Closed
Bug 1188992
Opened 9 years ago
Closed 9 years ago
[OTA] Latest Flame build is a day behind
Categories
(Firefox OS Graveyard :: Gaia::System, defect)
Tracking
(b2g-master verified)
VERIFIED
FIXED
Tracking | Status | |
---|---|---|
b2g-master | --- | verified |
People
(Reporter: Marty, Unassigned)
Details
(Keywords: qablocker, Whiteboard: [2.5-Daily-Testing], [b2g-build-support])
Attachments
(3 files)
Description: Using yesterday's '20150728030208' build, there is no OTA update available on the 'nightlytest' channel. According to nhirata on irc, this is due to an aliasing issue. <nhirata_> B2G-mozilla-central-nightly-latest <nhirata_> 20150729030209 <nhirata_> that shuld be the build <nhirata_> oh it's not aliased <nhirata_> 20150727162816 <nhirata_> file a bug where flame isn't being aliased for flame-kk <nhirata_> and NI me please <nhirata_> I just changed it so you have the right build. <nhirata_> 20150729030209 Repro Steps: 1) Update a Flame to 20150728030208 2) Connect to a WiFi or Data network 3) Change the update channel to 'nightlytest' 4) Check for updates Actual: No system updates are found Expected: The latest Flame build system update is found successfully. Environmental Variables: Device: Flame 2.5 Build ID: 20150728030208 Gaia: 14e32276025b0310d3e89027320cf4b2a24cedfb Gecko: 33dc8a83cfc0 Gonk: 41d3e221039d1c4486fc13ff26793a7a39226423 Version: 42.0a1 (2.5) Firmware Version: v18D User Agent: Mozilla/5.0 (Mobile; rv:42.0) Gecko/42.0 Firefox/42.0 Repro frequency: 5/5 See attached: Logcat
Reporter | ||
Updated•9 years ago
|
Flags: needinfo?(nhirata.bugzilla)
{ "platforms": { "aries-userdebug": { "locales": { "en-US": { "buildID": "20150803170922", "platformVersion": "42.0a1", "displayVersion": "42.0a1", "appVersion": "42.0a1", "completes": [ { "fileUrl": "https://queue.taskcluster.net/v1/task/mimzLu8_SqqPEFA26Sw1Bg/runs/0/artifacts/public/build/b2g-aries-gecko-update.mar", "from": "*", "hashValue": "4baeeff61875b96f742953eb494e2f42d59270408a1386b145867cd5ecae0d56fff663133593fa39f23e76f78be6ed1d61789279076f19eb4954c3a877f9abb7", "filesize": 125875211 } ] } } }, "aries": { "locales": { "en-US": { "buildID": "20150803171137", "platformVersion": "42.0a1", "displayVersion": "42.0a1", "appVersion": "42.0a1", "completes": [ { "fileUrl": "https://queue.taskcluster.net/v1/task/t8LDwvzhR5ihqWZAw6O2uA/runs/0/artifacts/public/build/b2g-aries-gecko-update.mar", "from": "*", "hashValue": "a40ed60cc861771aa22b5f84da95ebee7802b58f0538bf069d05d0b80cd98cedfbc3f8842d762e350b4566121a72d22dc252a1c1e0a8a997e0c046887743ee2f", "filesize": 125679252 } ] } } }, "flame": { "locales": { "en-US": { "buildID": "20150803132043", "platformVersion": "42.0a1", "displayVersion": "42.0a1", "completes": [ { "fileUrl": "https://queue.taskcluster.net/v1/task/-t1J0bTXTpyXvpcH2xfnvg/runs/0/artifacts/public/build/b2g-flame-gecko-update.mar", "from": "*", "hashValue": "d60aee0a563a6b92e2f5e8f152bff9dce881dcd1a2f95371d15facf172f3072cbd46cd7853ce903464dfe2584faea19367a6088ac8ec883a860501e26fd565d5", "filesize": 94557431 } ], "appVersion": "42.0a1" } } }, "flame-userdebug": { "locales": { "en-US": { "buildID": "20150803170724", "platformVersion": "42.0a1", "displayVersion": "42.0a1", "completes": [ { "fileUrl": "https://queue.taskcluster.net/v1/task/ePvjJbhTRAuHGV-zy9bYUA/runs/0/artifacts/public/build/b2g-flame-gecko-update.mar", "from": "*", "hashValue": "41c74c3af3e9db006de2415a6a8b7e74c3aa4c5346ff03d3d5d9c4825bad778658855522bc172ef0504cebab8b3e8f6974075c60418782c267ac62c804f2a26c", "filesize": 94798429 } ], "appVersion": "42.0a1" } } } }, "hashFunction": "sha512", "name": "B2G-gecko-nightly-latest", "schema_version": 4 } Changed by b2gbld 12 days ago
FYI, not sure, but I think there's something up with buildbot.
