Issue 455749 in chromium: "your profile could not be opened correctly" after latest update

[chro...@googlecode.com]

Status: Untriaged
Owner: ----
Labels: Type-Bug Pri-2 Cr-OS-Systems OS-Chrome

New issue 455749 by semenz...@chromium.org: "your profile could not be
opened correctly" after latest update
https://code.google.com/p/chromium/issues/detail?id=455749

Customer reports this:


I'm now getting this pop-up when I log into the Chromebook:
Your profile could not be opened correctly.
Some features may be unavailable (or something like that).
I can't screen-shot it because it takes over the screen.
This started after an update was installed this A.M.
Fearing some other problem, I changed my Google password.
I can still log in and get access to everything (apparently), but even
after multiple logins (power-downs) I still get that pop-up (multiple times
-- perhaps once per tab?) at the start.

Any clues?

--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings

[chro...@googlecode.com]

Comment #1 on issue 455749 by semenz...@chromium.org: "your profile could

not be opened correctly" after latest update
https://code.google.com/p/chromium/issues/detail?id=455749

Version 40.0.2214.93 (64-bit)
Platform 6457.83.0 (Official Build) stable-channel swanky
Firmware Google_Swanky.5216.238.5

Also reports this:

And when I first log in after power on, the lower right corner has
the "account details out of date -- please sign in again" thing (plus the
other pop-ups).
But when I do sign out and sign in again, I get the same thing with the
other pop-ups (without the "account details out of date").

[chro...@googlecode.com]

Updates:
Cc: etco...@gmail.com

Comment #2 on issue 455749 by semenz...@chromium.org: "your profile could

not be opened correctly" after latest update
https://code.google.com/p/chromium/issues/detail?id=455749

(No comment was entered for this change.)

[chro...@googlecode.com]

Updates:
Cc: dva...@chromium.org mler...@chromium.org

Comment #3 on issue 455749 by sad...@chromium.org: "your profile could not

be opened correctly" after latest update
https://code.google.com/p/chromium/issues/detail?id=455749

Possibly a dup of issue 449740 and/or issue 465218, although those two
other issues are on Linux.

[chro...@googlecode.com]

Comment #4 on issue 455749 by Dechamel...@gmail.com: "your profile could

not be opened correctly" after latest update
https://code.google.com/p/chromium/issues/detail?id=455749

Also experiencing this. Major changes that may or may not be related to
this error message for me:
I am unable to download files from the internet.
The websites I visit frequently have lost their autofill.

[chro...@googlecode.com]

Comment #5 on issue 455749 by semenz...@chromium.org: "your profile could

not be opened correctly" after latest update
https://code.google.com/p/chromium/issues/detail?id=455749

Happened again to my friend.

https://feedback.corp.google.com/product/208/neutron?lView=rd&lRSort=1&lROrder=2&lRFilter=1&lReportSearch=@semenzato&lReport=6302778028

The log has a lot of errors like these:

[722:722:0927/203022:ERROR:logging.h(779)] Failed to call method:
org.freedesktop.DBus.ObjectManager.GetManagedObjects: object_path=
/org/chromium/peerd: org.freedesktop.DBus.Error.ServiceUnknown: The name
org.chromium.peerd was not provided by any .service files
[722:722:0927/203022:WARNING:object_manager.cc(361)] org.chromium.peerd
/org/chromium/peerd: Failed to get managed objects
[722:722:0927/203022:ERROR:logging.h(779)] Failed to call method:
org.freedesktop.DBus.ObjectManager.GetManagedObjects: object_path=
/org/chromium/privetd: org.freedesktop.DBus.Error.ServiceUnknown: The name
org.chromium.privetd was not provided by any .service files
[722:722:0927/203022:WARNING:object_manager.cc(361)] org.chromium.privetd
/org/chromium/privetd: Failed to get managed objects
[722:722:0927/203022:VERBOSE1:display_configurator.cc(829)]
SetDisplayPower: power_state=ALL_ON flags=0, configure timer=Stopped
[722:722:0927/203022:ERROR:logging.h(779)] Failed to call method:
org.chromium.CryptohomeInterface.TpmIsOwned: object_path=
/org/chromium/Cryptohome: org.freedesktop.DBus.Error.ServiceUnknown: The
name org.chromium.Cryptohome was not provided by any .service files

[chro...@googlecode.com]

Updates:
Cc: dkr...@chromium.org

Comment #6 on issue 455749 by semenz...@chromium.org: "your profile could

not be opened correctly" after latest update
https://code.google.com/p/chromium/issues/detail?id=455749

The kernel log shows that the TPM is working happily.

