Jump to content

Do disks being spinned up/down affect your choices?


settings

Recommended Posts

Just wondering is anybody else the same as me when it comes to using their unraid server.

When I want to watch something I will have a quick check of the unraid home page to see if any disks are spinning then I will either watch a movie/tv show on the same disk depending what the outcome is.

Another thing if say my brother wants a file I will put it on the cache disk to prevent me spinning up the parity and another disk in the raid.

Likewise if I only have a tv show or two downloaded I will wait to accumulate some more until i move them from the cache.

The main reason I do this is after reading that the most strain on the harddrives occur when they are being spun up. I have the default time of 1hr to spin down the drive. I would rather leave this as they are generally only used in the evening and wont be spun down until they are finished with.

Am I being ridiculous?  :)

Link to comment

I organize my files carefully on the system. I.E. I do not use usershares.

 

The problem with my environment is I have hundreds of thousands of files. directory caching does not work. There are too many files. Even with 8GB of ram.

 

So if I accidentally use a user share the whole environment spins up.

I tried the dir_cache mechanism. It kept all the drives spun up (even with 8GB).

 

So I resorted to carefully organizing my files.

 

New rips are on disk3 ALWAYS and that is spun up from when I get home till bed.

all my small files are on disk1, that is spun up from 6:00am till 21:00.

My parity and cache drive share the same pair of drives with the Areca controller.

They are spun up from 6:00am till 21:00.

 

If I want to watch certain TV shows. I login and spin that drive up only.

Sometimes I'll copy a whole directory to a temp area on the cache or a constantly spinning drive.

Other times I set symlinks to an index directory so only the drives I need to access for that application spin up.

 

Frankly speaking, most of my failures have not occurred due to spin ups.

They have occurred due to bad spots on the disk or a sudden power off.

This gets detected during reads and writes and has nothing to do with spinups/spindowns.

 

I know this, yet I still elect to organize my data.

 

My drives currently ALL spin up once a day for a find in making a file list.

I have not had a failure because of that yet.

Link to comment

I wrote a tool to keep my drives spinning regardless of the spindown selection

 

it uses an old program I wrote called istime which checks if the current time is within parameters.

If there is any interest I can provide the binary and shell script.

 

 

 

root@atlas /etc/cron.hourly #ls -l /etc/cron.hourly/keepspin.sh 
-rwxr-xr-x 1 root root 338 Nov 21 13:13 /etc/cron.hourly/keepspin.sh*

root@atlas /etc/cron.hourly #cat /etc/cron.hourly/keepspin.sh 
#!/bin/bash


while read FILE AFTER BEFORE
do  [ -z "${FILE%\#*}" ] && continue
    [ ! -z "${AFTER}"  ] && AFTER="-a ${AFTER}"
    [ ! -z "${BEFORE}" ] && BEFORE="-b ${BEFORE}"
    /boot/custom/bin/istime ${AFTER} ${BEFORE} && ( date > ${FILE} ; rm ${FILE} )
done <<-EOF
/mnt/cache/.istime 06:00 21:00
/mnt/disk1/.istime 06:00 21:00
EOF

 

 

Link to comment

I wrote a tool to keep my drives spinning regardless of the spindown selection

 

it uses an old program I wrote called istime which checks if the current time is within parameters.

If there is any interest I can provide the binary and shell script.

 

 

 

root@atlas /etc/cron.hourly #ls -l /etc/cron.hourly/keepspin.sh 
-rwxr-xr-x 1 root root 338 Nov 21 13:13 /etc/cron.hourly/keepspin.sh*

root@atlas /etc/cron.hourly #cat /etc/cron.hourly/keepspin.sh 
#!/bin/bash


while read FILE AFTER BEFORE
do  [ -z "${FILE%\#*}" ] && continue
    [ ! -z "${AFTER}"  ] && AFTER="-a ${AFTER}"
    [ ! -z "${BEFORE}" ] && BEFORE="-b ${BEFORE}"
    /boot/custom/bin/istime ${AFTER} ${BEFORE} && ( date > ${FILE} ; rm ${FILE} )
done <<-EOF
/mnt/cache/.istime 06:00 21:00
/mnt/disk1/.istime 06:00 21:00
EOF

 

 

 

Dosen't this defeat one of the huge benifits of Unraid, the power savings by not having to have all of your disks spun up?