Flags: needinfo?(wcosta)
Comment 3•9 years ago
|
||
Sorry for the late response, iirc, after mozharness went in-tree, blob name is now B2G-gecko-nightly-latest.
Flags: needinfo?(wcosta)
Thanks Wander. I wasn't aware of this... Would there be a way that we can get a heads up on these changes before they take place? Also flame is still a day behind. "flame": { "locales": { "en-US": { "buildID": "20150820202733", "platformVersion": "43.0a1", "displayVersion": "43.0a1", "completes": [ { "fileUrl": "https://queue.taskcluster.net/v1/task/NBHLNIaWRgSZq5yN2zO87g/runs/0/artifacts/public/build/b2g-flame-gecko-update.mar", "from": "*", "hashValue": "f12e40a3c801d85870b826cfcf59988b64788e83a4f41a1784e88f8fb3c0d5188ed9a8a3b0db5a59d78fcb6a957c426430f53ebab6bedc186fee4b1bc589ea5c", "filesize": 95680379 } ], "appVersion": "43.0a1" } } },
Flags: needinfo?(wcosta)
Flags: needinfo?(nhirata.bugzilla)
Flags: needinfo?(mshal)
Summary: [OTA] Latest Flame build is not being aliased for flame-kk on nightlytest channel → [OTA] Latest Flame build is a day behind
Comment 5•9 years ago
|
||
After chatting with wcosta and bhearsum, it sounds like the problem (or at least part of the problem) may be from how the balrog rules are configured. In particular, the update query is using "nightlytest", while balrog is configured to use "nightly-test". bhearsum also noted a few other issues, which he can explain better than I :). In any case, this doesn't appear to be a problem with buildbot at the moment.
Flags: needinfo?(mshal) → needinfo?(bhearsum)
Comment 6•9 years ago
|
||
To address the immediate issue, it looks to me that there's a mismatch between channel name in the rule ("nightly-test"), and the one in the update URL ("nightlytest"). The latter is what we typically use, so I'd advise changing the one in the rule. I also noticed some other weird things that may or may not be causing issues with this or other things: * Many other rules have "-test" in their name: dogfood-test, nightly-b2g34-test, nightly-b2g37-test * There's a rule with "nightly-latest" for its channel, is that even used? * It looks like all of the releases submitted by taskcluster start with "B2G-gecko". There's probably going to be some naming collisions if those are actually for different branches. Eg: if some are based on "master" and some are based on "b2g37", B2G-gecko-$buildid isn't a unique enough name. * The "aurora" channel rule for flame-kk is pointing at mozilla-b2g34. This one looks like it might be dead now, since most things point at B2G-gecko releases rather than B2G-mozilla-whatever
Flags: needinfo?(bhearsum)
Updated•9 years ago
|
Flags: needinfo?(wcosta)
I ended up changing some of the rules at comment 4: * The nightly-latest and dogfood-latest have been updated to use B2G-gecko-nightly-latest * The dogfood-test and nightly-test are used for QA to change to an OTA test build. nightlytest moved to nightly-latest * dogfood and nightly are for public consumption. I did that so both channels conform rather than having nightlytest and dogfood-latest as it was confusing people. Aurora has been dead for a while, 2.1, 2.2 has discontinued on pvtbuild, there's a build to have task cluster pick those up. The comment 4 was in regards to the wrong build for the release and I believe it may have to do with the explanation of : "It looks like all of the releases submitted by taskcluster start with "B2G-gecko". There's probably going to be some naming collisions if those are actually for different branches. Eg: if some are based on "master" and some are based on "b2g37", B2G-gecko-$buildid isn't a unique enough name." I think there's some collision happening; it would be great if we could get some sort of logging that we can look at to troubleshoot the issue...
