Chef ?

Anybody know anything about "chef" ?

http://wiki.opscode.com/display/chef/Home

   Paul

--
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.

Virtual Infrastructure Workshop at LISA

If anyone is interested, this year's LISA workshop will be on  
"Virtual Infrastructures".
This is slightly different from the previous workshops, in that it is  
not focussed
only on configuration - but of course, virtual infrastructures  
present lots of
interesting configuration problems ...

    http://homepages.inf.ed.ac.uk/group/lssconf/iWeb/lssconf/2008.html

   Paul

--
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.

Internet Network Management Workshop

Please consider submitting to this workshop. It is Orlando, FL in 
October. -- Sanjai

=======================================================================
CALL FOR PAPERS
Internet Network Management Workshop (INM) 2008
Sponsored by IEEE Computer Society and IEEE Communications Society
Co-located with ICNP 2008
Orlando, Florida
October 19, 2008
http://www.cs.rice.edu/~eugeneng/inm08
=======================================================================

New extended deadlines:
    Paper registration: 5:00pm EDT, July 7, 2008
    Paper submission: 5:00pm EDT, July 14, 2008
    Notification of acceptance: August 6, 2008
    Camera ready submission: September 5, 2008

In many ways, computer network management remains the least understood
aspect of computer networking. There is a lack of well-established
principles for guiding the design of networks for manageability. There is
also a lack of scientific theories for analyzing the state of a network
and for the evolution of network state.

The Internet Network Management (INM) workshop provides an opportunity to
elevate participants' collective experience with IP networks into ideas,
principles, and theories that can be leveraged in today's networks, or can
be carried forward into the clean-slate design of future networks that

Re: lssconf-discuss Digest, Vol 32, Issue 3

On 28 Nov 2007, at 12:00,  "Steven Jenkins"  
<steven.jenkins <at> gmail.com> wrote:

> Have you tried experimenting to see a minimally complete type system
> for the evolution and learning of a systems administration ontology?
> I worked on a very, very simple system that looked like this:
>
> .........

Have you written anything about this?

    Paul

Request for discussion: configuration management taxonomy

_______________________________________________
lssconf-discuss mailing list
lssconf-discuss <at> inf.ed.ac.uk
http://lists.inf.ed.ac.uk/mailman/listinfo/lssconf-discuss

(unknown)

Apologies for the spam, but some of you may be interested in this  
conference ... It is rather network-oriented, but it now includes  
"management of autonomic networks and systems" which includes  
configuration ...

   Paul

Begin forwarded message:
> From: IARIA Invitation <invitation <at> iaria.org>
> Date: 30 July 2007 19:34:25 BDT
> To: dcspaul <at> inf.ed.ac.uk
> Subject: ICN 2008 || The Seventh International Conference on  
> Networking || Cancun, Mexico, April 13-18, 2008
>
> Invitation:
>
> Please consider to contribute and distribute to the appropriate  
> groups the following
>
> CALL FOR PAPERS, TUTORIALS, PANELS
>
> ICN 2008, The Seventh International Conference on Networking
>
> April 13-18, 2008 - Cancun, Mexico
>
> Site: http://www.iaria.org/conferences2008/ICN08.html
>
> Submit a paper: http://www.iaria.org/conferences2008/SubmitICN08.html
>
> Submissions will be peer-reviewed, published by IEEE CPS, posted in  
> IEEE Digital Library, and indexed with the major indexes.
>
> Extended versions of selected papers will be invited for  
> specialized journals.
>
>
>
> Important deadlines:
>
> Submission deadline: November 5, 2007
>
> Notification of acceptance: December 15, 2007
>
> Registration/camera ready: January 20, 2008
>
>
>
> ICN 2008 Topics (details in the CfP on site):
>
> 1. Communication theory
>
> 2. Communications switching and routing
>
> 3. Communications modeling
>
> 4. Communications security
>
> 5. Computer communications
>
> 6. Distributed communications
>
> 7. Signal processing in communications
>
> 8. Multimedia and multicast communications
>
> 9. Wireless communications (satellite, WLL, 4G, Ad Hoc, sensor  
> networks)
>
> 10. Next generation networks [NGN] principles
>
> 11. Storage area networks [SAN]
>
> 12. Access and home networks
>
> 13. High-speed networks
>
> 14. Optical networks
>
> 15. Peer-to-peer and overlay networking
>
> 16. Mobile networking and systems
>
> 17. MPLS-VPN, IPSec-VPN networks
>
> 18. GRID networks
>
> 19. Broadband networks
>
> 20. Quality of service, service level agreement [QoS/SLA]
>
> 21. Reliability, availability, serviceability [RAS]
>
> 22. Traffic engineering, metering, monitoring
>
> 23. Voice over IP services
>
> 24. Performance evaluation, tools, simulation
>
> 25. Network, control and service architectures
>
> 26. Network signaling, pricing and billing
>
> 27. Network middleware
>
> 28. Telecommunication networks architectures
>
> 29. On-demand networks, utility computing architectures
>
> 30. Applications and case studies
>
> 31. NGN protocol design and evaluation
>
> 32. NGN Standard Activities [ITU, TMF, 3GPP, IETF, etc.]
>
> 33. NGN Device Instrumentation
>
> 34. Network Management, scheduling and policy
>
> 35. NGN policy-based control
>
> 36. Networks policy-based management
>
> 37. Management of autonomic networks and systems
>
>
> ======================
>
> IARIA Publicity Board
>
> ======================

