unRAID Server Release 5.0-rc6-r8168-test Available

[pantner]

Maybe in 5.1 - unRAID should support mirror features.. like in Raid 10.

 

Why? It also has parity which is much smarter than RAID-1 and more space efficiant (ie, you don't need an equal size for the mirror as for the data. Some people have 18-24-60TB setups!).

I believe there's talk about a 2-drive parity system which will allow for 2 data disk failures.

but im not sure how official that is. That would be preferable to mirror/stripe features IMHO.

 

You could also manually setup a mirror, add that disk to its own user share, exclude other disks from it, and it from other shares. Or just access it directly through the disk# way.

[mvdzwaan]

Maybe in 5.1 - unRAID should support mirror features.. like in Raid 10.

 

Why? It also has parity which is much smarter than RAID-1 and more space efficiant (ie, you don't need an equal size for the mirror as for the data. Some people have 18-24-60TB setups!).

I believe there's talk about a 2-drive parity system which will allow for 2 data disk failures.

but im not sure how official that is. That would be preferable to mirror/stripe features IMHO.

 

You could also manually setup a mirror, add that disk to its own user share, exclude other disks from it, and it from other shares. Or just access it directly through the disk# way.

 

It can be done manually (and I have done it for just my personal pictures collection which I would rather have as copies on two or more disks with the additional parity security, but don't forget you can easily loose data with an unraid setup if 2 disks fail).

 

Only thing in unraid which makes this really not practical is that you get 'duplicate' file messages for all these files. So any support for such a feature would only have to include the option per share to not check for duplicates.. It would of course be great to just have a user share with an extra option 'keep on 1|2|3|4|...... disks'...

 

[eek]

 

It can be done manually (and I have done it for just my personal pictures collection which I would rather have as copies on two or more disks with the additional parity security, but don't forget you can easily loose data with an unraid setup if 2 disks fail).

 

 

 

I just adopt the old Windows Home Server approach of keeping things I really want in a different directory on a separate hard disk. In my case disks 1-4 are my main array disks 5 consists of a backup directory with copies of all my photos documents and music. Rsync runs in the background hourly to copy any changes in those user folders to /mnt/user/backup/Music and /mnt/user/backup/Photos.

 

 

[Joe L.]

You can 'hot-swap' in that if you 'Stop' the array, you can perform any disk movements and then 'Start' the array.  No need to reboot - unRAID will recognise new devices on starting the array.

Is that true? Can you just pull out drives if the array is stopped, and the os etc is still running? No need to powerdown first? I thought you should at least enable AHCI in the BIOS as well?

Yes, true, on the latest few versions of the 5.0beta series.  You MUST have AHCI enabled in the BIOS for it to notice the drive removal.  (The removal of a disk in that fashion is equivalent to that disk failing.  If more than one in the array is removed, you 'll not be able to re-start the array.)

[PeterB]

Only thing in unraid which makes this really not practical is that you get 'duplicate' file messages for all these files.

 

I thought that only happened if the duplicate files are all in the same user share?

 

If you want to keep security copies like this, surely it would make sense for the copies to be on a totally different disk share (a drive not involved in the user share) with, perhaps, a different directory tree (to help eliminate confusion when dealing with the copies).

[mr-hexen]

 

 

I just adopt the old Windows Home Server approach of keeping things I really want in a different directory on a separate hard disk. In my case disks 1-4 are my main array disks 5 consists of a backup directory with copies of all my photos documents and music. Rsync runs in the background hourly to copy any changes in those user folders to /mnt/user/backup/Music and /mnt/user/backup/Photos.

 

i'm interested in seeing a more detailed how-to on your script etc can you start a new thread in User Customizations ??

 

thanks.

[jowi]

Yes, true, on the latest few versions of the 5.0beta series.  You MUST have AHCI enabled in the BIOS for it to notice the drive removal.  (The removal of a disk in that fashion is equivalent to that disk failing.  If more than one in the array is removed, you 'll not be able to re-start the array.)

So it is also possible to dynamically extend the array by stopping the array, and insert a new disk e.g. using an empty hotswap bay, and restart the array? Without rebooting? (Offcourse the disk will have to be precleared first before it can be assigned etc)

[PeterB]

So it is also possible to dynamically extend the array by stopping the array, and insert a new disk e.g. using an empty hotswap bay, and restart the array? Without rebooting?