Flags: needinfo?(mshal)
Flags: needinfo?(bhearsum)
Comment 8•9 years ago
|
||
(In reply to Naoki Hirata :nhirata (please use needinfo instead of cc) from comment #7) > I ended up changing some of the rules at comment 4: > * The nightly-latest and dogfood-latest have been updated to use > B2G-gecko-nightly-latest > * The dogfood-test and nightly-test are used for QA to change to an OTA test > build. nightlytest moved to nightly-latest > * dogfood and nightly are for public consumption. > > I did that so both channels conform rather than having nightlytest and > dogfood-latest as it was confusing people. > > Aurora has been dead for a while, 2.1, 2.2 has discontinued on pvtbuild, > there's a build to have task cluster pick those up. > > The comment 4 was in regards to the wrong build for the release and I > believe it may have to do with the explanation of : > "It looks like all of the releases submitted by taskcluster start with > "B2G-gecko". There's probably going to be some naming collisions if those > are actually for different branches. Eg: if some are based on "master" and > some are based on "b2g37", B2G-gecko-$buildid isn't a unique enough name." > > I think there's some collision happening; it would be great if we could get > some sort of logging that we can look at to troubleshoot the issue... I'm not really sure where to point you here, I have very little knowledge on these b2g builds. https://github.com/mozilla/gecko-dev/blob/master/testing/taskcluster/scripts/phone-builder/build-phone-ota.sh looks like it might be involved? But I'm not really sure. There's sections in logs like https://s3-us-west-2.amazonaws.com/taskcluster-public-artifacts/vvxdielpQnG2Uv9U5-k-tA/0/public/logs/live_backing.log that says " Balrog request to http://balrog/api/releases/B2G-gecko-nightly-2015082702214", so if all branches use the same base string (B2G-gecko-nightly), it's pretty clear to me that there's collisions happening.
Flags: needinfo?(bhearsum)
Comment 9•9 years ago
|
||
:wcosta, do you know or can you point us to someone who knows how these builds are deciding what to submit to balrog? I'm out of my league here :/
Flags: needinfo?(mshal) → needinfo?(wcosta)
Comment 10•9 years ago
|
||
(In reply to Michael Shal [:mshal] from comment #9) > :wcosta, do you know or can you point us to someone who knows how these > builds are deciding what to submit to balrog? I'm out of my league here :/ This all happens inside mozharness: https://dxr.mozilla.org/mozilla-central/source/testing/mozharness/scripts/b2g_build.py?offset=200#1068-1104 https://dxr.mozilla.org/mozilla-central/source/testing/mozharness/mozharness/mozilla/updates/balrog.py?offset=2000#17-84 And tools: https://github.com/mozilla/build-tools/blob/master/scripts/updates/balrog-submitter.py https://github.com/mozilla/build-tools/blob/master/lib/python/balrog/submitter/cli.py
Flags: needinfo?(wcosta)
mshal, I have a hard time tracking what is actually updating the rule. I think if we had some patches for logging what actually updates balrog, it might shed some clarity why there's a disparity.
Comment 12•9 years ago
|
||
Any progress on this? Dogfooding builds are now 3 weeks old (20150820200530).
