Improve DGA/LotL doc's Installation Procedures - #10476
Conversation
1 - Noted that one of the listed ingest pipeline methods should be used 2 - added the missing mapping that is required for the ML job card to appear as an option 3 - added optional step of creating a data view that is specific to the windows process data (rather than selecting `logs-*` which hits impertinent indices) 4 - noted that the ML job card can also be missing if the mapping is not in place
|
@rseldner Thank you for your contribution to this documentation! I do not believe the index mapping changes are still necessary as they now are handled in custom.yml Also, when would a user need to use the Rollover API? |
|
I found the mapping was needed while assisting a user in troubleshooting the missing ML job card despite adding the ingest pipeline and confirming the problemchild fields had been enriched. I reproduced this myself and can demonstrate. I will reach out internally. The rollover would be needed after defining the mapping in the index/component template so that the subsequent index picks up the mapping. |
|
@rseldner Can you confirm the version of Kibana? |
|
User was on Elastic's Govcloud |
|
Please also add what data from what Agent is required. I once lost time testing this while it needs Agent events to work? Or has that changed since |
|
Thank you for the PR, @jmcarlock on our team has been investigating some related details, and will provide more feedback + keep you posted |
|
@rseldner , I changed the instructions to runtime mappings, which doesn't require a rollover. Can you give it one more test on your end? We also needed to make the changes for DGA, which had the same issue. I also bumped the version number, added a changelog entry, and tested both with elastic-package and with cloud deployments. |
sodhikirti07
left a comment
There was a problem hiding this comment.
New instructions to add Mappings lgtm!
|
At least in the current state where these fields are not getting mapped, We are missing the point that if the mappings do not exist (runtime or not), they need to be added to the relevant index/component template. Doing Adding runtime mappings to an index do not get around this and do not quite make sense to me (if we are adding a mapping to an index, why make it a runtime mapping? We prefer query overhead v.s. storage overhead?) Unless you meant to instead apply a runtime field on the data view to emit these field values from Sidenote, shouldn't |
per @rseldner, will require additional changes
|
@rseldner , ok, I understand! Thank you for the explanation. I will change the method back to your suggestions and test again. Will also change |
|
Previews of the documentation changes can be viewed at: |
|
Thank you for the review, @susan-shu-c ! |
💚 Build Succeeded
History
cc @jmcarlock |
|
sodhikirti07
left a comment
There was a problem hiding this comment.
@jmcarlock Nicely done! LGTM!
susan-shu-c
left a comment
There was a problem hiding this comment.
Thanks all for the contributions, and Gus for the research and testing!
|
Package dga - 2.0.3 containing this change is available at https://epr.elastic.co/search?package=dga |
|
Package problemchild - 2.1.4 containing this change is available at https://epr.elastic.co/search?package=problemchild |
|
☝️ ➕ Thanks for your attention on this @jmcarlock ! |
* Update README.md 1 - Noted that one of the listed ingest pipeline methods should be used 2 - added the missing mapping that is required for the ML job card to appear as an option 3 - added optional step of creating a data view that is specific to the windows process data (rather than selecting `logs-*` which hits impertinent indices) 4 - noted that the ML job card can also be missing if the mapping is not in place * remove non-functional field mappings * add correct index name for elastic defend and remove windows agent reference * add runtime mapping install for DGA * change lotl instructions to use runtime mappings * add changelog, bump manifest version * update lotl docs * small fixes * add dga instructions * fix wording * finish network log sentence * fix field mappings in dga * add custom pipeline instructions * formatting fixes * formatting try 2 * formatting fix 3 * formatting fix 4 * fix typo * fix tab indentation for numbering * clarify UI vs instruction steps * remove problemchild reference from dga --------- Co-authored-by: Gus Carlock <10844131+jmcarlock@users.noreply.github.com>
* Update README.md 1 - Noted that one of the listed ingest pipeline methods should be used 2 - added the missing mapping that is required for the ML job card to appear as an option 3 - added optional step of creating a data view that is specific to the windows process data (rather than selecting `logs-*` which hits impertinent indices) 4 - noted that the ML job card can also be missing if the mapping is not in place * remove non-functional field mappings * add correct index name for elastic defend and remove windows agent reference * add runtime mapping install for DGA * change lotl instructions to use runtime mappings * add changelog, bump manifest version * update lotl docs * small fixes * add dga instructions * fix wording * finish network log sentence * fix field mappings in dga * add custom pipeline instructions * formatting fixes * formatting try 2 * formatting fix 3 * formatting fix 4 * fix typo * fix tab indentation for numbering * clarify UI vs instruction steps * remove problemchild reference from dga --------- Co-authored-by: Gus Carlock <10844131+jmcarlock@users.noreply.github.com>




Proposed commit message
The Living off the Land integration's installation procedure could use these improvements:
logs-*which hits impertinent indices)Ask for Reviewers: