Thursday Jul 03, 2008
Hacking around with luupgrade on a Nevada SXCE (OpenSolaris) server when /opt is a ZFS filesystem but the root file system is UFS. In this example the file system (mounted at /opt) is called optpool:
YMMV...
comments, corrections, improvements welcome...
Disable services that use /opt and any others that make sense.
Backup /opt (e.g. with rsync)
Unmount (zfs unmount) the optpool filesystem
Run luupgrade
Change the mountpoint property (zfs set) of optpool to a temporary mountpoint. And make sure it mounted there, or zfs mount it manually.
Go to that mountpoint and rename directories that need to be moved out of the way, e.g. perhaps SUNWmlib.
Mount the new BE (lumount)
Note that luupgrade, being a bit confused, probably created a new /opt and installed some stuff there. Copy that stuff over to where you temporarily mounted optpool. E.g:
cd /.alt.new_be/opt && /bin/tar -cpf - SUNWmlib | (cd /tmpmountpoint && /bin/tar -xpvf -) # YMMV
Repeat as necessary, some possibilities (i.e. these showed up in /opt in a b64a->b88 luupgrade for example): SUNWtvnc SUNWvgl SUNWjavadb TurboVNC VirtualGL
Now do luactivate and reboot (init 6)
Move (mv) the bogus /opt directory out of the way.
Change the mountpoint property (zfs set) of optpool back to /opt, and make sure it mounted.
init 6 again
Review the state of SMF services, svcs -x, etc. of course.
Friday Jun 06, 2008
If you subscribe to this blog, you might want to unsubscribe.
I'm not sure what to do about this space. I've quit paying much attention to it and I don't miss it. I don't read blogs much any more either. I've replaced that part of my informaion consumption with listening to netcasts (aka podcasts) on-the-go -- i.e. on my mp3 player/phone (I recently bought a Nokia 5300 in case you're curious).
Also, I guess, there are aspects of blogging that leave me feeling cynical about it. Sometimes really cynical. It sure seems look-at-me'ish and quest-for-hits'ish and monologue'ish. Doesn't it? In other words, not very social.
(By the way, at times, I've been just as guilty of that here as anybody. In fact, one could argue that I'm doing it with this post. Though like anything, it's a matter of degrees really.)
So I thought about this space, and narrowed it down to 3 choices: delete it, leave it as-is (an archive of old posts), or mirror my other (i.e. non-blogging) posts here, such as Twitter. I'm not sure which one I'll eventually settle on, but I'm going to try the latter first and see if that feels right. My theory is that because blogs.sun.com is high on google's radar, people who might like to participate in the public conversations that I participate in will have a better chance of finding them.
Friday Jun 06, 2008
Fri Jun 06 2008
@stevel re: vimperator ... Nice! Can't believe I hadn't heard of it -- otoh, it only started a year ago. Anyway, thanks for the pointer.
Tue Jun 03 2008
@erwann The part of my life that I increasingly, but reluctantly, spend in a browser may have just taken the biggest leap forward ever.
Tue Jun 03 2008
@erwann http://goosh.org/ Holy shit! Does that actually work as advertised? (I just typed help)
Tue Jun 03 2008
Letting the Chicago OpenSolaris mailing-list know about this: http://tinyurl.com/6kugry
Mon Jun 02 2008
@stevel yeah yeah, someday I'll switch. I s'pose I should put-up & write a send/recv wrapper that mimics the CLI and actions of rsync -a ..
Mon Jun 02 2008
Here, zfs beats rsync: http://tinyurl.com/64tkqf, and for servers I admin someday I'll switch, but rsync's KISS advantage still wins for now
Mon Jun 02 2008
Click on the May 31st performance.
Mon Jun 02 2008
Video of my wife's barbershop quartet @ Kennedy Center: http://tinyurl.com/5dy89x. Fast-forward to 33:15 though (past the girls choir)
Mon Jun 02 2008
http://tinyurl.com/5tcbwb - Chicago user group guy reporting strange 'su - postgres' problem on (presumably) Solaris 10.
Sat May 31 2008
They won 1st at the international women's barbeshop (aka sweet adelines international) contest this year.
Sat May 31 2008
Live webcast of my wife's a capella quartet at the Kennedy Center in DC: ...
Thu May 29 2008
Working with lucreate et al. to prep an opensolaris.org backend box for an upgrade.
Thu May 22 2008
@davest Beer is the base, plus lemon or lime, and usually on the rocks.. then it varies. Here's Wikipedia's take: http://tinyurl.com/66a2lb
Thu May 22 2008
@davest If you're staying the W, across the street & left a block or so is pub where I discovered micheladas -- "hand-crafted". I'm hooked.
Thu May 22 2008
Alternatively, as I mentioned earlier, one's following number should reflect reality, not a silly wink-and-a-nod favor exchange.
Thu May 22 2008
... It is REALLY refreshing how Pirillo and others blow up that silliness by just auto-following everyone. ...
Thu May 22 2008
Instead, what we have a lot of is this annoying, mostly self-aggrandizing "my followers:following ratio is bigger than yours" ...
Thu May 22 2008
... that is, if the number of people you TRULY like to follow (eg via devices/IM) is say 40, shouldn't your follow number be near that? ..
Thu May 22 2008
... Personally, I like that. Here's why: Given the follow/following "measure" tends to evoke the bad kind of self-centeredness in people...
Thu May 22 2008
Twitter noobie here wondering about follow etiquette: Q: Why do some tech celebs, like Chris Pirillo, just auto-follow everyone? ...
Friday Oct 12, 2007
Dear Joan -- you're amazing, you're wonderful, and now you're a World Champion Quartet Tenor!
Congratulations Joan and the incredible Four Bettys quartet -- 2007 International (yes, that's worldwide) Quartet Champions!
Pictures:
Monday Oct 08, 2007
The following was actually intended as a comment here:
What is Project Indiana? Explained
But I couldn't get the submit button to accept my comment for some reason, so here it is...
Ben, I agree with John Levon here. Allow me to use gmake (as you mentioned) as an example to illustrate:
My understanding is that Indiana does not provide a way to put gmake (for example) anywhere other than where its been reviewed/approved to be. What I think you're referring to is the ability to customize how add-on software is added on. But gmake isn't an add-on. It's In, with a capital `I'. Gmake on OpenSolaris lives in /usr/bin, and that's just as true for the forthcoming (Indiana-spawned) new distro as it is today for (its predecessor and mentor) Solaris Express.
Of course a separate project could roll its own separate version of Indiana that eschews one or more ARC decisions, e.g. taking gmake out of /usr/bin and putting it somewhere else, but that's no different than what can be done today. I believe Belenix currently has it in /usr/foss/bin for example.
Monday Oct 08, 2007
An exciting event will be happening later this month: Indiana's new OpenSolaris distro is due for its first preview (milestone) release. Also, the OpenSolaris/Indiana Developer Summit is happening this weekend.
With this in mind, allow me to post a reprise or two of some of the core issues that were on a lot of our minds a while back as this new initiative struggled to find its footing.
Probably the most core of the core issues is this: Is the Indiana project a true collaborative effort? There are many shades to that question, but overall, I believe it is. I base this on the the following two things. (Note, the second rests on the first.)
Monday Jul 23, 2007
With DTrace-like and ZFS-like impact, I believe this forthcoming packaging/installation project will frickin' shake the UNIX/Linux earth.
It's going to single-handedly define Indiana.
I've been talking about the importance of packaging systems on this blog for years. See here or here for a sampling... or click on one of those fat links in my tag cloud. Long story short: Package management is foundational to distro popularity and growth. For Linux (or *BSD) laptop/desktop users, the biggest differentiator of their preferred distro is the feature that gives them network-based package management. Period. Nothing else comes close.
The thing about Indiana (in my opinion, which is hardly a secret any more) is that on the heels of things like Nexenta, BeleniX, and Solaris Express (OpenSolaris) Developer Edition, the novelty of the Indiana ideas, especially for the first month or so, weren't exactly something to write home about. But now Indiana literally equates to the new installation/packaging project... at least to my way of thinking.
Something earth-shattering is indeed going to happen.
The new OpenSolaris installation/packaging system...
For The Win!:-)
Monday Jul 16, 2007
Prompted partly by anticipation (and delight) of the distro constructor project being discussed on indiana-discuss recently, I bought an "economy model" laptop over the weekend. The point was to find a barebones (cheap), 32bit laptop to serve a an install/packaging/etc sandbox, and since I researched quite a few models, I thought I'd post a write-up of some of the component specs I uncovered (in this price-class) and related thoughts.
