FreeBSD RAID

I just finished reconfiguring my network file server - a long process.  I'm running FreeBSD, and the geom utilities for RAID configuration.

I have four disks - 3x 320GB, 1x 1TB.  I placed the 3 matching disks into a stripe, or RAID0.  This RAID0 disk is then in a mirror/RAID1 with the 1TB drive.

Configuring the stripe & mirror is done with the gstripe and gmirror commands.  Since the array is also a boot/root disk, I have entries in /boot/loader.conf:

geom_stripe_load="YES"
geom_mirror_load="YES"

These instantiate the stripe & mirror prior to reading fstab and completing boot.  Also /etc/fstab is set to the array device:

/dev/mirror/gm0b        none            swap    sw           0       0
/dev/mirror/gm0a        /               ufs     rw           1       1
/dev/mirror/gm0e        /tmp            ufs     rw           2       2
/dev/mirror/gm0f        /usr            ufs     rw           2       2
/dev/mirror/gm0d        /var            ufs     rw           2       2

The 1TB drive, ad7, contains 1 slice ad7s1.  The gm0 mirror is between the stripe, st0, and that slice.  I suppose I could have added the entire ad7 to the mirror instead.  ad7 is larger than st0, so in that situation gm0 would have been created with st0, and then ad7 would be inserted.

gmirror bases the size of the mirror on the first device it is created with:

gmirror label -v -b prefer gm0 /dev/ad7s1

In this case, I built gm0 with ad7s1, then inserted the stripe:

gmirror insert gm0 stripe/st0

One caveat is, when building a mirror on a slice (like ad7s1) it must be mounted read-only or not at all.  If you build it on a device (like ad7) then you can simply use

sysctl kern.geom.debugflags=17

The system will then let you write to the disk metadata.
To build on the root slice, I rebooted into single user mode, which mounts root as read-only.

I like this setup.  I've got RAID1 redundancy, and the 3-disk stripe is quite peppy.  Lastly, to take advantage of the speed of the stripe, I set the priority of the stripe in gm0 higher than ad7s1.  Since gm0 uses a 'prefer' balance mode, it will read from the stripe first.  I believe writes complete on the stripe as well, but I'm not certain.  I'm getting decent write speeds with dd commands.  I'll have to do a more thorough performance test some time in the future.

judgin'

alot of judgements that are made by judges are controversial;  some say you find the right judge, and you can have whatever ruling you desire.

how about we replace the judge with some crowdsourcing.  take the facts of the case, and devise them into a test.  present the facts on a website as they are to the judge, and allow potential "judges" to take the test.  if they pass, proving understanding of the case, then they can submit their judgement.

multiple judgements can be statistically combined to arrive at a final judgement.  of course these "judges" would have to have equal familiarity with case law to make comparable judgements.  the system could be applied to non-technical judgements.. or only basic, straight-forward questions can be asked.

this could work well for something like patent processing.

now that i think about it, crowd-sourcing has so many applications.  it seems ripe for an open architecture to be developed for specific applications to build on.  wonder if such a thing already exists.

barriers to entry

there are many barriers to entry in any sector of the economy, and those hinder competition.

a universal barrier is money, and its availability in most cases is dependent on lenders.  someone with an idea must make a proposal to a lender, and the odds of lending to a legitimate prospect depends on the risk and potential return in the lender's view.

an accurate prediction of the involved risk requires a comprehensive analysis of factors in both the involved sector and the proposed idea.  perhaps a NDA (non disclosure agreement) is utilized at this juncture to secure the entrepreneur's intellectual capital, but the lender has to agree to be bound to secrecy of whatever is revealed.  some lenders may not be willing to take that risk, as then they can't utilize any of that information in making future decisions (i think, though i don't know the legal intricacies)

plus, the entrepreneur has to trust the lender to abide by the NDA.  who's monitoring the lender?  perhaps if the government facilitated these exact exchanges, it would foster confidence in approaching lenders and encourage more innovation in the economy.

pizza and tires

katie & i shared our 5 year anniversary yesterday.  we went to pizza hut last night.  mm pizza.

i also had pizza yesterday for lunch at the pizza palace with my parents.  checking out the place before deciding to use it for a wedding rehearsal dinner.

i also had pizza the day before yesterday for dinner.  with my parents, since i was in smyrna picking up a package.  lots of pizza.

that package was a camber kit for my civic.  im going to correct the excessive stock camber in the hopes of achieving proper wear on my next set of tires.  the current set is incredibly, horribly loud at highway speeds.  quality engineering, honda.

the next set of tires are sitting in my truck currently.  the camber kit will be installed tomorrow, then i'll get the new tires mounted & balanced, and get an alignment.. so we should have a nice quiet ride to virginia for our honeymoon next month.

positive perspective on the context of this thought train

i find myself seeking guidance in my own state of mind and arriving at my consciousness repeating this sentence over and over.. not certain it has a deeper significance, but i suppose i must trust myself to some extent..

i must maintain a positive perspective on the context of this thought train..

knowledge acquisition

Maybe it would be possible to structure the development of concepts and ideas in the human mind to a manner such that it would permit the communication of other concepts and ideas in a nontraditional form of communication.  Such a form could be in some kind of technology integrated into our bodies.  Thus, rich communications could be automated and fed into a human mind.  Perhaps simply repeated input would allow memorization, so such a system could be used to efficiently infuse knowledge into people.

Talk about speeding up development of humankind.  That would be revolutionary.

universal symbols

Further consideration led me to the idea that there are many universal symbols in existance.  Such as a heart for love, a peace sign, etc.  Many corporate logos are universal in the companies they represent.  I wouldn't consider corporate logos characters per se, as they can't be combined to communicate many thoughts.

Interesting.  It seems the more ways a given symbol can be utilized, the more likely it is that I would consider it part of a language.  For instance, I could make up a word.  But since noone uses it but me, I wouldn't consider it part of the English language.  A popular word, on the other hand, I would.  A corporate symbol has too few legitimate uses in communication, so I wouldn't consider it part of language.  But yet there are words in the English dictionary with very few useful situations.  These probably shouldn't be considered part of the language either, but these guys printing the dictionary are quirky, aren't they.

There is an interesting parallel to all of this in the concept of data compression.  The goal is to create a minimal character set to represent data in the smallest space.  Ahah, this realization makes my earlier blog's consideration of ways to make communication more efficient rather special.

It seems the idea of making communication more efficient is a very good goal.  So then, the embedded nature of culture and language is seen to be a bad thing.  Culture resists change, but more efficient communication requires the language to change.

I bet there are dudes that research what an optimal form of language is.  Commence www search..