 

With the later 5.0 betas, yes!  Obviously you need to assign the disk before restarting.

[Frank1940]

So it is also possible to dynamically extend the array by stopping the array, and insert a new disk e.g. using an empty hotswap bay, and restart the array? Without rebooting?

 

With the later 5.0 betas, yes!  Obviously you need to assign the disk before restarting.

 

Has anyone actually had the guts to try all these forms of 'hot swapping' and can provide testimony about their experience and results?  It is one thing for something to theoretically possible and quite another for it to work in actual practice.  I haven't seen any statements from LimeTech advertising this functionality.

 

(Not picking particularly on you, PeterB, as a number of other persons have made comments regarding the ease of 'hot swapping'.)

[PeterB]

Has anyone actually had the guts to try all these forms of 'hot swapping' and can provide testimony about their experience and results?

 

I'm not quite sure what you mean/imply by the words I've highlighted, but regarding hot-swapping:

 

Tyrindor says that he does it 'all the time'.

 

I've certainly done it on a number of occasions.  I've stopped the array, moved all my data drives from one controller to another, and restarted the array - it works perfectly.

 

Joe L has stated that it is possible in the latest 5.0 betas.

 

From Tom's release notes, from 5.0-beta4 to 5.0-beta5:

- webGui: support hot plug of hard drives when array is Stopped (limited h/w testing).

 

[Frank1940]

Has anyone actually had the guts to try all these forms of 'hot swapping' and can provide testimony about their experience and results?

 

I'm not quite sure what you mean/imply by the words I've highlighted, but regarding hot-swapping:

 

Tyrindor says that he does it 'all the time'.

 

I've certainly done it on a number of occasions.  I've stopped the array, moved all my data drives from one controller to another, and restarted the array - it works perfectly.

 

Joe L has stated that it is possible in the latest 5.0 betas.

 

From Tom's release notes, from 5.0-beta4 to 5.0-beta5:

- webGui: support hot plug of hard drives when array is Stopped (limited h/w testing).

 

Thank you, PeterB.  First, I must confess that I did not completely read all of the release notes.  ( I started testing unRIAD about beta 10 or 11.)  I completely missed the statement that Tom had acknowledged that his software was compatible with 'hot swapping'. 

 

But even more reassuring is the statement that you personally have actually done so on several occasions and not had a problem.  When I read the comments in this thread, they all seemed to be in a form which lead me to think that people were making statements that since hot swapping now in the Linux kernel and was now a feature on many new (if not all) motherboards that it would 'just' work with unRAID. 

 

I now stand corrected and I really do appreciate hearing of your successful experiences.   

[abs0lut.zer0]

Only thing in unraid which makes this really not practical is that you get 'duplicate' file messages for all these files.

 

I thought that only happened if the duplicate files are all in the same user share?

 

If you want to keep security copies like this, surely it would make sense for the copies to be on a totally different disk share (a drive not involved in the user share) with, perhaps, a different directory tree (to help eliminate confusion when dealing with the copies).

 

i too am still getting this under the latest test ....

i thought the duplicate issue was supposed to fixed in the beta's ?

 

[PeterB]

i too am still getting this under the latest test ....

i thought the duplicate issue was supposed to fixed in the beta's ?

 

The circumstance which I encountered has been fixed - that of renaming a file on one disk in the user share to be the same as the name of a file on another disk in the user share.

 

The only way it should be possible to get duplicates is to place a file in a user share and then place another copy of the file in the user share directory, but accessed as a disk share.  There is little that can be done to prevent this.  Any other mechanism could be considered as a bug.

 

Are you able to reproduce the duplicate 'issue' at will - do you know the circumstances which cause it?  If so, Tom will tell you how to enable debugging so that he can trace the problem within the unRAID code.

[abs0lut.zer0]

i too am still getting this under the latest test ....

i thought the duplicate issue was supposed to fixed in the beta's ?

 

The circumstance which I encountered has been fixed - that of renaming a file on one disk in the user share to be the same as the name of a file on another disk in the user share.

 

The only way it should be possible to get duplicates is to place a file in a user share and then place another copy of the file in the user share directory, but accessed as a disk share.  There is little that can be done to prevent this.  Any other mechanism could be considered as a bug.

 

Are you able to reproduce the duplicate 'issue' at will - do you know the circumstances which cause it?  If so, Tom will tell you how to enable debugging so that he can trace the problem within the unRAID code.

