687109 - AUS3 snippet importer caused us to serve invalid partials to 64-bit linux users on 20110915030845

Status: RESOLVED WONTFIX

(Release Engineering :: General, defect)

Description

[[e-mail] (:bhearsum)]

I think this has something to do with the fact that the nightlies were respun very early today, around 7am PST. After running the importer around 10am today, dev-master01 was serving me the following:
<updates><update type="minor" version="9.0a1" extensionVersion="9.0a1" buildID="20110916071142"><patch type="complete" URL="http://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/2011/09/2011-09-16-07-11-42-mozilla-central/firefox-9.0a1.en-US.linux-x86_64.complete.mar" hashFunction="sha512" hashValue="ec82d62892fed806bd5bced7db36464b38e2fcff899752d857e8a1842b0c5427bb16fa811b3ad6c5f599d1a4df55aa5fc4a7d196f98d7fa4346d823caa8bf83f" size="18728159"/><patch type="partial" URL="http://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/2011/09/2011-09-16-07-11-42-mozilla-central/firefox-9.0a1.en-US.linux-x86_64.partial.20110916030907-20110916071142.mar" hashFunction="sha512" hashValue="afb4a30795317696fd11dd10131e34f98f61e47af4e158752d39317543072d435a92a4fdb309f4df6cd7a5061a02c995275fffa5dd5b5f5d5018418a7630a9c6" size="965133"/></update></updates>


While aus3.m.o was serving:
<updates><update type="minor" version="9.0a1" extensionVersion="9.0a1" buildID="20110916071142"><patch type="complete" URL="http://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/2011/09/2011-09-16-07-11-42-mozilla-central/firefox-9.0a1.en-US.linux-x86_64.complete.mar" hashFunction="sha512" hashValue="ec82d62892fed806bd5bced7db36464b38e2fcff899752d857e8a1842b0c5427bb16fa811b3ad6c5f599d1a4df55aa5fc4a7d196f98d7fa4346d823caa8bf83f" size="18728159"/></update></updates>

...for the same inputs.

Comment 1

[[e-mail] (:bhearsum)]

Looks like it's this line of the shell script that does the imports that mucked this up:
previous=`ls $LOCAL_SNIPPET_DIR/$BRANCH/WINNT_x86-msvc | tail -n2 | head -n1`

and because we had a Linux-only respin this morning, the buildid that this line grabbed wasn't appropriate for Linux builds. Since platform specific respins are a pretty rare thing, I'm calling this WONTFIX, and we can live with the rare cases when this does happen.

Once build machines are capable of talking to AUS3 directly, this won't be an issue.