Array stays online while parity is syncing.

 

So I guess somehow the system handles writes. How?

Does parity stop the ongoing pass and go and fix the new bytes, or are those somehow "marked" for resyncing after normal sync is finished?

Something else?

(forget the easy way out now with cache disk - what happened up to today or for systems without cache disk)

 

I suspect system performance is low during this process right?

 

I couldn't find anything on this subject.

 

(realized that I start way too many threads... maybe because I was building my unRAID up to today - now parity is building, I should be ok in about 10 hours...)

Array stays online while parity is syncing.

 

So I guess somehow the system handles writes. How?

It writes the parity based on the normal steps of reading both the old and new data and parity for the block being written and then calculating the new parity value.

 

If the parity block has already been correctly calculated as the parity process proceeds through the disk blocks of data, it will be updated correctly.  if the parity disk still had garbage and had not yet been calculated, it would initially use the "garbage" that was there, and then, when the parity sync process got to that same block calculate a correct value based on the data on the other drives.

Does parity stop the ongoing pass and go and fix the new bytes, or are those somehow "marked" for resyncing after normal sync is finished?

Something else?

It does not need to do anything special other than do each block of bytes in turn on the disks.

(forget the easy way out now with cache disk - what happened up to today or for systems without cache disk)

Cache disk has nothing to do with how parity is initially calculated.  It is simply a convenient place to temporally write data until it is moved into place by the mover script. It is nice in that it is integrated with the "user shares" so files there show up in the proper folders, even if not yet protected from a drive failure. (if cache drive fails before the files on it are moved, they are lost)

I suspect system performance is low during this process right?

Not as bad as you might think.  For most people, you will never know a parity check is going on, or even that a drive has failed and the other drives are all working together to supply you data..  I've served 4 DVD ISO images from a "broken" disk, while rebuilding that same disk from parity.  It worked just fine.  I'll never be playing 4 different movies at the same time normally.  (two at max...only if my wife and I cannot agree on what to watch)

 

(realized that I start way too many threads... maybe because I was building my unRAID up to today - now parity is building, I should be ok in about 10 hours...)

You are fine.  Glad things are settling down.  with luck your array will be around a long time serving files.

 

Joe L.

Thanks man.

 

For the cache. I know, I didn't say it has anything to do with cache, only that because of the cache disk's existence, now parity will be touched less (only when the mover runs).

 

For the parity that is building on my system... well this didn't deserve a post by itself (as I don't have a syslog about it, so people will whine about my newbie feedback), but when I got home, I checked my unRAID's local terminal (just because it is easy - being one of my KVM positions), then went to my Windows machine to check on parity's progress (it had started 3 hours before then) by double clicking browser's shortcut (that goes to http://my-unraid-name/main.htm)... and guess what?

 

I GOT A "SHUTTING DOWN" message!!! I thought it was wrong, I refreshed the page (which again I repeat was main.htm, NOT some shutdown link), only to see that web was already dead. Popped to unRAID (local) terminal to see that indeed system was shutting down!

 

Why? Any idea?

 

What I did next? (after I switched back on, only to start parity from the beginning) I changed all my web links to my server to be WITHOUT main.htm in case there is SOME risk with this.

 

Any idea? (without syslogs or whatever)

 

Anyway, I expect it will be finished by morning (I am at 25% and is midnight now here) and everything will be well and job done.

I am at 196mil. KB, and at 312mil. parity will handle four less disks (my 320GB) and possibly speed up, then at 500mil. mark it will just handle one 750GB disk.

 

In case someone reads Greek, here is my blog entry about my "odyssey" with the whole project:

 

http://nulusios.blogspot.com/2008/04/nas-or-something-like-that-part-ii.html

 

In the end I DO propose that people DO install an unRAID (my odyssey is just a guide of what not to do... except maybe the internationalization problem).

 

 

 

In case someone reads Greek, here is my blog entry about my "odyssey" with the whole project:

 

All Greek to me NLS but I liked the bit about 'roasting Tom' 

All Greek to me too... Maybe you should update the internationalization settings  (j/k)

(now if I could only "really" read it! LOL)

hehehe

 

Paul that part is somewhat correctly translated.

I pressed on Tom to include ntpd as my mobos clock doesn't seem to keep the time.