unRAID Server Release 6.0-beta5a-x86_64 Available


Recommended Posts

From my understanding, a split level of 0 means to keep all data for a directory on the same disk, so if that disk is full you will get a out of space message and it wont use other disks.  I use a split level of 999 so it can spread all data across any of the disks.

Link to comment
  • Replies 148
  • Created
  • Last Reply

Top Posters In This Topic

That is your problem.  Split level 0 is a "special" option.  Unraid will not create the folder structure for you, you must do it yourself.  So if you did not create a folder "my stuff" on both of your drives, then unraid will not do it for you and will not start copying data to the "other" drive despite your allocation method

 

http://lime-technology.com/wiki/index.php?title=Un-Official_UnRAID_Manual#Split_Level_0

 

You will either need to create the folder you want on the other drive, or change to whatever number split you need.

Link to comment

I see jphipps sort of beat me to it.  Just keep in mind that with a split level like 999, or anything relatively high, you run the risk of contantly spinning up drives (and thus waiting) as you travers directories.  Not a huge deal, but also not great planning.  For example if you're listening to an artists collection, why not set it up so at least entire albums stay on the same disc, if not even the entire artist.  Otherwise you have to switch discs literally between songs.  Same for tv shows ... keep a season on a single disc but allow multiple seasons to span discs.  Or photos, if you organize them by event, or subject, etc ... you get the point.

 

Which is, there is a logic to setting split level and the wiki has some great examples to give ideas.

Link to comment

Not sure what you mean about needing to create the directory manually.  I have never had to do that, it has always done it automatically..

 

True, there are a lot of pros and cons about splitting.  I have mine with split of 999 and most free allocation.  I wasn't as concerned about the disks spinning up, my thought was more spindles for the data.  Not always good on writes since there is only 1 parity disk, but for reads, if I have multiple reads going on, it is a high likelihood that the files are on separate disks, so I get more data throughput.  It is also bad because if you did have a catastrophic failure and did happen to loose a disk ( Not too likely with unraid.. ) you would loose some of the data in multiple directories, as a posed to complete directories.  But I figure better to loose a little then all of a directory...

 

 

Link to comment

Like I said, Split 0 is a special case.  read that link, I'm not making it up :)

 

Regardless, you've solved you concern with split 999, I just want you to consider if that is really what you want, especially as you grow. 

 

As for spindle reads and speed ... that will only matter if you have a lot of concurrent activity going on.  If you do, gotcha.  If you don't, you'll never notice the speed difference since your network will quickly become the bottle neck anyway.

 

I also agree on the failure issue ... that is why I too use most-free and let all my data split across all drives.  But I do it such that common sequential access situations will not result in spinning up more drives than needed.  Like when binge watching a TV show season.

Link to comment

From past experience, some things need to be in one directory to work well (dvd rips come to mind, had no amount of trouble with them split across several drives back when I just ripped the VIDEO_TS folders for my old Sage system to read).  I think I have mine set somewhere around 3 or 4, and usually don't have any issues.  Didn't realize the special case with the '0' split level, may set some of my data shares like that, so I can manage where they go..

 

Link to comment

There is, and I've updated it ... mentioned in my first post in an edit along with other updates.

There are some IOMMU-related kernel settings which have been turned on for -beta6.  May or may not help.

Tom, can you also activate VFIO modules on next release?

 

//Peter

 

Link to comment

There is, and I've updated it ... mentioned in my first post in an edit along with other updates.

There are some IOMMU-related kernel settings which have been turned on for -beta6.  May or may not help.

Tom, can you also activate VFIO modules on next release?

 

//Peter

 

For upcoming -beta6a using 3.14.4 kernel:

 

CONFIG_VFIO_IOMMU_TYPE1=m

CONFIG_VFIO=m

CONFIG_VFIO_PCI=m

CONFIG_VFIO_PCI_VGA=y

CONFIG_KVM_VFIO=y

 

That about it?

Link to comment

There is, and I've updated it ... mentioned in my first post in an edit along with other updates.

There are some IOMMU-related kernel settings which have been turned on for -beta6.  May or may not help.

Tom, can you also activate VFIO modules on next release?

 

//Peter

 

For upcoming -beta6a using 3.14.4 kernel:

 

CONFIG_VFIO_IOMMU_TYPE1=m

CONFIG_VFIO=m

CONFIG_VFIO_PCI=m

CONFIG_VFIO_PCI_VGA=y

CONFIG_KVM_VFIO=y

 

That about it?