Link to comment

I wrote a tool to keep my drives spinning regardless of the spindown selection

 

it uses an old program I wrote called istime which checks if the current time is within parameters.

If there is any interest I can provide the binary and shell script.

 

 

 

root@atlas /etc/cron.hourly #ls -l /etc/cron.hourly/keepspin.sh 
-rwxr-xr-x 1 root root 338 Nov 21 13:13 /etc/cron.hourly/keepspin.sh*

root@atlas /etc/cron.hourly #cat /etc/cron.hourly/keepspin.sh 
#!/bin/bash


while read FILE AFTER BEFORE
do  [ -z "${FILE%\#*}" ] && continue
    [ ! -z "${AFTER}"  ] && AFTER="-a ${AFTER}"
    [ ! -z "${BEFORE}" ] && BEFORE="-b ${BEFORE}"
    /boot/custom/bin/istime ${AFTER} ${BEFORE} && ( date > ${FILE} ; rm ${FILE} )
done <<-EOF
/mnt/cache/.istime 06:00 21:00
/mnt/disk1/.istime 06:00 21:00
EOF

 

 

 

Dosen't this defeat one of the huge benifits of Unraid, the power savings by not having to have all of your disks spun up?

 

 

Look carefully. I only have 2-3 drives spinning all the time between 6:00 and 21:00.

Point is, These are my most active accesses and I do not want to wait.

The benefit of spinning them down slows me down when I need quick access to them.

 

Considering that I used to have all these RAID1 mirrored sets spinning 24x7, I'm actually way ahead of the game now with a 20 drive server where only about 3 drives are spinning all the time now.

 

I've replaced allot of OS drives with SSD and moved the other ancillary files (/home) to an NFS access to this server.

I also use a few vmware machines with the My Documents folder mapped to the unRAID server.

So I need them spinning.

 

It's kinda cool as my slackware dev vmware system automounts my source tree on unRAID.

I build a package, switch terminals and it's already on my unRAID server for installation.

 

One of these days (perhaps black friday) I'll score a cost effective 120GB SSD, move my home and source trees to that as a LOCAL or APPS drive, then let the hybrid cache/parity pair spin down.

 

 

 

Link to comment

You guys realize you can change the spindown timer of each drive individually, right?  Click on the drive's link on the unRAID main page (parity, disk1, disk2, etc.).  Whatever spindown timer you set there will override the global spindown timer.  If you have certain drives you need spun up at all times, just disable the timer there.

 

I can still see the usefulness of a scheduling script like Weebo's, but I personally don't need something like that.  I do put some thought into my data organization - all files that I need constant access to I place in a user share that is confined to the minimum number of disks.

Link to comment

I wrote a tool to keep my drives spinning regardless of the spindown selection

 

it uses an old program I wrote called istime which checks if the current time is within parameters.

If there is any interest I can provide the binary and shell script.

 

 

 

root@atlas /etc/cron.hourly #ls -l /etc/cron.hourly/keepspin.sh 
-rwxr-xr-x 1 root root 338 Nov 21 13:13 /etc/cron.hourly/keepspin.sh*

root@atlas /etc/cron.hourly #cat /etc/cron.hourly/keepspin.sh 
#!/bin/bash


while read FILE AFTER BEFORE
do  [ -z "${FILE%\#*}" ] && continue
    [ ! -z "${AFTER}"  ] && AFTER="-a ${AFTER}"
    [ ! -z "${BEFORE}" ] && BEFORE="-b ${BEFORE}"
    /boot/custom/bin/istime ${AFTER} ${BEFORE} && ( date > ${FILE} ; rm ${FILE} )
done <<-EOF
/mnt/cache/.istime 06:00 21:00
/mnt/disk1/.istime 06:00 21:00
EOF

 

 

 

Dosen't this defeat one of the huge benifits of Unraid, the power savings by not having to have all of your disks spun up?