Re: testing large systems using virtualization (fwd)

LSC: Luke Crawford
OL: off-list mail

forwarding off-list comments on-list;  I'm not 100% sure, but I bet others 
would be interestted.

This list is mostly academia, no?  the SysAdmins on the ground are often 
not academic at all (I've got no formal education to speak of at all; 
this is extremely common for SysAdmins)  I think both academia and 
the profession of system adminstration could benifit greatly if there was 
more communication between the two tiers.

(getting some formal education has been on my to-do list for a while, but 
I've got a lot of other stuff on my  plate.)

...

really, at this scale, doing anything at all on the per-node basis is generally 
untenable.   We even re-image boxes using tools that specify a range of nodes. 
(one of the things I'm working on is automating the detection of bad hardware; 
we have a small army of hardware guys, but they usually need detail on the 
level of "replace drive X"  to actually fix anything.  right now, we have a 
small army of expensive contractors doing this sort of thing, but I believe 
most of it can be automated away.)

You are correct that things get a whole lot more complicated (way more 
complicated than the tools I have access to can handle) when you change 
switch/network configs and server configs at the same time.  We don't do that. 
(It'd be nice if we could, but we work with what we have.  I do believe the 
'transactional rollback' with unix and switch configs within the same 
transaction is a reasonably solvable problem.)

yes, binary rollback is still only half-solved in most environments (which is 
silly,  for example, just using the DJB package system makes it a trivially 
solved problem)   But certainly for configs, the 'rollback all / rollback none' 
strategy works fairly well.

OL> I'm probably in over my head here, but I imagine some application
OL> connectivity properties might be quite difficult to test if they're 
OL> fully internal (multi-tier apps), because naively I would think I can 
OL> test whether things are available to be called, but testing if 
OL> theyre actually being called appropriately might be harder. I would 
OL> need a sample incoming request stream to see how the config is 
OL> dealing with it. This isn't my area of expertise in any sense, but 
OL> when people talk about end-to-end behaviours I have typically 
OL> imagined that they meant something that would be difficult to
OL> code a shell script to run on a particular client to test for.

hm.  like I said, in my environment, it's a rare app that can't be tested with 
'telnet'  -   in one environment, we had the developer insert an "ok 
cookie" (a serial from the DB)  into a HTML comment;  we would alarm if 
that cookie wasn't there in the HTML output-  as the cookie was drawn 
from the same db as the real data, it was a pretty good indication of 
"backend bits OK"

we had another test account that was supposed to not change at all;  if 'diff' 
of a known good page and the current page returned  anything, my cellphone 
would get a text.  I would also get an alarm if a diff of the same page 
returned from different servers returned anything.

Personally, I think that before you even think about config management, you 
need a monitoring system that tells you about problems from your customer's 
point of view.  Monitoring is key to everything, and it's generally an easier 
problem than large-scale configuration management.

OL> I guess the question I was asking was: how will you drive this test 
OL> cluster? Where will the incoming request stream come from? I did 
OL> some work in distributed fault tolerance at one point, writing a 
OL> system to stress peoples' web services (the "trendy" WS-I/WSRF type) 
OL> and much of the work I did there was coming up with a plausible 
OL> request stream for these things. Maybe there's an easy get out in 
OL> this domain I don't know about.

The standard way to do it in my world is to capture live requests from 
production, usually with tcpdump (or if it's easy, some sort of 
application-level logging, but tcpdump is usually easier 'cause you don't need 
dev cooperation.  either way, we grab live data.)  and then replay the real 
load against the test server.  (verifying results can be complicated, but 
usually we just check for the appropriate return codes and a reasonable size 
and call it good enough... sometimes we add in "OK cookies"  and test 
those, or hit it with perl regexps.)

testing large systems using virtualization

I've just got myself a new job;  I'm in a group with something like 
a 10,000 to 1 ratio of servers to administrators.  This is by far the 
largest installation I've ever had root on.

What concerns me is that while I see some powerful automation, I don't 
see any safety nets.  "Don't screw it up" is the order of the day.  That's 
fine if you have people that are good enough;   but bay area labor is 
tight right now; (speaking of which; I get referral bonuses) they've hired 
me, and I see no evidence that I am an anomaly.  I need safety nets. 
and we're loosing the graybeards, we really need to put more effort into 
preventing a fat finger from doing an incredible amount of damage.  That, 
and I want to do some development on our tools, and I am a little 
frightened of cowboying it on this scale.  Now, my background is in 
virtualization;  I've been running a vps provider for the last two years, 
and I've spent the last year doing Xen consulting.

so this might be partly a "I've got a hammer, all problems look like 
nails"  thing, but it is a pretty nice hammer, so I think it's at least 
worth looking into.

does anyone else use virtualization to simulate massive systems for the 
purpose of testing configs?    What other approaches do other people use 
to install 'safety nets' that prevent an undercaffinated admin from 
taking out a huge number of servers?  what is the state of the art here?

Anyone used/tried OpForce?

Has anyone used or tried OpForce, also known as Vertias Configuration
Manager (now from Symantec, but previously known as "Clarity" from
Relicore)? If so, what did you think of it?

I'm trying to convince my management that they need to get Windows
servers out of the production system if they want me to give them the
automation of system installation/configuration/management that they
require, and one thing I'm claiming is that there's no good tool that
can do both Windows and *nix. Well, not do them the way we all
understand it needs to be done. (At least I don't have to convince them
they need to automate everything.)

So of course, one of the managers fires back with "Have you looked at
OpForce?"

Please reply directly to me and I'll summarize to the list.

Thanks,
AdamM

Reminder: ACM SIGCOMM INM Workshop deadline is April 26/May 3

Abstracts are due April 26 and papers May 3. The CFP is at 
http://research.microsoft.com/~mort/inm07/cfp.html

--

-- 
Sanjai Narain, Ph.D.
Senior Research Scientist
Information Assurance and Security Department
Telcordia Technologies, Inc. 
1 Telcordia Drive, Room 1N-375
Piscataway, NJ 08854
732 699 2806 (T)
908 337 3636 (M)
narain <at> research.telcordia.com

_______________________________________________
OPS-NM mailing list
OPS-NM <at> ietf.org
https://www1.ietf.org/mailman/listinfo/ops-nm

Fwd: Tutorial Workshop on System Configuration and LCFG

Just in case anyone is interested and will be be close enough to  
Edinburgh !

   Paul

Begin forwarded message:

> From: Chris Cooke <cc <at> inf.ed.ac.uk>
> Date: 5 April 2007 10:53:06 BDT
> To: cocentric <at> lists.ed.ac.uk, scicos <at> lists.ed.ac.uk,  
> medicos <at> lists.ed.ac.uk
> Subject: Tutorial Workshop on System Configuration and LCFG
> Reply-To: Chris Cooke <cc <at> inf.ed.ac.uk>
>
> The University of Edinburgh is holding a one day tutorial workshop  
> on System Configuration and LCFG on 13 June 2007.
>
> "System Configuration" is the process of transforming a collection  
> of raw hardware and software into an integrated "system" which  
> continuously and correctly serves a specific function - for example  
> a computing "cluster" or a student laboratory.  LCFG is one tool  
> which helps to automate this process.
>
> The workshop will introduce the underlying principles of system  
> configuration, and will show how LCFG can be used to help automate  
> the process. Throughout the day talks will be interspersed with  
> hands-on sessions including the creation of new components, as well  
> as the configuration of real systems, using components provided  
> with the LCFG distribution. You will also be shown how to extend  
> LCFG for your own site and given some pointers on starting an LCFG  
> installation.
>
> To get the most from the day you should ideally have some  
> experience of system administration, especially on Unix-like  
> systems, and of scripting in either Shell or Perl.
>
> The presenters will be Paul Anderson, Kenneth MacDonald, Stephen  
> Quinney and Alastair Scobie.
>
> For more information see:  http://www.lcfg.org/june07
>
> 	-- Chris Cooke.
>
> Computing Officer, School of Informatics, University of Edinburgh.
>
>
>