Great, that is was I have enabled, thanks Tom, I also enabled all IOMMU modules, but I see you already done that;-)

 

//Peter

 

Link to comment

My cache mover refuses to move the files over. I start it manually, it'll move 2-3 files, then stop. Not sure if it's caused by accessing the cache or writing new files to the cache but this didn't happen on 5.0.

 

Is it possible that one of your array disks is almost full? This has been known to prevent mover operations for others.

 

Link to comment

My cache mover refuses to move the files over. I start it manually, it'll move 2-3 files, then stop. Not sure if it's caused by accessing the cache or writing new files to the cache but this didn't happen on 5.0.

 

Is it possible that one of your array disks is almost full? This has been known to prevent mover operations for others.

 

They all have 800GB+ free. If I click the script, it'll move over 2-3 more files, then stop again. I am going to click it this time and not access the server in anyway. It almost seems like the mover is stopping due to user read/write activity.

 

EDIT: Fixed it. Somehow my share got set to cache "only". Odd... I haven't been in that share settings for months.

Link to comment

Did php-cgi build get dropped?

Yes but php-fpm is enabled.

 

Ahhh.  It broke my webserver config which was looking for php-cgi, so family could not get to my media system... nothing but 500 errors.

 

Would have saved me a bit of good bit of grief had that been in the release notes. ;)

Link to comment

Last week, I disabled network bonding, to eliminate a source of possible problems while working on other issues.  I now only have one network cable connected to my machine, but I have 2 NIC on the motherboard, and a 3rd on an Intel NIC installed in a PCI slot.

 

Last week, I rebooted the machine, using the GUI, and when the machine finished booting, I could not access it from my network.  I tried several things, but it was just not accessible.  I finally thought to move the ethernet cable from one motherboard slot to the other, and all was well.  I had full access and was happy.

 

Today, i rebooted again from the GUI and once again I can't access the server.  I just switched the cable back to the other slot in the motherboard, and once again have access.

 

Seems like a bug to me :(

Link to comment

My father-in-law is running 5.0.5 and it does the same thing.  He has an ASUS motherboard with 2 onboard NICs, and most times when he reboots the server, it switches to the other one, and he has to switch the cable over to the other NIC to get the network back...

Link to comment

Last week, I disabled network bonding, to eliminate a source of possible problems while working on other issues.  I now only have one network cable connected to my machine, but I have 2 NIC on the motherboard, and a 3rd on an Intel NIC installed in a PCI slot.

 

Last week, I rebooted the machine, using the GUI, and when the machine finished booting, I could not access it from my network.  I tried several things, but it was just not accessible.  I finally thought to move the ethernet cable from one motherboard slot to the other, and all was well.  I had full access and was happy.

 

Today, i rebooted again from the GUI and once again I can't access the server.  I just switched the cable back to the other slot in the motherboard, and once again have access.

 

Seems like a bug to me :(

I wonder if linux isn't doing the same thing with nic's that it does with drives and enumerates whichever nic initializes first as eth0.  You might try hiding all but the one you want with pciBack.hide and see if it works more consistently that way.  Assuming what I guessed is even remotely accurate of course.
Link to comment

Last week, I disabled network bonding, to eliminate a source of possible problems while working on other issues.  I now only have one network cable connected to my machine, but I have 2 NIC on the motherboard, and a 3rd on an Intel NIC installed in a PCI slot.

 

Last week, I rebooted the machine, using the GUI, and when the machine finished booting, I could not access it from my network.  I tried several things, but it was just not accessible.  I finally thought to move the ethernet cable from one motherboard slot to the other, and all was well.  I had full access and was happy.

 

Today, i rebooted again from the GUI and once again I can't access the server.  I just switched the cable back to the other slot in the motherboard, and once again have access.

 

Seems like a bug to me :(

 

When there are multiple enabled NIC's there are race conditions in linux affecting how they get enumerated.  The network setup scripts always uses 'eth0'.  There are few ways to solve this but easiest is just turn on bonding.  You don't have to have cables connected to all the ports, in fact you only need one cable, but it won't matter which port you plug it into.

Link to comment

Did php-cgi build get dropped?

Yes but php-fpm is enabled.

 

Ahhh.  It broke my webserver config which was looking for php-cgi, so family could not get to my media system... nothing but 500 errors.

 

Would have saved me a bit of good bit of grief had that been in the release notes. ;)

 

In the words of Robin Seggelmann, "oops".

Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.