My bargain-hunting method was the weekly-ad, loss-leader, in-store-purchase routine. The retail price (MSR) range turned out to be roughly $550-$700 USD. The price range after discount and mail-in rebate (though some were available w/out rebate) was roughly $400 - $450.
So without further ado, and with corrections and other improvements highly encouraged...
Key:
PDC = Pentium Dual Core
AMD = AMD Sempron 3500+
PCM = Pentium Celeron M
GMA = Intel Graphics Media Accelerator 950
VIA = VIA Chrome9 HC IGP
ATI = ATI Radeon
Y/N = Does the optical drive write DVDs
Models:
Compaq Presario, C502US: PCM, GMA, N?
Acer Aspire 3680: PCM, GMA, N?
Compaq Presario T2080: PDC, GMA, Y
Everex StepNote VA2000T: PDC, VIA, Y?
* Toshiba Sat. A135-S2326: PDC, ATI, Y?
Acer Aspire 3100: AMD, ATI, ?
I wound up getting the Compaq Presario T2080 at Best Buy.
The asterisk on the Toshiba means it is purported to have Atheros wireless. Most, or all of the rest have Broadcom. (I think that means the NDIS driver is required?). To narrow things down, first I decided to stick with GMA graphics based on something I read in laptop-discuss. Unfortunately that ruled out getting Atheros wireless. That left the 2 Presarios and the Aspire 3680, and I figured the Dual Core Pentium would be better than the Celeron M.
The Sunday ads I perused for this were Best Buy, Walmart, Circuit City, Office Depot, Staples, and CompUSA. Here are the stores that I know currently carry each brand. This is way incomplete though... IOW, no doubt each brand is carried at more than the stores listed here...
Aspire: Walmart, Staples ...
Presario: CompUSA, Best Buy ...
StepNote: Walmart, Circuit City ...
Satellite: Office Depot ...
Some more caveats off the top of my head...
- Advertised specs are often wrong.
- Cheap laptops are big and clunky.
- Installed RAM was either .5 or 1GB, but I didn't keep track of which.
- I think the HDD was 80 GB in all of them, but I'm not sure.
- Some of these might be available for on-line purchase only.
- I think one or two of them has a 14.x-inch screen.
- Loss-leader shoppers are frowned upon by electronics stores, be prepared to be treated that way.
Model names/codes can get really tricky (e.g. "Toshiba Satellite A135-S2326") and are probably not valid outside the U.S. So it might be helpful if anybody has knowledge of equivalent model names in other countries.
So now I have my nifty new cheapo... er... economy lappy but I haven't had time to do much with it yet! What I did get a chance to do though was download and fire up OpenSolaris (Nevada b67) via the BeleniX
0.6.1 LiveCD (also, I stuck with Xfce and the Realtek ethernet NIC for now) and everything, including LAN/WAN/DHCP discovery worked great!
Sunday Jun 24, 2007
Just thought I'd blog a copy of my recent input to the "Indiana" (New Binary Distro) requirements gathering task (being led by David and Glynn, per my last post).
By the way, in the interest of focusing discussion it seems like a good idea, as I've done here, to collect and consolidate ones votes on this subthread.
Date: Fri Jun 22 08:43:23 2007
From: Eric Boutilier
Subject: Re: [indiana-discuss] Indiana Problem Statement
...
On Fri, 22 Jun 2007, Glynn Foster wrote:
>
>
> David.Comay at Sun.COM wrote:
>> Glynn,
>>
>> I have a couple of questions on the specific requirements below. On
>> general question is are the requirements within a section (such as
>> "Installation") ordered by priority?
>
> Some of the requirements probably need some thought about the order - especially
> those requirements that go hand in hand with having a useful system ie. you
> can't really have one without the other.
>
> In the case of Installation, I believe the current ordering to be ok.
>
>> Also, is there a general prioritization here or could priorities be
>> assigned to the individual items as you see them? For example, are
>> there a set of items that are "must haves" for a meaningful, first
>> Indiana distribution?
>
> I believe having a set of requirements that provides a good step on the ladder
> without shooting ourselves in the foot on subsequent releases is obviously
> preferred. However, I'm just as open to suggestions for how to prioritize. As
> I've mentioned before, I'd be happy for a first release if we only had an
> install and package management infrastructure that was developed to the stage
> where it would be possible to extend later on. I would personally rate those 2
> sections higher than some of the others because in essence, they are enablers
> for future work.
