https for Elastic v8+. (default: http://localhost:9200)
  --user USER           The Elasticsearch user. (default: disabled)
  --passwd PASSWD       The Elasticsearch password. Note this will put your password in this shell history file.  (default: disabled)
  -l LINES, --lines LINES
                        Lines to buffer for RESTful operations. (default: 10,000)
  -n NAME, --name NAME  The name of the system to add to the index for uniqueness. (default: empty string)
  -k KEYWORDS [KEYWORDS ...], --keywords KEYWORDS [KEYWORDS ...]
                        A list of text fields to add a keyword subfield. (default: service)
  -a LAMBDAFILTER, --lambdafilter LAMBDAFILTER
                        A Python lambda function, when eval'd will filter your output JSON dict. (default: empty string)
  -f FILTERFILE, --filterfile FILTERFILE
                        A Python function file, when eval'd will filter your output JSON dict. (default: empty string)
  -y OUTPUTFIELDS [OUTPUTFIELDS ...], --outputfields OUTPUTFIELDS [OUTPUTFIELDS ...]
                        A list of fields to keep for the output.  Must include ts. (default: empty string)
  -d DATASTREAM, --datastream DATASTREAM
                        Instead of an index, use a data stream that will rollover at this many GB.
                        Recommended is 50 or less.  (default: 0 - disabled)
  --compress            If a datastream is used, enable best compression.
  -o fieldname filename, --logkey fieldname filename
                        A field to log to a file.  Example: uid uid.txt.  
                        Will append to the file!  Delete file before running if appending is undesired.  
                        This option can be called more than once.  (default: empty - disabled)
  -e fieldname filename, --filterkeys fieldname filename
                        A field to filter with keys from a file.  Example: uid uid.txt.  (default: empty string - disabled)
  -g, --ingestion       Use the ingestion pipeline to do things like geolocate IPs and split services.  Takes longer, but worth it.
  -p SPLITFIELDS [SPLITFIELDS ...], --splitfields SPLITFIELDS [SPLITFIELDS ...]
                        A list of additional fields to split with the ingestion pipeline, if enabled.
                        (default: empty string - disabled)
  -j, --jsonlogs        Assume input logs are JSON.
  -r, --origtime        Keep the numerical time format, not milliseconds as ES needs.
  -t, --timestamp       Keep the time in timestamp format.
  -s, --stdout          Print JSON to stdout instead of sending to Elasticsearch directly.
  -b, --nobulk          Remove the ES bulk JSON header.  Requires --stdout.
  --humio HUMIO HUMIO   First argument is the Humio URL, the second argument is the ingest token.
  -c, --cython          Use Cython execution by loading the local zeek2es.so file through an import.
                        Run python setup.py build_ext --inplace first to make your zeek2es.so file!
  -w, --hashdates       Use hashes instead of dates for the index name.
  -z, --supresswarnings
                        Supress any type of warning.  Die stoically and silently.

To delete indices:

	curl -X DELETE http://localhost:9200/zeek*?pretty

To delete data streams:

	curl -X DELETE http://localhost:9200/_data_stream/zeek*?pretty

To delete index templates:

	curl -X DELETE http://localhost:9200/_index_template/zeek*?pretty

To delete the lifecycle policy:

	curl -X DELETE http://localhost:9200/_ilm/policy/zeek-lifecycle-policy?pretty

You will need to add -k -u elastic_user:password if you are using Elastic v8+.

Requirements

  • A Unix-like environment (MacOs works!)
  • Python
    • requests Python library installed, such as with with pip.

Notes

Humio

To import your data into Humio you will need to set up a repository with the corelight-json parser. Obtain the ingest token for the repository and you can import your data with a command such as:

python3 zeek2es.py -s -b --humio http://localhost:8080 b005bf74-1ed3-4871-904f-9460a4687202 http.log 

The URL should be in the format of: http://yourserver:8080, as the rest of the path is added by the zeek2es.py script automatically for you.

JSON Log Input

Since Zeek JSON logs do not have type information like the ASCII TSV versions, only limited type information can be provided to ElasticSearch. You will notice this most for Zeek "addr" log fields that are not id$orig_h and id$resp_h, since the type information is not available to translate the field into ElasticSearch's "ip" type. Since address fields will not be of type "ip", you will not be able to use subnet searches, for example, like you could for the TSV logs. Saving Zeek logs in ASCII TSV format provides for greater long term flexibility.