This doesn't block dogfood. Please see https://wiki.mozilla.org/B2G/QA/Spark for Aries or https://wiki.mozilla.org/B2G/QA/Flame_OTA#Nightly for Flame for the information in regards to what's blocking them.
Comment 14•9 years ago
|
||
Are the channel names "nightly test" and "nightly latest" (mind the spaces) correct under https://wiki.mozilla.org/B2G/QA/Flame_OTA#Nightly ?
Comment 15•9 years ago
|
||
:nhirata - it looks like both the flame-kk-ota-* builds (running in Taskcluster) and the flame nightly build (running in buildbot) are submitting updates to balrog. Is that intended? Or should we be turning off the flame nightlies in buildbot? Separately, what kind of logging are you looking for? We have the build logs for both when submitting things to balrog, but I don't know if that has the info that you're looking for.
Flags: needinfo?(nhirata.bugzilla)
I think we want to remove flame-kk-ota-* from updating the latest, right? Because I think the goal is to have Nightly builds... I could be mistaken. We don't have a test server, do we? Basically a log file stating which script updated the information for the releases.
Flags: needinfo?(nhirata.bugzilla)
Comment 17•9 years ago
|
||
(In reply to Naoki Hirata :nhirata (please use needinfo instead of cc) from comment #16) > I think we want to remove flame-kk-ota-* from updating the latest, right? We can disable the flame-kk-ota-* builds in taskcluster for now. Is that acceptable? The only difference between the -ota- version of these builds is that it submits to balrog. :) I realise that we would like to get these jobs moved over, but we have a couple dependencies that need to be resolved in Q4 before that can happen.
Flags: needinfo?(nhirata.bugzilla)
Sounds good! I think disabling it would help. We can turn it back on once the dependencies are done. Thanks!
Flags: needinfo?(nhirata.bugzilla) → needinfo?(sdeckelmann)
Updated•9 years ago
|
Flags: needinfo?(sdeckelmann)
Whiteboard: [2.5-Daily-Testing] → [2.5-Daily-Testing] [tc-emulator]
Updated•9 years ago
|
Whiteboard: [2.5-Daily-Testing] [tc-emulator] → [2.5-Daily-Testing] [tc-build-support]
Updated•9 years ago
|
Whiteboard: [2.5-Daily-Testing] [tc-build-support] → [2.5-Daily-Testing], [b2g-build-support]
Comment 19•9 years ago
|
||
Agreed - let's disable the flame-kk-ota builds
Following up on this bug, can we have https://bugzilla.mozilla.org/show_bug.cgi?id=1188992#c17 done please? If you want us to file a new bug in regards to that with a dependency on this one, we can do this.
Flags: needinfo?(sdeckelmann)
Comment 21•9 years ago
|
||
(In reply to Naoki Hirata :nhirata (please use needinfo instead of cc) from comment #20) > Following up on this bug, can we have > https://bugzilla.mozilla.org/show_bug.cgi?id=1188992#c17 done please? > If you want us to file a new bug in regards to that with a dependency on > this one, we can do this. This was done in bug 1212812
Status: NEW → RESOLVED
Closed: 9 years ago
Resolution: --- → FIXED
Comment 22•9 years ago
|
||
This issue can repro on the Flame KK v2.5 user build by the following STR. STR: 1. Update Flame to 20151014150203. 2. Connect to a WiFi or Data network. 3. Tap on the Setting app on the homescreen. 4. Tap Developer button on the Settings. 5. In the Update Channel, Change the update channel to 'nightlytest'. 6. Tap Check Now button on the Device Information. Actual results: No system updates are found. See attachment: verified_logcat_2039.txt. verified_Flame KK_v2.5.3gp Reproduce rate: 10/10 Device: Flame KK 2.5 user(Affected) Build ID 20151014150203 Gaia Revision c64d6e345aa477f2a24b676e84a47020a00d015d Gaia Date 2015-10-14 15:54:16 Gecko Revision https://hg.mozilla.org/mozilla-central/rev/4f4615ffec6a6a7ec40ff61ffda90a46c53f8d31 Gecko Version 44.0a1 Device Name flame Firmware(Release) 4.4.2 Firmware(Incremental) eng.cltbld.20151014.184235 Firmware Date Wed Oct 14 18:42:47 EDT 2015 Firmware Version v18D v4 Bootloader L1TC000118D0
Comment 23•9 years ago
|
||
Comment 24•9 years ago
|
||
Updated•9 years ago
|
QA Whiteboard: [MGSEI-Triage+]
Comment 25•9 years ago
|
||
(In reply to Tony from comment #22) > This issue can repro on the Flame KK v2.5 user build by the following STR. > STR So are you saying that this is broken still despite bug 1212812?
