[prev in list] [next in list] [prev in thread] [next in thread]
List: osdl-storage-sig
Subject: [Storage_sig] NFSv4 testing project status
From: Bryce Harrington <bryce () osdl ! org>
Date: 2005-12-13 0:50:03
Message-ID: 20051213005003.GA18723 () osdl ! org
[Download RAW message or body]
NFSv4 Testing Status - Nov 2005
------------------------------------------------------------------------
Main Achievements
=================
* Released Crucible 1.4
* Spam protection for NFSv4 wiki
* Began new hardware acquisition process
Monthly Statistics
==================
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV
NFSv4 Kernel Releases: 7 7 7 2 3 7 2 0 2 4 1
NFSv4 List Traffic: 148 210 233 185 165 250 186 161 141 154 233
NFSv4 Conference Calls: 1 3 4 4 DONE
NFSv4 New Bug Reports: 9 11 25 6 4 2
OSDL Cross-Compile Runs: 12 ~84 88 22 33 39 22 0 24 40 28
OSDL Regression Tests Run: 12 8 0 8 16 8
OSDL Issues Found: 2 9 11 11 3 5 13 0 4 14 1
OSDL Releases Delivered:
Upstream Patches 0 3 6 0 0 0 7 0 0 0 0
Test Matrix ~10 2 4 4 DONE 1 1
Documents 0 1 0 1 1 2 3 0 0 1 0
Presentations 2 1 0 1 0
Tools/Scripts 0 0 2 1 4 1 0 4 1 2 4
Test Matrix States:
NEW 132 199 263 251 260 245 249 249 251 251 251
OPEN 16 18 33 56 56 72 60 60 62 62 62
IN PROGRESS 3 3 6 6 5 5 8 8 8 8 8
NEAR DONE 0 0 1 1 3 3 3 3 3 3 3
DONE 7 7 5 6 13 13 16 16 16 16 16
Description of Activities at OSDL
=================================
I was away on vacation or holiday for much of the month, but a lot of
activity went into improving the nfsv4 test harness, 'Crucible'. The
development effort focused on generalizing its multi-sut test
capabilities so it can coordinate tests being run on multiple machines.
In other words, when a new NFSv4 patch is released, we can now cause it
to be compiled and installed on an arbitrary number of machines, then
configure some to work as servers and others to run as clients, and then
invoke a variety of tests on them.
Also, a number of improvements have been made to how SUTs can be
managed. A checkout functionality has been implemented to enable
machines to be taken out of the automated queue, so a developer or
tester can do manual testing if needed. (An example of this occurred
this month; an issue I found during the automated testing was of
interest to Bruce, so I checked the machine out to him for several days
so he could do some deeper investigation.)
The main motivation for much of the above work is in anticipation of
growing the hardware devoted to NFSv4. I'm acquiring a set of 8 new x86
client machines, and have gotten an AMD-based server. I plan to acquire
several additional systems next year, once these are in place. A
secondary motivation was that with the prior architecture for running
tests, I was having trouble diagnosing issues in iozone and LTP; now I
can run them in isolation and experiment with some additional variables
and hopefully will be able to narrow in on the problems more swiftly.
I've started work on two new testing tools: 1) Test::Parser is a system
for gathering parsers for different test outputs, that will make it
easier to produce unified reports from a wider variety of tests
(currently, I have to manually review and summarize test results from
each test I run.) 2) WWW::PkgFind is the component used for
automatically retrieving the NFSv4 patches, kernel code, etc. This has
been a key part of the NFSv4 testing framework, and I'm repackaging it
in a way that'll make it more widely usable for others that might be
setting up similar systems.
Our NFSv4 wiki had been getting spammed quite a bit this last month,
which was using up time from NFSv4 developers and testers to revert it.
I spent some time researching into protection mechanisms for this, and
implemented 6 different (complimentary) solutions. While this hasn't
totally eliminated spam, it's cut it way down and has provided us with
better tools to quickly eliminate it.
We've also received sponsorship money from NetApp to hire a sysadmin for
supporting the NFSv4 testing effort. We've begun the process of
creating a job req and are planning to begin interviewing after the
first of the year. This person will help in procuring/racking/installing/
networking/configuring the new NFSv4 test hardware; creating scripts for
power, console, and network management; setting up kerberos/ldap
services; ensuring these testing platforms are accessible for the NFSv4
testers and developers; etc. This will probably be a ~6-month contract
position. If you know of someone in the Portland area with the right
qualifications, let me know! :-)
General Community NFSv4 Testing
===============================
Activity on the list picked up significantly in November, with a lot of
names I've not seen before. Many of the questions have been basic usage
issues or questions about capabilities, but there have also been a
number of legitimate problems reported by the community. For example,
Bruce noticed a lot of confusion around the /etc/exports file format,
and is currently exploring an alternate syntax that'll be more intuitive
for people experienced with other *nix flavors.
Much discussion on the list also surrounded two new features: "Support
for stackable file systems on top of nfs" and "FS-Cache".
Bull has added two new nfsv4 tests to the Linux Test Project (LTP) for
locks and ACLs. I believe the lock tests are the result of Vincent's
efforts testing that area; he had found some issues earlier but Trond's
patches have fixed those bugs, providing the conclusion that NFSv4
locking works.
_______________________________________________
Storage_sig mailing list
Storage_sig@lists.osdl.org
https://lists.osdl.org/mailman/listinfo/storage_sig
[prev in list] [next in list] [prev in thread] [next in thread]
Configure |
About |
News |
Add a list |
Sponsored by KoreLogic