New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 718592 link

Starred by 1 user

Issue metadata

Status: Untriaged
Owner: ----
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Chrome
Pri: 2
Type: Bug

Blocking:
issue 711158



Sign in to add a comment

Not enough free space on /usr/local/autotest | intermittently on gale/jetstream/related

Project Member Reported by jen...@chromium.org, May 4 2017

Issue description

https://uberchromegw.corp.google.com/i/chromeos/builders/veyron_minnie-chrome-pfq/builds/1098

Failed to install image during autoupdate test, due to not enough disk space error. "Not enough free space on /usr/local/autotest - 0.607GB free, want 0.700GB".

05/04 10:51:34.307 DEBUG|          ssh_host:0284| Running (ssh) 'test -x /usr/local/autotest/bin/autotest'
05/04 10:51:34.694 DEBUG|          ssh_host:0284| Running (ssh) 'test -w /usr/local/autotest'
05/04 10:51:35.084 DEBUG|          autotest:0119| Found existing autodir at /usr/local/autotest
05/04 10:51:35.085 INFO |      base_classes:0352| Checking for >= 0.7 GB of space under /usr/local/autotest on machine chromeos4-row9-rack9-host5
05/04 10:51:35.085 DEBUG|          ssh_host:0284| Running (ssh) 'df -PB 1000000 /usr/local/autotest | tail -1'
05/04 10:51:35.476 DEBUG|             utils:0297| [stdout] /dev/mmcblk0p1          11057  9869       607      95% /usr/local
05/04 10:51:35.518 ERROR|            repair:0332| Failed: Legacy host verification checks
Traceback (most recent call last):
  File "/usr/local/autotest/client/common_lib/hosts/repair.py", line 329, in _verify_host
    self.verify(host)
  File "/usr/local/autotest/server/hosts/repair.py", line 55, in verify
    host.verify_software()
  File "/usr/local/autotest/server/hosts/cros_host.py", line 1481, in verify_software
    super(CrosHost, self).verify_software()
  File "/usr/local/autotest/server/hosts/abstract_ssh.py", line 722, in verify_software
    self.AUTOTEST_GB_DISKSPACE_REQUIRED)
  File "/usr/local/autotest/client/common_lib/hosts/base_classes.py", line 356, in check_diskspace
    raise error.AutoservDiskFullHostError(path, gb, free_space_gb)
AutoservDiskFullHostError: Not enough free space on /usr/local/autotest - 0.607GB free, want 0.700GB
05/04 10:51:35.519 INFO |        server_job:0184| 	FAIL	----	verify.cros	timestamp=1493920295	localtime=May 04 10:51:35	Not enough free space on /usr/local/autotest - 0.607GB free, want 0.700GB
 
Blocking: 711158
Project Member

Comment 2 by sheriffbot@chromium.org, Aug 24 2017

Labels: Hotlist-Recharge-BouncingOwner
Owner: ----
The assigned owner "akeel@chromium.org" is not able to receive e-mails, please re-triage.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot
Cc: seobrien@chromium.org yueherngl@chromium.org shuqianz@chromium.org josephsih@chromium.org
This happened again in the latest run of gale-paladin:

https://uberchromegw.corp.google.com/i/chromeos/builders/gale-paladin/builds/3882
Owner: shuqianz@chromium.org
Status: Started (was: Untriaged)
will take a look
Cc: -yueherngl@chromium.org jrbarnette@chromium.org
Labels: -Pri-3 Pri-2
I login to the server, and run the following command

root@chromeos6-row21-jetstream-host3 $ df -PB 1000000 /usr/local/autotest | tail -1
/dev/mmcblk0p1           2054   325      1607      17% /usr/local

The /usr/local only consumes 17%, not the percentage shown in the task. During the test runtime, is there anything under autotest consuming the disk? Richard?
> I login to the server, and run the following command

The failure is a verify failure on the DUT.  The shortage
of space is on the DUT.

Sorry, I man I login to the DUT, the output is from the DUT
> Sorry, I man I login to the DUT, the output is from the DUT

Oh, so it is.  Having enough space on the DUT is actually normal
after the failure.  Here's the relevant history:
    2017-09-15 12:11:50  OK http://cautotest/tko/retrieve_logs.cgi?job=/results/hosts/chromeos6-row21-jetstream-host3/61405945-repair/
    2017-09-15 11:51:11  -- http://cautotest/tko/retrieve_logs.cgi?job=/results/hosts/chromeos6-row21-jetstream-host3/61405793-provision/
    2017-09-15 09:48:56  -- http://cautotest/tko/retrieve_logs.cgi?job=/results/142270284-chromeos-test/
    2017-09-15 09:37:40  OK http://cautotest/tko/retrieve_logs.cgi?job=/results/hosts/chromeos6-row21-jetstream-host3/61404918-provision/

So, the provision failure triggered repair, and repair fixed the problem.
If you look in the repair logs, you'll see that the space check verifier
failed repeatedly, until finally repair powerwashed the DUT and
re-installed the software.

hmm, what caused this issue? and how to avoid it?
> hmm, what caused this issue? and how to avoid it?

Basically, something in stateful got too big.  A few points of note:
  * This is jetstream, which has a much smaller stateful than a
    typical chromebook.
  * The stuff that takes up space in the results directory can be
    seen by looking at "result_summary.html" in the provision logs.
    Some of that stuff likely contributed to the overflow.
  * Stateful includes a lot of software packages specific to test
    images.  If someone's recently added new packages to test images
    (or if package upgrades made stuff bigger) that will cause
    trouble.

Owner: ----
Status: Untriaged (was: Started)
Cc: -seobrien@chromium.org
Summary: Not enough free space on /usr/local/autotest | intermittently on gale/jetstream/related (was: Not enough free space on /usr/local/autotest)
Happened on gale-paladin. OP was about a different board, but that was a while ago. Repurposing this bug to be about gale/jetstream related siblings.

https://luci-milo.appspot.com/buildbot/chromeos/gale-paladin/3922
Cc: ra...@google.com
Labels: -Build-PFQ-Failures -Hotlist-Recharge-BouncingOwner

Sign in to add a comment