902985 - Unable to get/save contacts via the API immediately after pushing a reference database

Status: RESOLVED WONTFIX

(Core Graveyard :: DOM: Contacts, defect, P5)

Description

[Dave Hunt [:davehunt] [he/him] ⌚BST]

You will need a device running an engineering build so you can use Marionette to execute asynchronous JavaScript. Also, please note that data will be wiped from the device.

Steps to reproduce:
1. Download https://github.com/mozilla-b2g/gaia/raw/master/test_media/reference-workload/contactsDb-200.sqlite
2. Connect B2G device with master engineering build via USB
3. Run the following from the command line:

$ adb forward tcp:2828 tcp:2828
$ adb shell stop b2g
$ adb shell rm -r /data/local/indexedDB/chrome/idb/*csotncta*
$ adb push contactsDb-200.sqlite /data/local/indexedDB/chrome/idb/3406066227csotncta.sqlite
$ adb shell start b2g

4. Wait for B2G to start
5. Run the following to install the Marionette client:

$ pip install marionette_client

4. Start a Python interactive shell and run the following:

>>> from marionette import Marionette
>>> m = Marionette()
>>> m.start_session()
u'6-b2g'
>>> print m.execute_async_script("SpecialPowers.addPermission('contacts-read', true, document); var req = window.navigator.mozContacts.find({}); req.onsuccess = function () { marionetteScriptFinished(req.result) };", special_powers=True, script_timeout=60000)

Expected:
Details of the 200 contacts should be output to the console

Actual:
The script times out

If you run the last Marionette command a second time, it is successful. It seems there's an issue performing contacts API calls on a fresh reference database.

Comment 1

[Ben Kelly [:bkelly, not reviewing]]

Looks like this is using contacts API directly, so updating component.

Comment 2

[Gregor Wagner [:gwagner]]

The reference database probably uses an old DB schema and we have to iterate over all contacts and modify them when we do a schema upgrade. This might take some time. Can you try to increase the timeout to see if it works?

Comment 3

[Dave Hunt [:davehunt] [he/him] ⌚BST]

(In reply to Gregor Wagner [:gwagner] from comment #2)
> The reference database probably uses an old DB schema and we have to iterate
> over all contacts and modify them when we do a schema upgrade. This might
> take some time. Can you try to increase the timeout to see if it works?

The strange thing is that this can time out after 10 seconds and trying again immediately with a 10 second timeout works. Trying with a 60 second timeout fails, and I have also tried a 3 minute timeout without success.

Comment 4

[Jon Hylands [:jhylands]]

From what I saw of the conversation yesterday, it appears that the first call to the API fails (regardless of the timeout), and the second one succeeds.

Comment 5

[Reuben Morais [:reuben]]

(In reply to Gregor Wagner [:gwagner] from comment #2)
> The reference database probably uses an old DB schema and we have to iterate
> over all contacts and modify them when we do a schema upgrade. This might
> take some time. Can you try to increase the timeout to see if it works?

More than 60 seconds for a 200 contacts DB? I think it's more likely that the upgrade code is exploding, leaving things half-broken and causing the error on the first transaction.

Comment 6

[Gregor Wagner [:gwagner]]

(In reply to Reuben Morais [:reuben] from comment #5)
> (In reply to Gregor Wagner [:gwagner] from comment #2)
> > The reference database probably uses an old DB schema and we have to iterate
> > over all contacts and modify them when we do a schema upgrade. This might
> > take some time. Can you try to increase the timeout to see if it works?
> 
> More than 60 seconds for a 200 contacts DB? I think it's more likely that
> the upgrade code is exploding, leaving things half-broken and causing the
> error on the first transaction.

If it's an emulator running a debug build... But yeah something is strange here. Can we get a regular logcat output?

Comment 7

[Dave Hunt [:davehunt] [he/him] ⌚BST]

Attachment: logcat

Comment 8

[Dave Hunt [:davehunt] [he/him] ⌚BST]

(In reply to Gregor Wagner [:gwagner] from comment #6)
> If it's an emulator running a debug build... But yeah something is strange
> here. Can we get a regular logcat output?

This is a device running an engineering build.

I've attached a logcat, however there's a difference between my steps to reproduce and the way I ran this. I used b2gpopulate, which is how this issue was initially discovered. Essentially it does exactly the same as the steps to reproduce by pushing the reference database and then attempting to insert additional contacts via the API.

Comment 9

[Dave Hunt [:davehunt] [he/him] ⌚BST]

Is the attached logcat useful? Anything else I can do to assist with this?

Comment 10

[Gregor Wagner [:gwagner]]

(In reply to Dave Hunt (:davehunt) from comment #9)
> Is the attached logcat useful? Anything else I can do to assist with this?

Nothing obvious in the logcat. Maybe we need a reboot instead of b2g stop.
Jan, Ben do you know if stopping the parent process is enough after replacing a DB?

Comment 11

[Ben Turner (not reading bugmail, use the needinfo flag!)]

Stop/Start should be fine I think.

Comment 12

[Firefox Bug Husbandry Bot]

https://bugzilla.mozilla.org/show_bug.cgi?id=1472046

Move all DOM bugs that haven’t been updated in more than 3 years and has no one currently assigned to P5.

If you have questions, please contact :mdaly.

Comment 13

[Sylvestre Ledru [:Sylvestre]]

DOM: Contacts isn't used anymore. 
Closing all remaining bugs.