Any idea Darren?

[chro...@googlecode.com]

Comment #7 on issue 455749 by dkr...@chromium.org: "your profile could not

be opened correctly" after latest update
https://code.google.com/p/chromium/issues/detail?id=455749

No idea -- cryptohome looks happy.

[chro...@googlecode.com]

Updates:
Labels: Cr-UI-Browser-Profiles

Comment #9 on issue 455749 by dkr...@chromium.org: "your profile could not

be opened correctly" after latest update
https://code.google.com/p/chromium/issues/detail?id=455749

(No comment was entered for this change.)

[chro...@googlecode.com]

Updates:
Cc: gwen...@chromium.org

Comment #12 on issue 455749 by semenz...@chromium.org: "your profile could

not be opened correctly" after latest update
https://code.google.com/p/chromium/issues/detail?id=455749

But wouldn't a failing SSD create other problems as well? Not just this
one, repeatedly? Also, Gwendal, would the driver output anything to the
log?

Here's a feedback report:

https://feedback.corp.google.com/product/208/neutron?lView=rd&lReport=6609858684

[chro...@googlecode.com]

Comment #13 on issue 455749 by semenz...@chromium.org: "your profile could

not be opened correctly" after latest update
https://code.google.com/p/chromium/issues/detail?id=455749

There are many of these in the log.

2015-11-25T06:45:09.266744-08:00 DEBUG kernel: [ 32.770335] Valid
eCryptfs headers not found in file header region or xattr region, inode
261831

I searched for those errors and found this bug:

https://code.google.com/p/chromium/issues/detail?id=217829

which I think caused by a full disk.

Earl is it possible that your disk is full?

[chro...@googlecode.com]

Comment #14 on issue 455749 by semenz...@chromium.org: "your profile could

not be opened correctly" after latest update
https://code.google.com/p/chromium/issues/detail?id=455749

The chrome log also has these, which may support the disk full theory.

[713:713:1125/064512:ERROR:extension_pref_value_map.cc(74)] NOTREACHED()
hit.
[713:1031:1125/064525:ERROR:block_files.cc(111)] Failing CreateMapBlock
[713:1031:1125/064525:ERROR:backend_impl.cc(559)] Create entry failed 6243
[713:1031:1125/064525:ERROR:block_files.cc(111)] Failing CreateMapBlock
[713:1031:1125/064525:ERROR:backend_impl.cc(567)] Create entry failed 6245
[713:1031:1125/064526:ERROR:block_files.cc(111)] Failing CreateMapBlock
[713:1031:1125/064526:ERROR:backend_impl.cc(559)] Create entry failed 6412
[713:1031:1125/064526:ERROR:block_files.cc(111)] Failing CreateMapBlock
[713:1031:1125/064526:ERROR:backend_impl.cc(567)] Create entry failed 6413

[chro...@googlecode.com]

Comment #15 on issue 455749 by etco...@gmail.com: "your profile could not

be opened correctly" after latest update
https://code.google.com/p/chromium/issues/detail?id=455749

I'm just one of the users this is happening to. But I also happen to know
a lot about SSDs.

First, my disk was nowhere near full (of user data) when this happened --
and it's happened to me multiple times. (Each time, I have to PowerWash,
which conveniently empties the disk...)

My suspicion is that the problem occurs in writing the profile data
(perhaps the last shutdown). When the problem occurs, it is sticky, so it
is not likely due to timeout on reads.

From what I see in the reports above, I'd be suspicious that the SQL
database is doing something different from other types of accesses. For
example, non-4KB-aligned (from the SSDs idea of alignment) operations. If
there was some generic SSD problem, then wouldn't this happen more
frequently and affect other saved data, including potentially the Chrome OS
system when it is updated. Yet I've seen no evidence of other corruption
(no cached pages that fail to load, no mysterious crashes, ...). What's
the size of the profile data vs. all this other data, and what are the odds
that the profile data alone would fail multiple times with no failures
(observed) elsewhere?

I'm not sure what the ("Failing CreateMapBlock") errors reported above are,
but it would be good to understand what mode the SSD is operating in (is it
a SATA SSD? is it using command queuing?), and exactly what if any errors
the SSD is reporting and in what circumstances. That would seem to be key
to getting to the bottom of this.

