Christopher D. Clausen wrote: > > Oh, so I should email the list with something like: > "testing 1.5.0702 on Windows 2003 SP2. No problems encountered so far." > ?
Yes please do. Here is a quote from the announcement e-mails that we
send for each release:
"Please assist us by testing this release candidate and providing
positive or negative feedback. Bug reports should be filed to
[EMAIL PROTECTED] Reports of success should be sent to
[EMAIL PROTECTED]"
Positive feedback is crucial for two reasons:
(1) it informs the gatekeepers that previous problems have indeed been
addressed
(2) it provides a level of confidence for other openafs users who might
be afraid to install releases that no one else has run. they in
turn will then be more likely to test the release on their systems.
> Yeah, some sort of list of things to check would be useful.
It is not going to be possible to test all possible things. However,
this is a basic list of things that it would be good if people could
test on Windows (both as an upgrade and as a clean install):
(1) Test AFS Shell Extension.
(2) Test Integrated Login
(3) Obtain tokens
(4) Test file operations using the Windows Explorer (copy, paste, new
file, new folder, rename, delete, view photo album, play mp3)
(5) Execute an application out of AFS
(6) Read/Write data from/to AFS using your favorite application.
(Office, ProEngineer, AutoCAD, PhotoShop, ...)
(7) (Optionally) Execute the AFS StressTest
/afs/athena.mit.edu/user/j/a/jaltman/Public/OpenAFS/Support/StressTest.zip
>>> I can (and have been trying to as often as possible) test clients on
>>> Windows 2003 Server (SP1 and SP2 beta x86) as they are released. I
>>> do not currently have any 64-bit OSes running nor do I have any
>>> machines running Windows XP or Windows 2000. If no one else has the
>>> time, I can setup test systems in a VM and make sure a basic install
>>> works, but I would not be actively using software in these VMs. I
>>> would assume that testing in actual environments would be preferable.
>> Testing in actual environments would be preferable in some cases.
>> However, VMs are certainly nice for the ability to perform snapshots.
>> Especially when you want to be able to test the behavior of a clean
>> install.
>
> Would it be possible to get access to the automated testing framework?
> I'm almost sure such tests could be run in VMs as well. Or is this not
> needed b/c others are already running such tests?
I don't think the stress test needs to be run by most people. I run the
test a few times a week. The purpose of the test is to check for
deadlocks. The test is run at MIT in a special environment in which
volumes used for the test are moved and released on a very frequent basis.
Does someone want to volunteer to put together a web form to automate
the reporting?
Jeffrey Altman
smime.p7s
Description: S/MIME Cryptographic Signature
