Hi,

 

I confess to being a neophyte w/regard to Linux and to some UnRaid matters.  I would like to add a Cache Drive to my UnRaid which has been working "flawlessly" for ~ 3.5 years, but the steps seem to me to be arcane.  I would appreciate help with the following scenario if possible:

 

• 1) I plan to add a larger Parity Drive that has been PreCleared w/backup flash (thank you Joe L), current Parity Drive is 1.0 Terabyte; replacement Parity Drive 1.5 Terabytes - both SATA II.
  
• 2) I would like to remove 2 160GB PATA Drives and utilize 1 or both for the Cache Drive.
  
• 3) I would then like to add the current Parity (1 Terabyte) drive to the Array for Data to eventually become my primary I/O drive (main data storage over network).

 

Process ?:

Stop Array, shut down, add new 1.5 Terabyte Drive, boot Server, "Stop" array if started, format 1.5 Terabyte Drive, "Unassign" current Parity Drive (1 Terabyte) and "Assign" new Parity Drive, "Unassign" both 160 GB Pata Drives (data previously copied to another drive), Start Array w/Parity (check/rebuild).  After complete add old Parity Drive (1 Terabyte).  Does this seem correct?

 

If so, how do I then add one/both of the 160 GB drives back as Cache Drive(s).  I am using User Shares and have Movies, Music, TV Shows, and Data Server as my shares. 

 

Nothing that I've found provides actual direction of how to do the above and/or add the Cache Drive or the actual steps to follow.

 

Thanks in advance to all.

 

Dave

Maybe you're performing a few steps but not listing them. I THINK the way you have the steps listed, you would loose the entire contents of both 160GB drives. EDIT: Just reread your post and noticed you already moved the data off the 2 160GB drives.

 

The way I would approach it to contain the least number of stop/start cycles and retain the data until the new array is validated is:

 

A) STOP ARRAY

B) SHUTDOWN SERVER

C) ATTACH new Parity Drive

D) START SERVER

E) PRECLEAR new 1.5TB drive that will become the new parity drive

 

while preclear is going on the new drive, in parallel you can:

A) move the data off of the 160GB drives to the other data drives in the array

 

Once you are finished preclearing and moving the data, then you can:

A) stop array

B) remove 2 data drives and old parity drive from the array

C) assign new parity drive

D) start array

E) let parity calc finish on new array

 

Once parity check completes

A) start preclear on old parity drive and your future cache drive

B) stop array

C) assign old parity drive as a new data drive

D) assign future cache drive to array

E) start array

 

CAVEAT: I would verify this with someone who's done these steps more than I have, such as Joe L.

The process you use is greatly determined by the length of time you are willing to go without parity protection.

 

In general, you should NOT change/replace multiple drives at a time.

 

So... with that in mind

 

Before you do anything, get a screen-print of the devices page before you start any changes.

Perform a full parity check.  If any parity errors occur, post a syslog, do not pass "go" do not collect $200.

Don't do any changes until you can perform a parity check with zero errors.

 

Then, Stop the array, power down, install the new drives in spare slots if possible.  If not, only install the new parity drive.  If you have a spare slot, use it.  If you do not have a spare physical slot, but do have a spare port on the disk controller, connect it uninstalled physically so you can pre-clear it.

 

Once pre-cleared, Stop the array, assign the new parity drive instead of the old and press "Start" to compute parity on it.  It does NOT get formatted, as the parity drive has no file-system.

 

Once parity is calculated and stored on the new parity drive, perform a parity check on your array.  (The initial parity calc wrote to the parity disk, the subsequent parity check will read it to be sure it is readable.  Do not proceed until you can do a parity check on the new parity drive with no errors.

 

To replace a smaller data drive with a larger one

Stop the array,

power down

replace the drive

power up

Press "Start" to begin the reconstruction of the old data onto the replacement drive.  There is NO formatting step since reconstructing the old data drive also reconstructs the partitioning and formatting.

 

If all your drives are working properly, and you've copied all the data off the two smaller drives and you wish to remove them and not replace them at all, then you'll need to

Stop the array

un-assign the smaller drives being removed

PRess the button labeled as "restore" (It is poorly labeled. It is actually a Set Disk Configuration button.)  Pressing the button will immediately invalidate parity.  When you subsequently press "Start" a new full parity calculation will begin.  Again, when it is complete you will want to perform a full parity check to ensure parity written can be read properly.

