tree f10466d35306e2ae0a9f06e44ca079d7d8063941
parent fdf862eacb58b233abe2f98b60b467457e243176
author Abdullah Alamoudi <bamousaa@gmail.com> 1519230039 -0800
committer abdullah alamoudi <bamousaa@gmail.com> 1519235589 -0800

[NO ISSUE][TXN] Prevent deadlock in Metadata transactions

- user model changes: no
- storage format changes: no
- interface changes: no

Details:
- Flushes in metadata datasets are triggerred by entity update
  logs, unlike regular transactions where flushes are triggerred
  by entity commit logs.
- Because entity update logs can be writting to disk before the
  operation completes, there is a chance that an operation that
  caused the component to be full exits after the log is flushed
  and so, a flush operation is not scheduled.
- This change proposes a simple fix. The fix is that metadata
  operation will also check if a flush is needed and will schedule
  one if needed.

Change-Id: I07a18840dc54fe052b7bd294595f816f6d8a4d2f
Reviewed-on: https://asterix-gerrit.ics.uci.edu/2413
Sonar-Qube: Jenkins <jenkins@fulliautomatix.ics.uci.edu>
Reviewed-by: Murtadha Hubail <mhubail@apache.org>
Tested-by: Jenkins <jenkins@fulliautomatix.ics.uci.edu>
Contrib: Jenkins <jenkins@fulliautomatix.ics.uci.edu>
Integration-Tests: Jenkins <jenkins@fulliautomatix.ics.uci.edu>