At times the variable directory (also known as vardir and represented by the environment variable $SP_SYS_VARDIR) starts filling up fast and if space utilization reaches a high value of say, 90% or up, the issue needs to be looked at immediately.
Not doing so can fill a disk and cause queue corruption with potential data loss.
Factors responsible for this typically include a problem with SharePlex processes like Capture, Read, Export or Post.
Sometimes SharePlex on Source is brought down for a long period and upon restart it starts processing from the archive log it last processed.
And depending on how far behind it is vis-à-vis Oracle and/or whether supplemental logging is enabled on Source, the messages do not move beyond Capture until Capture catches up with Oracle.
At other times, it is just a huge batch job that can result in Capture falling behind.
Or it can be slow Post or issues with the Post process that cause messages to accumulate in the Post queue.
Until the underlying issues are resolved, the messages stay in the affected queue(s) and can cause the variable directory to fill up to precarious levels.