Used by the UI and API to hint at a user that not every event will be attached to a case. Supports values up to 10,000 (the default limit on the number of documents returned by a single ES search).
To accomodate the config screen, the annotation now specifies it as a multiline string with a yaml syntax. The user can edit the yaml to add or remove queries. The UI will parse the YAML before use.
Also updated the IntegrityFailure queries to specify table columns more relevant to a sync failure than the default ones.
Each engine pulls the same repo into the same location and shows the summaries.
Which repo and where to keep them is advanced, but turning AI summaries on or off is not.
The so-detection index needs it's refresh_interval reset during an upgrade. If the index doesn't exist, the config change will set it correctly when it is created.
`maxScrollSize` defines the "page size" of each scroll request.
`bulkIndexerWorkerCount` defines how many worker threads a bulk indexer should use. 0 or fewer indicates that 1 thread per CPU core should be used.
New fields have been added to manager and then duplicated over to SOC's config in the same vein as how proxy was updated earlier this week.
AdditionalCA holds the PEM formatted public keys that should be trusted when making requests. It has been implemented for both Sigma's zip downloads and Sigma and Suricata's repository clones and pulls.
InsecureSkipVerify has been added to help our users troubleshoot their configuration. Setting it to true will not verify the cert on outgoing requests. Self signed, missing, or invalid certs will not throw an error.
If a sync errors out, the engine will wait `communityRulesImportErrorSeconds` seconds instead of the usual `communityRulesImportFrequencySeconds` seconds wait.
If `failAfterConsecutiveErrorCount` errors happen in a row when syncing detections to ElasticSearch then the sync is considered a failure and will give up and try again later. This assumes ElasticSearch is the source of the errors and backs of in hopes it'll be able to fix itself.
Each detection engine is getting a state file to help manage the timer over restarts. By default, the files will go in soc's config folder inside a fingerprints folder.
Removed some strelka/yara rules from salt.
Removed yara scripts for downloading and updating rules. This will be managed by SOC.
Added a new compile_yara.py script.
Added the strelka repos folder.
Detection mappings updated to include the removal of Note and the addition of Tags, Ruleset, and Language.
SOC defaults updated to use language based queries rather than engine and show the language column instead of the engine column in results.
Removed some strelka/yara rules from salt.
Removed yara scripts for downloading and updating rules. This will be managed by SOC.
Added a new compile_yara.py script.
Added the strelka repos folder.
Salt is getting a second timeout for operations known to take a long time such as sending and importing files. There's also an entry in soc_soc.yaml so the value can be changed in SOC's config page.
When salt-cp runs it's course and finds it can't send a file, it outputs a report saying as much but the exit code will be zero. Now we remove the filename and node from the response and look for `True` to know if it succeeded. Also, respect the cleanup flag on success or failure.
Check the status of the decryption process before importing.
No longer decrypt locally, issue salt command for the remote client to do the decrypting.
Using the new --json flag is not only more reliable than using a regex, the way the import script was written even re-imports will provide a url. This means that in more cases we can provide the results to the users (even if nothing changed).
sed doesn't remove ALL whitespace, only newlines. It's better to stop at the first whitespace than to stop at a particular, maybe-not-last query string parameter.
This folder is where a manager will initially store uploaded PCAP/EVTX files before sending to sensors. Sensors will store uploads in this folder on their own system.