[6911] in Release_7.7_team
Re: Lucid contingency plans
daemon@ATHENA.MIT.EDU (Garry P Zacheiss)
Wed Aug 25 15:40:38 2010
From: Garry P Zacheiss <zacheiss@MIT.EDU>
To: Jonathan D Reed <jdreed@mit.edu>
CC: "release-team@mit.edu" <release-team@mit.edu>
Date: Wed, 25 Aug 2010 15:40:30 -0400
Message-ID: <E2493020-59EE-4DD1-BDCC-F5D29F1022CD@mit.edu>
In-Reply-To: <245D42B3-B01E-4A06-B979-CD11033D0100@mit.edu>
Content-Language: en-US
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Hi Jon,
Have you spoken to Anne (Silvester) about the potential need for significant cluster support on Friday morning? I believe John Guy is currently on vacation, so she may have fewer resources than normal at her disposal.
Does the downgrade package that would potentially required in case "c" already exist? If not, can you estimate the amount of work required?
Garry
On Aug 25, 2010, at 3:25 PM, Jonathan Reed wrote:
> Greetings,
>
> We have identified 3 potential failure modes for the Lucid release:
>
> a) The upgrade bootstrap process fails.
> b) The upgrade bootstrap process succeeds, but the lucid installation fails before completion.
> c) The upgrade succeeds, but Lucid has a previously unknown flaw that renders it unusuable.
>
> For each of these, we have identified a contingency plan:
>
> a) Bootstrap process failure results in the machine staying at Jaunty (and trying again the next day), so no action is necessary.
>
> b) We have seen sporadic failures caused by unresponsive DHCP servers. We have also seen random network hiccups, but we don't believe the risk is larger than with any network-based installation. That having been said, we should ensure that hotline staff are available on the morning of Friday Aug 27 to perform manual installs on failed machines.
>
> c) Lucid machines have been in testing in W20 for the past 2 weeks. Users have been using them, and we have received sporadic test reports. We believe we have identified and corrected for all the known bugs at this point. Should we identify a serious bug that cannot be easily fixed, we would want to downgrade to Jaunty. If the auto-update mechanism is still functional, we can push out a packge that downgrades the machine to Jaunty. As with #2, hotline staff should be available on the morning of Friday Aug 27 to initiate manual installation.
>
> With this information, we are ready to proceed with tomorrow night's release. We are aware that this late in in the summer all options entail tradeoffs, but we believe this to be the best course at this point. If anyone has any significant concerns about proceeding with the release tomorrow night, please raise them at this time.
>
> -Jon