+1. In fact this would get my strongest support, FWIW. Namely, that the
1st release focuses mainly on two things: network repository based package
mgmnt and installation. (More below...)
>
>>> 3.1) Installation
>>>
>>> INS-1: Provide infrastructure to install Indiana from
>>> a single CD. The default install should be a
>>> graphical experience, with a text console install
>>> if preferred. The core install will be a basic
>>> operating system and desktop environment.
>>
>> There appear to be two distinct requirements being called out here.
>> The first, let's called it INS-1a, is around an installer that's build
>> around a CD rather than a DVD. The second, let's called it INS-1b,
>> concerns the default set of components/functionality delivered with the
>> CD. Is that correct?
>
> Sounds reasonable.
>
>> Concerning INS-1b, what are the specific features that the installed OS
>> should provide? I realize the common thing to do will be to use
>> something that satisfies PKG-2 to download other components but is
>> there an expectation around this initial installation? Some examples
>> here include: which (productivity) tools? which languages? which
>> hardware drivers besides the obvious ones concerning the disk and
>> network?
>
> I had hoped to work on an initial list of packages that would be available on a
> default install. I haven't unfortunately got to that work, but if we could
> assume that it includes everything from a core install, desktop environment and
> associated dependencies.
>
> I would personally be basing that work heavily from
>
> http://www.gnome.org/~gman/ubuntu-pkg-list.txt
>
> Aside from a few things (which presumably are part of the current package
> dependency list), it's a pretty good start.
+1.
> [As an aside, I was absolutely *shocked* when I put my Ubuntu CD into a running
> Windows environment to see they had additional space to add open source
> components that could be installed and run on Windows too. I'm seriously
> impressed with their ability to cram all that on a single CD]
>
>>> INS-3: Any installation should be aware of existing
>>> installed operating systems, prompt the user
>>> for an appropriate action and intuitive
>>> steps to carry out that action, particularly
>>> with respect to re-partitioning if
>>> necessary.
>>
>> Some additional clarification of this requirement would be helpful.
>> How integrated should this repartitioning tool be with the installer
>> itself? Which existing operating systems should the installer be aware
>> of? Which partition types need to be supported on x86/x64 systems for
>> installation (and subsequent boot?)
>
> No, I don't believe it's necessary to have a repartitioning tool with the
> installer. The installer should be nominally aware of Windows and Linux at the
> very least. I honestly don't know what the answer to the last question is -
> probably nothing more than what is currently supported in Solaris.
I vote to make it a higher priority. That is, given how
useful a repartitioning tool is, plus, given the low "cost" of
including one (assuming we could borrow the one Moinak developed).
>
>>> INS-4: A LiveCD should be available for a 'try first,
>>> install later' experience. The LiveCD functionality
>>> should be integrated into the core install CD.
>>
>> Given how little a CD can hold, how important is that the "live" media
>> actually be a CD versus something like a DVD? If the requirement is
>> really around a CD, again what are the attributes/features that the
>> live environment should contain? Which tools, languages, drivers,
Q: As with Network Auto-Magic (below), given that INS-4 (LiveCD) is already
part of OpenSolaris via the Live Media project, are there any specific
enhancements desired?
Regarding CD vs. DVD: I agree with those who would make CD a high priority
for the 1st release. I'm pretty sure (but admittedly not totally sure) the
BeleniX user community experience would support this position as well.
Also, e.g. citing Alberto Ruiz yesterday: "much much more people than
expected don't have DVD readers yet, " ...
> I believe crucial as a long term goal is to see the LiveCD be part of the
> install CD. However, that shouldn't stop anyone being able to construct a
> LiveDVD themselves with additional software available. The LiveCD does not need
> to contain any additional software that the install CD doesn't already have in
> my opinion.
>
>>> 3.4) Laptop Support
>>>
>>> LAP-1: Provide a utility to generate feedback for system
>>> configurations that are being installed, and build
>>> a basic public hardware compatibility list. Provide
>>> drivers for the most popular configurations.
>>
>> There also seem to be two distinct requirements specified here. The
>> first, let's called it LAP-1a, seem to express a requirement for a tool
>> which generates a system report and makes it available to be used to
>> build a HCL. Or is it to have a tool like the Sun Device Detection
>> Tool run before or during installation to identify whether the target
>> machine is suitable for a Indiana installation?
