Closed Bug 1057990 Opened 11 years ago Closed 11 years ago

GMPInstallManager installing openh264 plugin while testrun is underway

Categories

(Mozilla QA Graveyard :: Mozmill Tests, defect)

defect
Not set
normal

Tracking

(Not tracked)

RESOLVED WONTFIX

People

(Reporter: andrei, Unassigned)

References

Details

I've noticed lately that Firefox is installing the openh264 plugin sometime after it is started without any intervention. It might be worth investigating if this can affect our testruns. For example this was installed during one test where we would wait for an external resource to load. There is a possibility that this test failed because the bandwidth was occupied by the plugin thus the page took longer then expected to load: > 04:20:58 1408879193797 GMPInstallManager.simpleCheckAndInstall INFO Last check was: 1408879194 seconds ago, minimum seconds: 86400 > 04:20:58 1408879193797 GMPInstallManager._getURL INFO Using url: https://aus4.mozilla.org/update/3/GMP/%VERSION%/%BUILD_ID%/%BUILD_TARGET%/%LOCALE%/%CHANNEL%/%OS_VERSION%/%DISTRIBUTION%/%DISTRIBUTION_VERSION%/update.xml > 04:20:58 1408879193798 GMPInstallManager._getURL INFO Using url (with replacement): https://aus4.mozilla.org/update/3/GMP/33.0a2/20140824004001/Darwin_x86_64-gcc3-u-i386-x86_64/en-US/aurora/Darwin%2012.5.0/default/default/update.xml > 04:20:58 1408879193800 GMPInstallManager.checkForAddons INFO sending request to: https://aus4.mozilla.org/update/3/GMP/33.0a2/20140824004001/Darwin_x86_64-gcc3-u-i386-x86_64/en-US/aurora/Darwin%2012.5.0/default/default/update.xml > 04:20:59 1408879193900 GMPInstallManager.onLoadXML INFO request completed downloading document > 04:20:59 1408879193901 GMPInstallManager.onLoadXML INFO allowNonBuiltIn: false > 04:20:59 1408879193911 GMPInstallManager.simpleCheckAndInstall INFO Found 1 addons advertised. > 04:20:59 1408879193911 GMPInstallManager.simpleCheckAndInstall INFO Found addon: gmp-gmpopenh264 (isValid: true, isInstalled: false, isOpenH264: true, hashFunction: SHA512, hashValue: 57c2b23c97e32d21ff9f6f611803cfdf6ba4a97f7b403c031e979a79c1db1f507d9b72749824254892032f119a4d057cec926fbc4efb184c4f543dfd34ad72a2, size: 282436) > 04:20:59 1408879193931 GMPInstallManager.simpleCheckAndInstall INFO Addon installed successfully: gmp-gmpopenh264 (isValid: true, isInstalled: true, isOpenH264: true, hashFunction: SHA512, hashValue: 57c2b23c97e32d21ff9f6f611803cfdf6ba4a97f7b403c031e979a79c1db1f507d9b72749824254892032f119a4d057cec926fbc4efb184c4f543dfd34ad72a2, size: 282436) > 04:21:02 ERROR | Test Failure | { > 04:21:02 "exception": { > 04:21:02 "message": "controller.waitForPageLoad(URI=https://www.stopbadware.org/firefox?hl=zh-CN&url=https%3A%2F%2Fwww.itisatrap.org%2Ffirefox%2Fits-an-attack.html, readyState=loading)", > 04:21:02 "lineNumber": 27, > 04:21:02 "name": "AssertionError", > 04:21:02 "fileName": "resource://mozmill/modules/errors.js" > 04:21:02 } > 04:21:02 } > 04:21:02 TEST-UNEXPECTED-FAIL | testSecurity/testSafeBrowsingNotificationBar.js | testNotificationBar AFAIK this plugin is needed for WebRTC, so we might be safer to disable it completely for mozmill-tests for now (if it indeed can affect our tests stability)
I don't think there is a way to prevent the installation of the gmp plugin. But Nils might know more, and what we could do here.
Flags: needinfo?(drno)
According to bug 1046644 we should have now the user pref media.gmp-gmpopenh264.autoupdate allowing us to disable the download of the OpenH264 plugin.
Depends on: 1046644
Flags: needinfo?(drno)
It would be much better to not disable the autoupdate and instead set a dummy URL override like this: https://hg.mozilla.org/releases/mozilla-aurora/rev/c0e95adfb454#l1.13
Can I ask how much data we have to download for this plugin? I wonder if that really is a problem for us.
I'm not concerned about data-volume, but about interference (i.e. blocking the main thread bug 1039490). Unfortunately I have no testcase and can't be sure of its effects. Maybe we should not do any changes until we have a way of measuring the impact.
Interesting bug. It looks like nothing which can be fixed shortly. So my question still persists, in how much data we have to download. If its only a couple of kb, I don't see a problem for our CI machines in SCL3 given the good bandwidth.
This isn't very big, e.g. via here: https://aus4.mozilla.org/update/3/GMP/34.0a1/20140825171445/Darwin_x86_64-gcc3/en-US/default/Windows 7.0/default/default/update.xml ... it's 276kb for Darwin: http://ciscobinary.openh264.org/openh264-macosx64-master-300cceaabf0c423a3f464c963ac1b7abdae6e6b3.zip If you can keep this enabled and are not worried about hitting the network, then the added test-coverage would be nice to have.
That's indeed very small. So I don't see how this would cause network issues and timeouts for waitForPageLoad. Also unzipping the file should be done kinda quick. I would vote for keeping it in.
In that case let us leave this in. We should be aware that this happens and monitor further to see if there's any impact.
Status: NEW → RESOLVED
Closed: 11 years ago
Resolution: --- → WONTFIX
Product: Mozilla QA → Mozilla QA Graveyard
You need to log in before you can comment on or make changes to this bug.