Looks like there's two problems here:
- sql_unittests failed but didn't fail the build
- the original swarming task that all the failures were deduped from failed to dump the test results to the output file: https://chromium-swarm.appspot.com/task?id=3a1f8bed11419110

I think the "error: cannot open '::/output.json'" is the problem here. Since it's not currently failing, moving this out of the trooper queue.

+scottmg re cannot open ::/output.json

Bah, it's just running

https://cs.chromium.org/chromium/src/build/fuchsia/runner_common.py?rcl=682b6703be24b401ac1ea660c7363271a789c5f2&l=552

which is supposed to be extracting the output.json from a .minfs image to the local disk.

It looks like something might be failing on setup:

[00002.184] 02539.02620> object 0xffffff80002d4098 base 0xb0018000 size 0x1000 ref 1
[00002.187] 01280.01341> vc: new input device /dev/class/input/000
[00002.266] 01280.01341> vc: new input device /dev/class/input/001
[00002.270] 02661.02758> object 0xffffff80002dff28 base 0xb00fa000 size 0x1000 ref 1
[00002.472] 02735.02780> netstack: started
[00002.512] 02735.02780> netstack: socket dispatcher started
[00002.565] 02735.02780> netstack: watching for ethernet devices
[00003.028] 02735.03417> netstack: linkaddr: 52:54:00:63:5e:7b
[00003.045] 02735.03417> netstack: NIC 2 added using ethernet device "/dev/class/ethernet/000"
[00003.045] 02735.03417> netstack: NIC 2: ipv6addr: fe80::5054:ff:fe63:5e7b
[00003.093] 02735.03417> netstack: NIC 2: DHCP acquired IP 192.168.3.9 for 24h0m0s
[00003.103] 02735.03417> netstack: NIC 2: DNS servers: [192.168.3.3]
[00003.216] 03579.03606> Error opening block device
[00003.217] 01514.01603> Executing sql_unittests --test-launcher-retry-limit=0 --test-launcher-jobs=4 --test-launcher-summary-output=/volume/results/output.json
[00003.228] 02539.03348> sata0: dev info
[00003.228] 02539.03348>   serial=MQ0010 3
[00003.229] 02539.03348>   firmware rev=.2+5
[00003.229] 02539.03348>   model id=EQUMH RADDSI K
[00003.229] 02539.03348>   major=0xf0 ATA/ATAPI DMA 32 commands
[00003.229] 02539.03348>   LBA48 104857600 sectors, size=512
[00003.251] 03659.03685> IMPORTANT DEBUGGING NOTE: batches of tests are run inside their
[00003.251] 03659.03685> own process. For debugging a test inside a debugger, use the
[00003.251] 03659.03685> --gtest_filter=<your_test_name> flag along with
[00003.252] 03659.03685> --single-process-tests.
[00003.252] 03659.03685> Using sharding settings from environment. This is shard 0/1
[00003.252] 03659.03685> Using 4 parallel jobs.
[00003.298] 01101.01117> devmgr: new block device: /dev/class/block/000
[00003.314] 01101.01117> devmgr: minfs
[00009.411] 03659.03685> [1/83] SQLConnectionTest.Execute (339 ms)
[00009.412] 03659.03685> [2/83] SQLConnectionTest.ExecuteWithErrorCode (280 ms)
...

in particular "Error opening block device".

Sean/James: any wild guesses what might be going wrong there?

Oh, now that I re-read that log a few more times. I think what's happening is that we're probably trying to mount too soon and devmgr hasn't discovered the block device yet, so fails at t=3.216, and then does, and it's available at t=3.298s. So I just need to increase the msleep by 3.298-3.216 seconds! :/

Yeah, sorry about that. Increasing the sleep time is gross, but from the filesystem's perspective, it hasn't even started until devmgr sorts out what block devices are available and kicks off the MinFS process.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/b6476ac34a7f5473dc4a3909b6e76a5a4dd8802b

commit b6476ac34a7f5473dc4a3909b6e76a5a4dd8802b
Author: Scott Graham <scottmg@chromium.org>
Date: Wed Nov 29 20:22:25 2017

fuchsia: Increase msleep before trying to mount results block device

Fixed. Once and for all.

I SAID FIXED!

Bug:  789473 
Change-Id: I5f41023febd34c5e51ddbed9d668c505dedaed3f
Reviewed-on: https://chromium-review.googlesource.com/797612
Commit-Queue: Scott Graham <scottmg@chromium.org>
Commit-Queue: Wez <wez@chromium.org>
Reviewed-by: Wez <wez@chromium.org>
Cr-Commit-Position: refs/heads/master@{#520221}
[modify] https://crrev.com/b6476ac34a7f5473dc4a3909b6e76a5a4dd8802b/build/fuchsia/runner_common.py

Should maybe possibly probably be better now.