Jump to content

Mover Spinning Up All Drives???


Recommended Posts

(Running unRAID 4.7)

 

So not that I have my unRAID server up and running I'm going through the tedious process of moving data from my old WHSv1 NTFS drives to the new file system.  I have a 2TB cache drive on my server, so when I copy files to the server they go straight to the cache drive as expected.  When I invoke the mover script, or it starts as set, the data starts moving off the cache drive to the data disks within the unRAID array.  But what surprises me is that all 11 of my data disks are spinning up even though a single drive is being written to.

 

At first I had my spindown times set to 15 minutes.  Then I noticed that when moving a very large file that takes more than that spindown down (like a 40GB+ Blu-ray ISO) the non-used drives would spindown.  But as soon as the next file started moving all drives would spin up.  Clearly I don't want the drives all spinning up and down that frequently so I reset my spindown times to 30 minutes so that now the drives are all spinned up until the mover process is finished.

 

So to my question, am I correct it believing that as each file is being copied the unRAID system/driver is polling all drives to see where they stand relative to the high water mark?  Is that normal system behavior?  I'm sure that once I've got all my data moved this won't be a big deal since I won't be moving 1TB+ chunks of data off the cache drive on a daily basis, but I just want to make sure I'm not missing something in the way I have things setup.

 

Thanks, Kevin

Link to comment

Do you have spin-up groups enabled?

 

Yes they are enabled but I have each drive assigned to its own group, where group=<disk name>, e.g. "parity", "disk1", "disk2", etc.  I did that after I noticed multiple drive spinning up when accessing a file.  I thought it was behaving correctly, but maybe not???

Link to comment

Try turning spin-up groups off.

 

OK so here's what I just did.  Mover was in the middle of copying a 37GB ISO from the cache disk to disk12 in my 20-disk array.  I disabled spinup groups as suggested.  I then did a manual spindown of all disks except parity, disk12 and cache.  There was still about 30GB to go on the file after all disks were spun down.  So I refreshed the myMain window of unmenu periodically while the file move continued.  As soon as mover completed the rsync of the 37GB file to disk12, there was a pause in all activity in the myMain window with respect to disk reads & writes.  And then lo and behold as soon as the next file transfer showed up after a page refresh in the "System Log (last 15 lines)" section of the screen there were other drives that had been spun up.  After a couple more refreshes all disks were once again spinning.

 

I'm stumped.  This does not seem like "normal" behavior to me, but then again I've only been using unRAID for a few weeks now.

 

Any other suggestions as to what I might try?  Or is this just the way it is?

Link to comment

Try turning spin-up groups off.

 

OK so here's what I just did.  Mover was in the middle of copying a 37GB ISO from the cache disk to disk12 in my 20-disk array.  I disabled spinup groups as suggested.  I then did a manual spindown of all disks except parity, disk12 and cache.  There was still about 30GB to go on the file after all disks were spun down.  So I refreshed the myMain window of unmenu periodically while the file move continued.  As soon as mover completed the rsync of the 37GB file to disk12, there was a pause in all activity in the myMain window with respect to disk reads & writes.  And then lo and behold as soon as the next file transfer showed up after a page refresh in the "System Log (last 15 lines)" section of the screen there were other drives that had been spun up.  After a couple more refreshes all disks were once again spinning.

 

I'm stumped.  This does not seem like "normal" behavior to me, but then again I've only been using unRAID for a few weeks now.

 

Any other suggestions as to what I might try?  Or is this just the way it is?

It is completely normal.   As files are moved, those data blocks replace the older ones in the disk buffer cache eventually using the ones used to hold file and directory information.  When mover next needs to scan the disks to move the next file it must re-read the directories to see where to put it. Since the blocks are no longer in memory, it must access the physical disks and as a result, spin them up. That then will spin up some or all of the drives.   If you move more file data then you have free memory, this will happen.

 

About the only thing that might help is running cache_dirs.  It keeps accessing the disks, hopefully keeping the disk directory blocks in memory so they are not the least recently accessed and subsequently re-used for more recent data.

 

(Or, you could add about 40Gig more memory, and have more than the largest set of files you move in one shot)

 

Joe L.

Link to comment

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...