Example: https://chromium-review.googlesource.com/c/373164/

Error submitting change: 
BatchRefUpdate failed: TransitionalBatchRefUpdate[
  CREATE: 0000000000000000000000000000000000000000 bd229aac3baee3dab6299ce17a17c8a6a85fa528 refs/changes/64/373164/2  (REJECTED_OTHER_REASON: transaction aborted)
  UPDATE: ad686d6aa094591806045160d4243623da8cf5a7 bd229aac3baee3dab6299ce17a17c8a6a85fa528 refs/heads/master  (LOCK_FAILURE)
]

Please, unvote/vote on Commit Queue label to re-trigger on the same patchset.
Bot data: {"action": "cancel", "triggered_at": "2016-08-19T18:29:36.0Z", "revision": "7e70ec454f6a8b86995f0a358fafc1eb1565b56c"}

As far as I can tell, this is a retryable error. 

Additional context:

The tree was closed for ~50 minutes, and I suspect that multiple people had changes queued up. I investigated the closure (flake), and opened the tree. As soon as that happened 6 new changes were immediately committed by the CQ (I assume that 6 people didn't just run `git cl land` at exactly the same time). At the same time CQ rejected my change with the error above (I suspect that other changes at the same time may have failed for similar reasons, but cannot confirm).

I suspect this may have to do with googlesource's eventual consistency guarantees, and a retry loop with some backoff would have been able to submit this without issue.