DC Disaster recovery ignores index snapshots when Snappy or tarball archive is used

Problem

During disaster recovery, the pattern used for identifying index snapshot files matches only *.zip files. This pattern should include .tar and .tar.sz. Snappy and tarball archive was introduced in Jira 7.12 and users can choose the type of archiving used for creating snapshots. Currently, Snappy is the default option. This will produce snapshots that cannot be used by the disaster recovery mechanism. Since disaster recovery is a feature of Jira DC, this is a regression.

Expected behaviour

Disaster recovery mechanism should be able to use any index snapshot created by Jira itself.

Workaround

In a case where the only available snapshot is a tarball or Snappy archive, unarchiving and compressing it to a zip archive will work. However, if the snapshot size is large, which is expected in Jira DC, this method most likely wouldn't be able to satisfy the high availability requirements.

Notes

See  com.atlassian.jira.index.ha.DisasterRecoveryLauncher#restoreIndex and com.atlassian.jira.index.ha.IndexUtils#INDEX_SNAPSHOT_PATTERN