 

 

Look carefully. I only have 2-3 drives spinning all the time between 6:00 and 21:00.

Point is, These are my most active accesses and I do not want to wait.

The benefit of spinning them down slows me down when I need quick access to them.

 

Considering that I used to have all these RAID1 mirrored sets spinning 24x7, I'm actually way ahead of the game now with a 20 drive server where only about 3 drives are spinning all the time now.

 

I've replaced allot of OS drives with SSD and moved the other ancillary files (/home) to an NFS access to this server.

I also use a few vmware machines with the My Documents folder mapped to the unRAID server.

So I need them spinning.

 

It's kinda cool as my slackware dev vmware system automounts my source tree on unRAID.

I build a package, switch terminals and it's already on my unRAID server for installation.

 

One of these days (perhaps black friday) I'll score a cost effective 120GB SSD, move my home and source trees to that as a LOCAL or APPS drive, then let the hybrid cache/parity pair spin down.

 

 

 

 

Well that sheds a whole lot more light on the matter. I still have all of my files on my local pc (docs ect) and use synctoy to mirror them to Unraid. I guess it could get annoying waiting for drives to spin up while performing otherwise mundane pc tasks.

Link to comment

You guys realize you can change the spindown timer of each drive individually, right?  Click on the drive's link on the unRAID main page (parity, disk1, disk2, etc.).  Whatever spindown timer you set there will override the global spindown timer.  If you have certain drives you need spun up at all times, just disable the timer there.

 

I can still see the usefulness of a scheduling script like Weebo's, but I personally don't need something like that.  I do put some thought into my data organization - all files that I need constant access to I place in a user share that is confined to the minimum number of disks.

 

I'm aware of the setting, but after 9pm, I don't need the drive spinning for longer then an hour.

Also my script was invented before we had individual disk timers.

 

With my setup. even confining grouped files to disks with usershares still causes all the disks to spin up.

 

I think the user share checks the root of the split level across all disks first.

That annoyed the heck out of me so I stopped using user shares.

 

What happens is if I scan down a tree with hundreds of thousands of files, the directory entries eventually get flushed out.

 

I'm testing this out with a kernel tuning to see if I can prevent the spin ups. dcache_entries=(some crazy high value). and it's not working.

 

What I may do is a modified version of the dir_cache program which only does a maxdepth for the deepest splitlevel. That may help prevent spurious spinups.

Then again keep in mind all my source files, directories, objects, etc,etc for multiple machines points to a huge directory tree.

 

 

But to get back to the root question. I do not change my habit of what I'm going to use based on it's spinning state.

What I do, is organize objects so that what I want to access is in the state I want it.

 

Link to comment

I personally do not write to disk shares. I use them for users viewing my files that way I can control exactly where my files go and I know typically which drives are up and which are normally down.

 

In the past I've always kept my TV rips and DVD rips on separate drives, but I'm coming to the conclusion that sure that makes since in organization, but as far as when drives are spinning up and shutting down I could put new content on the same drive and older content on older drives so it limits whats up and what should be up.

 

Of course I haven't moved anything yet because its a pain in the butt, but I have been thinking about it. ;) As I'm adding another 1.5TB drive that I had stashed since 2TB drives are rather pricey right now.

Link to comment

I personally do not write to disk shares. I use them for users viewing my files that way I can control exactly where my files go and I know typically which drives are up and which are normally down.

 

Did you mean user shares? I.E. I do not write to user shares.

 

I only write to disk shares so I know where everything is and what is spinning.

Link to comment

Just wondering is anybody else the same as me when it comes to using their unraid server.

When I want to watch something I will have a quick check of the unraid home page to see if any disks are spinning then I will either watch a movie/tv show on the same disk depending what the outcome is.