 

At this point, you can stop the array a last time and assign the smaller disk as a cache disk if you think you need one. (most people do not, as 4.5.3 can write to parity protected disks as 40MB/s when used on good hardware without a cache drive.)

 

Joe L

Joe & Brit,

 

Thank you so much as always  !  It great to have experts like you to advise on proper procedures as I would definitely like to maintain my Parity/UnRaid safety.

 

Joe, I used your PreClear program on another "box" and it completed, though I'm not exactly sure what the report actually states.  I tried to go back up my console page but could not see the ending statement other than some "cryptic" comments about the drive.  I've not yet turned it off or closed/halted the session (PreClear) if there is anything I should review or send to you for analysis.  I did note that in the initial stages it was running ! 112-113 mb/s but towards the end during Post Read it was only ~ 63-65 mb/s and took slightly over 20 hours to complete (1.5 Terabyte - Seagate Barracuda 7200.11 ST31500341AS).  I've looked through your posts on the PreClear as well as the reports, but didn't find any notation of a "log" of results that I could post.  Assuming it is good (I have a copy of all the positions/placements/ids of current array), would you recommend that I stop the current array, take the current parity disk (1 TB) offline (unplug to protect), replace with the new one and re-build parity.  Then check before inserting the 1 TB back into the array?  Most of my drives are PATA, but the current/new ones are SATA IIs and are a bit faster.  I'm thinking of removing the 160s - because they are small and take up room + plus generate heat and block airflow a bit.

 

Again thank you both for your wisdom.

 

Dave

...stop the current array, take the current parity disk (1 TB) offline (unplug to protect), replace with the new one and re-build parity.  Then check before inserting the 1 TB back into the array? 

 

That is correct.  Always run a parity check immediately after running a parity build (which will happen automatically when replacing a parity disk).  The parity build writes all the parity info, the parity check makes sure that it can all be read back again.  You really don't know if you can trust your parity until you've done a parity check.

 

While I agree with Joe L. that cache drives are becoming less and less important in the unRAID environment, there are still many good reasons to use them, speed being only one.  I'm currently working on a thread about just that.

Rajahal,

 

Thanks for the comment, I had thought of using one of my older 160's but believe now that I would rather have the airflow.

 

Joe or Brit or Rajahal, in performing the initial parity run on the new drive and now (last 10 min) I have noticed a sequential beep with about a 3-5 sec pause.  I first thought it was due to elevated temps in the new Parity Drive, so I placed a fan directly over it bringing the temp down from 34C to 26C where it has remained throughout the process.  The "beep" sound seemed to stop (though I confess, sitting and watching/listening to my UnRaid check parity)was not a priority.  However, on the 2nd run to check the new drive for "real" the beeping again started shortly after commencement.  Any thoughts?  Incidentally the initial Parity run was successful albeit long (large drive). 

 

Also, when starting the program this morning, I moved the server to a location where I had an available monitor and noticed that my ProKey 4.5.3 would on occasion be bypassed (ie not recognized) on boot, thus not load the UnRaid Kernel, not sure if this is related but seems strange.

 

I'll let you know the outcome, but if anyone has heard of simlar problems that would be reassuring.

 

Dave

Quick Update:

 

• 1.5 TB now Parity & checked x 4
  
• 1.0 TB in Array (old Parity drive) - Total  3TB
  
• 2 x 160 GB Drives removed from Array
  
• Drives all re-ordered and Array Restored ~ 2 days total not including preClear
  
• Beeping has stopped - would guess possible power drain (too many drives or heat build-up) though still cannot confirm.
  
• Will repeat Parity Check one more time before replacing Array in headless location.

 

 

Thanks to all for help, for others with similar situations, Joe's & Brits  explanations above are wonderful and quite direct. Despite having had my unRaid since almost the beginning, I haven't had to tamper/adjust/or modify much!  For those on the Fence, just do it, it works simply and effectively!  THANK YOU TOM, JOE, BRIT, RAJAHAL ET. AL.   :D

 

Dave

 

If your beeping has stopped for no apparent, that's a bit weird.  As long as it is working, though, I wouldn't mess with it.  You may want to post one final syslog so that someone can give you the 'all clear'.

 

When I upgraded my server from 4.4.2 to 4.5.3, I noticed several more beeps during start up.  I immediately took them to be a 'feature,' as they seemed to coincide with normal unRAID events quite perfectly.

1st beep - startup, immediately after power on

2nd beep - array started, unRAID ready (web page also ready)

3rd beep (double beep, I think) - unRAID has received the power down command, and is shutting down.