Data Streams

You can use data streams instead of indices for large logs with the -d command line option. This option creates index templates beginning with zeek_. It also creates a lifecycle policy named zeek-lifecycle-policy. If you would like to delete all of your data streams, lifecycle policies, and index templates, these commands will do it for you:

curl -X DELETE http://localhost:9200/_data_stream/zeek*?pretty
curl -X DELETE http://localhost:9200/_index_template/zeek*?pretty
curl -X DELETE http://localhost:9200/_ilm/policy/zeek-lifecycle-policy?pretty

Helper Scripts

There are two scripts that will help you make your logs into data streams such as logs-zeek-conn. The first script is process_logs_as_datastream.sh and given a list of logs and directories, will import them as such. The second script is process_log.sh, and it can be used to import logs one at a time. This script can also be used to monitor logs created in a directory with fswatch. Both scripts have example command lines if you run them without any parameters.

$ ./process_logs_as_datastream.sh 
Usage: ./process_logs_as_datastream.sh NJOBS "ADDITIONAL_ARGS_TO_ZEEK2ES" "LIST_OF_LOGS_DELIMITED_BY_SPACES" DIR1 DIR2 ...

Example:
  time ./process_logs_as_datastream.sh 16 "" "amqp bgp conn dce_rpc dhcp dns dpd files ftp http ipsec irc kerberos modbus modbus_register_change mount mqtt mysql nfs notice ntlm ntp ospf portmap radius reporter rdp rfb rip ripng sip smb_cmd smb_files smb_mapping smtp snmp socks ssh ssl stun syslog tunnel vpn weird wireguard x509" /usr/local/var/logs
$ ./process_log.sh 
Usage: ./process_log.sh LOGFILENAME "ADDITIONAL_ARGS_TO_ZEEK2ES"

Example:
  fswatch -m poll_monitor --event Created -r /data/logs/zeek |  awk '/^.*\/(conn|dns|http)\..*\.log\.gz$/' | parallel -j 16 ./process_log.sh {} "" :::: -

You will need to edit these scripts and command lines according to your environment.

Any files having a name of a log such as conn_filter.txt in the lambda_filter_file_dir, by default your home directory, will be applied as a lambda filter file to the corresponding log input. This allows you to set up all of your filters in one directory and import multiple log files with that set of filters in one command with process_logs_as_datastream.sh.

The following lines should delete all Zeek data in ElasticSearch no matter if you use indices or data streams, or these helper scripts:

curl -X DELETE http://localhost:9200/zeek*?pretty
curl -X DELETE http://localhost:9200/_data_stream/zeek*?pretty
curl -X DELETE http://localhost:9200/_data_stream/logs-zeek*?pretty
curl -X DELETE http://localhost:9200/_index_template/zeek*?pretty
curl -X DELETE http://localhost:9200/_index_template/logs-zeek*?pretty
curl -X DELETE http://localhost:9200/_ilm/policy/zeek-lifecycle-policy?pretty

... or if using Elastic v8+ ...

curl -X DELETE -k -u elastic:password https://localhost:9200/zeek*?pretty
curl -X DELETE -k -u elastic:password https://localhost:9200/_data_stream/zeek*?pretty
curl -X DELETE -k -u elastic:password https://localhost:9200/_data_stream/logs-zeek*?pretty
curl -X DELETE -k -u elastic:password https://localhost:9200/_index_template/zeek*?pretty
curl -X DELETE -k -u elastic:password https://localhost:9200/_index_template/logs-zeek*?pretty
curl -X DELETE -k -u elastic:password https://localhost:9200/_ilm/policy/zeek-lifecycle-policy?pretty

But to be able to do this in v8+ you will need to configure Elastic as described in the section Elastic v8.0+.

Cython

If you'd like to try Cython, you must run python setup.py build_ext --inplace first to generate your compiled file. You must do this every time you update zeek2es!

About

A Python application to filter and transfer Zeek logs to Elastic/OpenSearch+Humio. This app can also output pure JSON logs to stdout for further processing!

Topics

Resources

License

Stars

Watchers

Forks

Packages

No packages published