[Tfug] Check file system and restore array

[Bexley Hall]

Hi Timothy,

> Been working on this cage for a couple years. Summer gets too hot, Then
> I seem to be more busy when it's cool. The first 2 floors have done for

Two *floors*??!  Yikes!

> some time and the cage is 2'x4'. The cat was born wild and we were
> trying to tame him, but he got hit by a car before he was tame enough to
> trap. We try to coax him out of the cage but he feels safe. Wants us to
> reach in to pet him and he plays with one of the other 2 cats through
> the wire, even when the door is open. So 3rd floor is just more room and
> climbing space for him.

Cage or "pen"?  (I think of a "cage" as a six-sided box; a "pen" as
four walls)

>> Be thankful -- we spent the night in the ER. <frown>
>
> Done that many times.

First time I have been there in the wee hours of the morning.
Actually think it is a better time to be there than "normal
hours".  Sure, fewer staff on hand -- but also less "customers"!

(Unfortunately, it was a tad chilly getting to and from the
car at those times.  First time I've had to deal with frost
on the windows since moving here!)

>> Of course, this is yet another change to the "failing" configuration
>> so adds yet another variable. Any reason not to use the SATA drives
>> from your father's machine -- and the same software configuration?
>
> His system is setup with 32bit linux and he may at some point decide to
> use it.

Yes, I recall you mentioning it was 32 (vs your 64).  But, that
wouldn't prevent you from pulling the *drives*!

OTOH, if he may use it, then silly to "cripple" it by doing that!

>> Why are you using RAID? (this is a DVR, right?) Is the "content"
>> that "precious"? (I don't think you would need it for bandwidth
>> reasons)
>
> It's not just the recordings. Those get deleted when the show is
> watched. It's getting it back up when a drive fails, plus I do use it
> for some storage. A sort of secondary backup for some things. Anyway, I
> am going back to the plan of restoring the array. Going to copy the
> programs, config files, etc., and some other data to the other pair just
> in case and then try to restore the array. While it might be good to
> start with a fresh clean OS install, I don't really want to have to
> resetup everything. I have a program install, to keep the clock set,
> Apache web is installed, Team speak server. Getting all the shortcuts
> put back, etc..

You might find imaging the disk (right after everything is set up)
to be a more effective option.  Certainly a lot easier to restore
an image than rebuild a RAID array.

I make an image of each system whenever it is at a "memorable state".
I.e., a point to which I would like to be able to restore the system
(if something gets corrupted -- e.g., Windows machines) or to
*rebuild* the system (e.g., disk crash, disk upgrade, etc.)

E.g., I am currently building a new "Documentation Workstation"
(Windows).  After finishing the Windows install (service packs, etc.),
I imaged the disk.  Then, after installing the drivers for the "bare"
machine (i.e., with no "additional" I/O's added.  Then, again, after
installing the umpteen gazillion Windows "updates".  Next, after
installing the additional I/O's and associated drivers.  Finally,
after installing the applications.

So, if I decide to repurpose the machine for some other use (i.e.,
different I/O's and/or applications), I can restore the image from
just before the "additional I/O's" step.  Then, add *different*
I/O's, drivers and applications.  (saves the time and tedium of
reinstalling windows, service packs, updates, etc.).

If, instead, I decide to put a larger disk in the machine, I can
install the new disk and then restore the "most recent" image
to the new disk.  Saves the trouble of having to partition the
disk, install the OS, drivers, apps, etc.

[If you take an image after you have finished setting all the options
that you want for all the applications, then you don't have to bother
restoring those settings, either]

Currently, I store images on external USB drives but you could
just as easily push the image to a file server on the other end of
a wire, etc.

>> If the drives have a "write inhibit" strap, you might consider
>> installing that and mounting them R/O -- just in case there is
>> something wonky in the software and/or system hardware. I lost
>> the *contents* of a disk to a driver bug in an early FreeBSD
>> release many years ago (and then promptly lost the contents of
>> a *second*, identical disk by repeating that same exercise! :< )
>
> I don't know of any write protect option on these. Even USB sticks don't
> have those anymore. And they call that progress. :( Haven't done any
> upgrades in at least a year, so any driver bugs would have shown by now.

Don't know what sorts of strapping options are available on SATA drives.
I primarily use different flavors of SCSI drives, here (and most/all
of those have write inhibit jumpers).

>> Things fail. As I mentioned before, even picking top shelf
>> components doesn't give you immunity from that. Only *you*
>> know what sort of use/abuse it was subjected to and what to
>> expect from it (longevity).
>
> The power around here drops to ~108v in the summer afternoons, from all
> the A/C's in the area is likely.

I wouldn't be surprised if you discovered the feed to your house was
*aluminum* (or, copper clad aluminum).

> But the power company can't do much
> about it because if they change the tap at the near by sub station, the
> voltage would be too high when the weather is cool. I try to run my
> computers only on UPS's now though the linux computer was run for some
> time without it when it was first built and power fluctuations are very
> hard on switch mode power supplies. My old UPS wasn't really enough to
> run two computers but I now have a nice Eaton 2500w rack mount I got
> from ebay new for around $150. So now all computer stuff, TV, etc are
> always on UPS.

I use small 1-1.5KVA UPS's -- one for each computer.  Also acts to
give me an "extension cord" with a main power switch so I can shut
down everything associated with a computer with one switch.

>> Again, the drives might not really be toast! There is a surprising
>> number of "dead" drives that actually are NDF (no defect found).
>> Like replacing a car battery when the problem is actually in the
>> alternator! :-(
>
> Yea, I don't think the drives are bad this time. It was a question of
> what it did to the file system when things went wonky. I never got that
> distinctive clicking they make when they went bad in the past this time.

Good luck!

Gotta go finish making my V-day candy...  <frown>

--don