Details
-
Question
-
Status: Closed
-
Major
-
Resolution: Invalid
-
3.0.0
-
None
-
Important
Description
Hi Spark team,
We are using spark 3.0.0 on AWS EMR service to run our spark jobs.
spark-core_2.12:3.0.0 has transitive dependency on commons-text 1.6 and this is flagged as critical severity CVE-2022-42889.
As per Jira SPARK-40801 , commons text has been upgraded and spark 3.4.0 is released.
We are dependent on AWS EMR service and changing EMR version and spark version is big task for us considering all downstream dependent applications
We know spark 3.0.0 is EOL for you but would really appreciate if could provide guidance on it.
We have few queries and need inputs from spark dev team to handle this issue on priority at our end
- Does spark-core 3.0.0 use StringSubstitutor API and do we need to worry about this?
- which lib or code within spark core 3.0.0 triggers StringSubstitutor method ?
- I searched for spark source code for usage of StringSubstitutor and found one reference here https://github.com/apache/spark/blob/master/core/src/main/scala/org/apache/spark/ErrorClassesJSONReader.scala in master branch but this class is not available in spark 3.0.0 tags. As per link - https://blogs.apache.org/security/entry/cve-2022-42889 , If you rely on software that uses a version of commons-text prior to 1.10.0, you are likely still not vulnerable: you are only affected when this software uses the StringSubstitutor API without properly sanitizing any untrusted input.
- Please confirm if spark 3.0.0 does not use StringSubstitutor API from commons-text but just have dependency marked in POM https://github.com/apache/spark/blob/3fdfce3120f307147244e5eaf46d61419a723d50/pom.xml#L506 for other API use from commons-text?
- in case , can we include the apache commons text 1.10.0 as explicit dependency on our applications POMs and add common text 1.6 in exclusions for spark-core , will it work ?