tree 49512f3b80021f552560e739169ce35355115377
parent d42b3255a6372f74c0fa75fa3f922009afbff459
author Prathmesh Prabhu <pprabhu@chromium.org> 1488237900 -0800
committer Prathmesh Prabhu <pprabhu@chromium.org> 1488240869 +0000

Revert "[autotest] Remove synch_id from database"

This reverts commit eb97ee7fffb4a2ef3155b07f2aa71f04d574b642.
But not quite. We never revert DB migration scripts, especially if
they've been run on any servers. So, this CL adds a new migration (111)
to revert the migration (110) that was already applied to the test
servers.
To reland, remove synch_id once again via a new migration.

Reason for revert:
This CL requires lab downtime for push-to-prod because host_scheduler
can not be cleanly restarted across this CL.

To push this CL to lab, one must:
- close the lab.
- migrate DB to the new schema (host_scheduler starts failing)
- push the CL to lab (host_scheduler recovers)
- open the lab.

Now is not the time to do this dance, so reverting with the hope to
reland at a better time.

BUG=chromium:696691
TEST=test_push host_scheduler can start with tip-of-tree code and with
tip-of-prod after applying both migration 110 and 111.

Change-Id: If2f5428f100af400d498eded9dfb08d6f22aa4dc
Reviewed-on: https://chromium-review.googlesource.com/447200
Tested-by: Prathmesh Prabhu <pprabhu@chromium.org>
Reviewed-by: Dan Shi <dshi@google.com>