This component acts as a bridge between Spark and Vertica, allowing the user to either retrieve data from Vertica for processing in Spark, or store processed data from Spark into Vertica.
Apache License 2.0
20
stars
23
forks
source link
[BUG] Spark jobs should only select UP nodes in current subcluster at the time the job starts #551
This is an issue emerging with v12 Eon mode and sandbox support. Current behavior appears to be to select all nodes; this doesn't work with sandboxing, where some nodes can't write to the main communal storage, and also may break during dynamic scaling, especially when a subcluster shrinks, which means some workers will point to a down node. The correct behavior is to check for Eon mode when the job starts, then select UP nodes in the current subcluster for workers to connect to.
This is an issue emerging with v12 Eon mode and sandbox support. Current behavior appears to be to select all nodes; this doesn't work with sandboxing, where some nodes can't write to the main communal storage, and also may break during dynamic scaling, especially when a subcluster shrinks, which means some workers will point to a down node. The correct behavior is to check for Eon mode when the job starts, then select UP nodes in the current subcluster for workers to connect to.