Flags: needinfo?(sdeckelmann)
Updated•9 years ago
|
Flags: needinfo?(mshuman)
Comment 26•9 years ago
|
||
(In reply to Tony from comment #22) > This issue can repro on the Flame KK v2.5 user build by the following STR. > STR: > 1. Update Flame to 20151014150203. > > 2. Connect to a WiFi or Data network. > 3. Tap on the Setting app on the homescreen. > 4. Tap Developer button on the Settings. > 5. In the Update Channel, Change the update channel to 'nightlytest'. > 6. Tap Check Now button on the Device Information. > > Actual results: No system updates are found. > See attachment: verified_logcat_2039.txt. verified_Flame KK_v2.5.3gp > Reproduce rate: 10/10 > > Device: Flame KK 2.5 user(Affected) > Build ID 20151014150203 > Gaia Revision c64d6e345aa477f2a24b676e84a47020a00d015d > Gaia Date 2015-10-14 15:54:16 > Gecko Revision > https://hg.mozilla.org/mozilla-central/rev/ > 4f4615ffec6a6a7ec40ff61ffda90a46c53f8d31 > Gecko Version 44.0a1 > Device Name flame > Firmware(Release) 4.4.2 > Firmware(Incremental) eng.cltbld.20151014.184235 > Firmware Date Wed Oct 14 18:42:47 EDT 2015 > Firmware Version v18D v4 > Bootloader L1TC000118D0 Hi, Tony, To my knowledge, we were using PVT build to do Flame OTA test because of some test purposes. So, I am not sure if this problem relates to PVT build. Can we try to use task cluster build to do Flame OTA test again? Thank you.
Flags: needinfo?(tianxu)
Comment 27•9 years ago
|
||
Hi, William, Thanks for your reminding. With William's help, this bug has been verified as "pass" on the old build of Flame KK 2.5 user(Build ID 20151020150236). Please check the link below for verification information. thanks. URI: https://aus5.mozilla.org/update/5/B2G/44.0a1/20151020003809/flame/en-US/nightly-test/Boot2Gecko 2.5.0.0-prerelease (SDK 19)/default/default/default/update.xml?force=1. Actual results: system updates are found. Reproduce rate: 0/10. Device: Flame KK 2.5 user(Pass) Build ID 20151020150236 Gaia Revision 11eb5d4cb2675d359d277ae17772bc75f7ccedbc Gaia Date 2015-10-20 16:22:28 Gecko Revision https://hg.mozilla.org/mozilla-central/rev/f397034950304b845175f8dab8fccbdd0e8bf995 Gecko Version 44.0a1 Device Name flame Firmware(Release) 4.4.2 Firmware(Incremental) eng.cltbld.20151020.184701 Firmware Date Tue Oct 20 18:47:13 EDT 2015 Firmware Version v18D v4 Bootloader L1TC000118D0
Flags: needinfo?(tianxu)
Updated•9 years ago
|
Comment 28•9 years ago
|
||
Thank you, everyone. Looks good to me.
Flags: needinfo?(whsu)
Flags: needinfo?(mshuman)
You need to log in
before you can comment on or make changes to this bug.
Description
•