Another thing if say my brother wants a file I will put it on the cache disk to prevent me spinning up the parity and another disk in the raid.

Likewise if I only have a tv show or two downloaded I will wait to accumulate some more until i move them from the cache.

The main reason I do this is after reading that the most strain on the harddrives occur when they are being spun up. I have the default time of 1hr to spin down the drive. I would rather leave this as they are generally only used in the evening and wont be spun down until they are finished with.

Am I being ridiculous?  :)

The server is there for me... not the other way around.  There is no way I'm going to look to see what is spinning to only select media from it.

 

If I want to use, I don't care if the disks are spinning... (In fact, in most cases, I want them to be spinning)

 

In my opinion, you are being a bit ridiculous/compulsive.  ???

Link to comment

I personally do not write to disk shares. I use them for users viewing my files that way I can control exactly where my files go and I know typically which drives are up and which are normally down.

 

Did you mean user shares? I.E. I do not write to user shares.

 

I only write to disk shares so I know where everything is and what is spinning.

 

Thanks for the catch. I do not write to User Shares. ;)

Link to comment

My usage pattern is that I pull some torrents every few days which means unraid spins up to seed and once download is complete the file is copied to cache and then once my ratio hits a decent level I shut down.  In general that means all (read 3) of my disks spin up.

 

After that I tend to pull a single stream at a time for viewing, and three times a week a drive spins up to accept a system snapshot from Acronis.

 

So, I personally don't worry too much about spin up because it doesn't happen that often and when it does I don't much care about any delays.  My only acknowledgment of the issue is I use cache_dirs to avoid waiting for spin-up when my BD set top powers up and searches my media directory tree.  I still have to wait for spin up once I select a video to watch, but at least I avoid spinning up all drives at once.

 

Beyond that, I don't much care because my usage pattern doesn't lend itself to a lot of spin up/down cycles.  If it did I'd first solve it with a longer spin down delay (or even Weebotech's script) before I started making my life more difficult.

 

A buddy of mine, new to unraid and this board, has a different usage pattern than mine (more active torrenting) and as such is mucking about with running his uncompressed torrents from the cache drive so that he only ever has one drive spun up.  For stuff he wants to keep and/or which might not stream well from compressed format, he manually moves it to the array.  

 

We've been going back and forth about the relative merits of that plan vs. just forcing all torrents to operate from one array drive (either using include, split level, or drive mounts) until such time as he runs out of room on a 2TB.  I'm more inclined to use that option since that puts the 2TB in the array vice as a cache drive.  But it does mean he is more likely to have two drives spun up at a given time while torrenting vs. only one.  

 

It is all an interesting thought experiment and if a single design choice can minimize spin up great.  However, I won't put in the effort on a recurring basis just to prevent spinup. Like Joe L. the server is there for me, and I'm not keeping 20 drives spinning 24/7 so I'm not going to spend more recurring brain power managing my drive access than it will actually save me.  

Link to comment

My compromise for minimizing power cycles - l keep all drives set to manual and spin everything down at bed time.  Whenever they spinup, they stay that way for the rest of the day.

 

Does anyone have a script that changes the spindown delay depending on the time of day?

Link to comment

My compromise for minimizing power cycles - l keep all drives set to manual and spin everything down at bed time.  Whenever they spinup, they stay that way for the rest of the day.

 

Does anyone have a script that changes the spindown delay depending on the time of day?

 

I have a keep busy script that can be used per time of day, wade through the full thread and see if my tool would help, if so I'll post the istime program (the script is already in the thread).

 

It could just as easily be down with cron.

Link to comment

just out of interest, how long does it take for your disks to spin up?

 

Few seconds I'm guessing. However when I'm watching a movie or a tv show I'm not in a super hurry. LOL

 

If I was to do some video editing I guess I would force the drive to spin up first then do what I need to do to insure there are no problems.

Link to comment

Archived

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

×
×
  • Create New...