>
> I believe the former. While the current Sun Device Detection tool is very useful
> in informing users about whether it is likely that there will be drivers to
> support their system, I believe the former is useful for collecting basic
> information about what our users are running on, and as such developing a set of
> priorities for hardware types that we may want to write support for in the future.
+1.
>
>> The other requirement, let's called it LAP-1b, seems to call out the
>> need for more drivers. If that's the case, then there needs to be much
>> more specific requirements around which devices are important.
>
> This should be considered a long term goal. Perhaps the requirement is to gather
> the list of 'known' currently available drivers that are suitable to ship with
> the distribution.
+1.
> Does this narrow things down, or do I need to do some investigation?
>
>>> LAP-3: The system should automatically connect to an available
>>> network, whether through a network cable or wireless
>>> connection.
>>
>> Given that Network Auto-Magic Phase 0 is already part of OpenSolaris,
>> what specific enhancements are meant by this requirement?
>
> Possibly none. Given that we have this infrastructure, and roadmap, I think
> including it 'as is' is fine.
+1.
Thanks,
Eric
P.S. +1000 to the progress this new sub-thread represents!
>
>>> LAP-4: Any system connected up to an external video projector
>>> should work without needing added configuration.
>>
>> "Any" covers a awfully lot of systems. :-) Is there a list of the more
>> important ones given that we're talking about video drivers here?
>
> This is similar to LAP-1b. I think we need to do what we can in the short term
> until we start generating better metrics for what systems our users are
> installing on. Sorry for it being so vague.
>
>
> Glynn
Thursday Jun 21, 2007
Are you wasting time/effort worried about incompatibility
of distros? Perhaps you need to come up to speed on one of the
most profound differences between OpenSolaris and Linux
(in my humble opinion of course :) ).
(The rest of this is copied from a reply I wrote on
indiana-discuss)
In short, in OpenSolaris land, the existing standard base of compatibility
is a completely different animal than Linus' is:
In terms of Linux, the attributes of the base (Linus' kernel project) led
to all those incompatible distros. By contrast, the intrinsic attributes of
our existing base and the environment from which it generates (which I
trust you've noted are radically different/better in this regard) leads to
far more, though admittedly not absolute, unification and compatibility
across distros and other projects, IMO.
Sunday Jun 17, 2007
OpenSolaris Posts and Discussions
June 6th - 10th, 2007
Friday Jun 15, 2007
OpenSolaris Posts and Discussions
June 13th - 15th, 2007
Friday Jun 01, 2007
Per my last post, I'd like to describe what I mean when I say non-Sun-derived (i.e. Community-derived) project.
But first, make no mistake. I believe -- as everyone does -- that
it's perfectly fine for many/most OpenSolaris projects to
be totally Sun-derived, Sun-designed. But not this one -- not the project that wants to lay claim to the title OpenSolaris reference distro!
OK, here's my definition of a non-Sun derived OpenSolaris project:
The design goals
(both initial and ongoing) of the project are derived a.) independent of
the desires of all levels of Sun management, and b.) independent of Sun's
Business goals, leading to a broadly held perception (by reasonable,
open-source savvy people) that the project's course is fundumentally not
Sun-management-controlled.
Regarding origin and impact of project
resources: if Sun pays for resources for a non-Sun derived OpenSolaris
project (per my definition above), that influences things like program
management, speed and quality of project execution, marketing, etc. (IOW,
definitely not the project's power to ultimately self-arbitrate.)
So that raises this question: What's the Wallstreet business
case in that? Why would the board and other shareholders
want Sun executives to direct lots of money in this way?
-
In the software world, volume wins.
-
Success of an open-source software project builds volume,
and does so in a way that no other software strategy can
match (not even close).
To my knowledge (for what it's worth), the two bullet
items above are core tenets of SUNW's software business
plan. Therefore, because an excellent OpenSolaris project
with excellent core contributors is excellent for
OpenSolaris, Sun should invest tons of resources in both kinds of projects (Sun-derived and Community-derived), even though Sun cannot ultimately
control many of them.
Friday May 25, 2007
OpenSolaris Posts and Discussions
May 22-24, 2007
Monday May 14, 2007
OpenSolaris Posts and Discussions
May 10-14, 2007
|
|
|
|
YMMV, but if you're not going to upgrade anyt...
@mads: I used lumake in this case, which doesn't h...