mysqldump backup do not report failure

[Yang Youseok]

When mysqldump strategy failed by any reasons, trove-guestagent do not report the failure. Different from other backup strategies which has failure check logic (check_process), mysqldump itself depends only on base.py raise_if_process_error().

But the problem is that stderr never catched as expected since there is redirect part in command (mysqldump ... >/tmp/mysqldump.log). If error occurred when sqldump command executed, all stderr messages are redirected to the log file and it makes success state for the backup.

version: newton-eol, seems to be same as version above.
How to reproduce: add invalid extra-opts for sqldump. In my case, I added --single-transaction with no binlog.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/549773

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/549773
Committed: https://git.openstack.org/cgit/openstack/trove/commit/?id=49fea478ee6d2fe67f6c248b8da9b0ce910184f8
Submitter: Zuul
Branch: master

commit 49fea478ee6d2fe67f6c248b8da9b0ce910184f8
Author: Yang Youseok <email address hidden>
Date: Mon Mar 5 23:16:33 2018 +0900

    Fix false-negative failure report for mysqldump backup

    If mysqldump command has redirect part for stderr, catching error logic
    by stderr doest not work as expected. Remove redirect part to report
    BackupError normally for sqldump failure.

    Change-Id: I2c10ae7b6b6094af7819594e889385b0ed77fe4b
    Close-Bug: 1753498