1174041 - [Flame] Shallow flashing to tinderbox KK central eng builds causes error: No space left on device

Status: RESOLVED WORKSFORME

(Firefox OS Graveyard :: Gaia::Build, defect)

Description

[Pi Wei Cheng [:piwei] (inactive)]

Description:
When shallow flashing latest builds from https://pvtbuilds.mozilla.org/pvt/mozilla.org/b2gotoro/tinderbox-builds/mozilla-central-flame-kk-eng/ , it gives an error saying that the device has run out of space when I try to flash it the second time.

STR:
1) Base the Flame to v18D-1 base image
2) Shallow flash to a latest build from the above link

- Observe the flash is successful and device enters FTU

3) Shallow flash the same build again

Expected: Flashing is successful and enters FTU

Actual: Flashing is unsuccessful. The terminal shows an error saying: "No space left on device".

Repro rate: 10 out of 10 for me.

Comment 1

[Pi Wei Cheng [:piwei] (inactive)]

This issue doesn't happen on v2.2 (from this link: https://pvtbuilds.mozilla.org/pvt/mozilla.org/b2gotoro/tinderbox-builds/mozilla-b2g37_v2_2-flame-kk-eng/ )

Comment 2

[Pi Wei Cheng [:piwei] (inactive)]

mozilla-inbound regression window:

Last Working
Device: Flame
BuildID: 20150603064143
Gaia: 477b5672811ed970a7476fe6f67dba546a302dce
Gecko: 24d30440d9eb
Version: 41.0a1 (3.0 Master) 
Firmware Version: v18D-1
User Agent: Mozilla/5.0 (Mobile; rv:41.0) Gecko/41.0 Firefox/41.0

First Broken
Device: Flame
BuildID: 20150603072142
Gaia: 477b5672811ed970a7476fe6f67dba546a302dce
Gecko: 91bba46e795a
Version: 41.0a1 (3.0 Master) 
Firmware Version: v18D-1
User Agent: Mozilla/5.0 (Mobile; rv:41.0) Gecko/41.0 Firefox/41.0

Gaia is the same so it's a Gecko issue.

Gecko pushlog:
http://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=24d30440d9eb&tochange=91bba46e795a

From the pushlog, it seems like Bug 1051146 is a possible culprit.

Comment 3

[KTucker [:KTucker][Inactive 3/4/2016]]

Kelly, can you take a look at this please? Looks like the landing for bug 1051146 might be the cause here.

Comment 4

[kdavis]

(In reply to KTucker [:KTucker] from comment #3)
> Kelly, can you take a look at this please? Looks like the landing for bug
> 1051146 might be the cause here.

Sounds likely that Bug 1051146 is the cause. 

Bug 1051146 installs a speech-to-text engine on B2G nightly, and part of this
engine is a statistic model of the English language. This model is on the order
of 20MB.

The question is what to do.

The model installation is behind a flag. So, it can be easily turned on or off,
but then people working on the UI/UX for our voice app will have to compile their
own gecko, which is also not ideal.

This Thursday we, the group working on a speech interface to B2G, have a team
meeting where we can discuss the proper resolution. I'll give an update then.

Comment 5

[kdavis]

It looks like we can get this solved through this Bug 1154072 Comment 31,
the partition size will be expanded.

Comment 6

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

QA wanted to see if this still occurs.

Comment 7

[Pi Wei Cheng [:piwei] (inactive)]

We haven't seen this since we moved to v18D nightly v4 base image. Fixed by changing base image and closing bug.