For example, one suspicious would be that the SSD is slow (due to a
combination of garbage collection and non-aligned operations) when the
system is shut down, and something is not waiting for the SSD to be truly
done with all outstanding ops when power is removed....

[chro...@googlecode.com]

Comment #17 on issue 455749 by jpodho...@btcatholic.org: "your profile

could not be opened correctly" after latest update
https://code.google.com/p/chromium/issues/detail?id=455749

For what it's worth: we are seeing this issue too (frequently) - we have
just over 1,000 managed Chromebooks and unfortunately have been plagued by
hardware issues (over a 20% failure rate just between 08/2015 and 12/2015).
Thus, I would be perfectly inclined to believe it is caused by faulty/flaky
SSDs and it actually fits in with the other issues we are seeing (0x5b
errors that are not fixed by factory shim reset and require new
motherboards).

However, EVERY time we have this issue it is corrected by simply removing
the user from the device and re-adding them so it's only been an
inconvenience. Our symptoms (in addition to the profile error at sign-in)
include: user cannot install any apps or extensions from web store, user
cannot download any files to local storage, and favicons on bookmark bar
are missing.

[chro...@googlecode.com]

Updates:
Labels: -Pri-2 Pri-1 M-50 Hotlist-ConOps

Comment #19 on issue 455749 by abod...@chromium.org: "your profile could

not be opened correctly" after latest update
https://code.google.com/p/chromium/issues/detail?id=455749

Seeing more reports of this. Bumping up priority.

I know there's not a lot to go on. Is there instrumentation that can be
added to help narrow down the problem?

[chro...@googlecode.com]

Updates:
Status: Available
Owner: ---
Cc: s...@chromium.org hash...@chromium.org

Comment #21 on issue 455749 by sdef...@chromium.org: "your profile could

not be opened correctly" after latest update
https://code.google.com/p/chromium/issues/detail?id=455749

Yes, this error can be presented when the history sql db is corrupted (for
lots of reasons as suggested earlier in the thread). Since the history code
does not remove nor attempt to the sql db, I guess it is normal to see this
error at each login, and for it to be resolved by wiping the account from
the local disk.

The history shutdown process is slow (as it needs to sync multiple sql db
to disk). I remember someone mentioning that there is a timeout when
shutting down ChromeOS. If the timeout is reached, then it does not wait
for all task to finish and just terminate the app. If this is true, then it
could explain the corruption. Here's the bug in the context of which I
remember this mentioned:
https://code.google.com/p/chromium/issues/detail?id=477975

Regarding #20, yes, I think a new owner more familiar with ChromeOS
shutdown should take a look and investigate whether there may be something
terminating the app without giving it opportunity to complete the pending
tasks.

[chro...@googlecode.com]

Comment #23 on issue 455749 by hash...@chromium.org: "your profile could

not be opened correctly" after latest update
https://code.google.com/p/chromium/issues/detail?id=455749

On Chrome OS, when chrome cannot finish shutdown within 3 seconds, the
session manager sends SIGKILL to force shutdown chrome.
This results in unclean shutdown which can leave any user data corrupted.

Fixing any sub task of issue 418627 or issue 474869 will make shutdown
faster, and it will give the history thread more time to finish clean
shutdown so there will be less instances of this problem.
Another unclean option is changing the 3 second timeout to something longer
like 5 seconds or 10 seconds.

For Googlers: This is a doc I composed about a year ago to mitigate this
problem.
https://docs.google.com/document/d/1e7enmQLE1jy2y0Q-054fqgW2RT4mXfK0i_ws8tKsUnw/edit#heading=h.6on13jd9eu3o
AFAIK the situation hasn't changed much since then, so still
content::DOMStorageArea (issue 480228) and base::ImportantFileWriter are
the main causes of browser crashes labelled
as "base::SequencedWorkerPool::Inner::Shutdown" (i.e. the SIGKILL is sent
by the session manager while chrome waits for the worker pool to finish
remaining tasks during shutdown).

[chro...@googlecode.com]

Comment #26 on issue 455749 by dhadd...@google.com: "your profile could not

be opened correctly" after latest update
https://code.google.com/p/chromium/issues/detail?id=455749

Issue chrome-os-partner:52685 has been merged into this issue.