 

well i have never done that as i only user total commander in windows or /mnt/user with mc  i never touch the separate disks... ?

[dimaestro]

So it is also possible to dynamically extend the array by stopping the array, and insert a new disk e.g. using an empty hotswap bay, and restart the array? Without rebooting?

 

With the later 5.0 betas, yes!  Obviously you need to assign the disk before restarting.

 

Has anyone actually had the guts to try all these forms of 'hot swapping' and can provide testimony about their experience and results?  It is one thing for something to theoretically possible and quite another for it to work in actual practice.  I haven't seen any statements from LimeTech advertising this functionality.

 

(Not picking particularly on you, PeterB, as a number of other persons have made comments regarding the ease of 'hot swapping'.)

 

---- current uptime -----

(from /usr/bin/uptime)

 

12:13:33 up 137 days, 17:30,  0 users,  load average: 0.14, 0.05, 0.06

 

 

In that time I've gone from a 1.5tb parity disk to a 2tb parity disk to a 3tb parity disk, put the old 1.5tb parity as a drive, then the 2tb parity as a drive. I've gone from no-cache to a cache, back to no-cache, back to having a cache.

 

The last time I rebooted was to put RC14 on. Previous uptime was near 400 days IIRC.

 

MY server is in an Antec 1080 case w/out hotswap bays. I just turn off the array, slap in a drive, screw it down, head back upstairs (server is headless) initiate a pre-clear, bring the array back up. When preclear is finished, I bring down array, assign a drive to it. (Of course running parity checks or rebuilds as needed)

 

**Correction - last power down was due to a 2 hour power outage, my UPS is only good for about 30 minutes. That was this spring during a good storm.

[optiman]

Ok, I've been voting to get this FINAL for a while now because I've had no issues.  That changed today.  My shares dissapeared so I went to my unraid server via web gui and I could see my shares, but my popcorn hour couldn't find the share and my pc also said the share did not exist.  So I decided to pick the movie share and just hit Apply under the section where you select who has access.  This used to be a good trip to get the share stuff going again in the older versions.  This time I got an error saying the share had been deleted!  WTF!  So went back to the share list and they were all gone.  I instantly just stopped the array and did a reboot.  After the reboot all shares were back and everything seems normal.  I have no idea why the server would just stop sharing the shares and why the web gui was reporting no shares exists.  My server had been running a couple days and I've not had this issue since before 4.7 Final and certinaly not with v5.

 

The good thing is my shares were really not deleted and the reboot fixed it.  the question is WHY....

 

syslog attached

 

I hope it helps idenfity what happened.

 

Thanks!

syslog-20120814-193624.zip

[mvdzwaan]

Ok, I've been voting to get this FINAL for a while now because I've had no issues.  That changed today.  My shares dissapeared so I went to my unraid server via web gui and I could see my shares, but my popcorn hour couldn't find the share and my pc also said the share did not exist.  So I decided to pick the movie share and just hit Apply under the section where you select who has access.  This used to be a good trip to get the share stuff going again in the older versions.  This time I got an error saying the share had been deleted!  WTF!  So went back to the share list and they were all gone.  I instantly just stopped the array and did a reboot.  After the reboot all shares were back and everything seems normal.  I have no idea why the server would just stop sharing the shares and why the web gui was reporting no shares exists.  My server had been running a couple days and I've not had this issue since before 4.7 Final and certinaly not with v5.

 

The good thing is my shares were really not deleted and the reboot fixed it.  the question is WHY....

 

syslog attached

 

I hope it helps idenfity what happened.

 

Thanks!

 

Are you running any plugins/add-ons (simplefeatures for instance) ?

[optiman]

Yes, I've installed most of them, except the iTunes ones.  I'm also runing Plex and APC, and that's it.  All of these have been in place since I went to v5.  I've actually had no issues with any of them or my server for that matter.  This is the real first issue I've had. 

 

I hope the log shows something about why the shares would demon would just stop and why the server would think it has no shares anymore, just weird.

[mvdzwaan]

You did notice the hundreds of

Aug 14 14:48:20 Tower kernel: scsi_verify_blk_ioctl: 32 callbacks suppressed
Aug 14 14:48:20 Tower kernel: hdparm: sending ioctl 2285 to a partition!
Aug 14 14:48:27 Tower last message repeated 5 times

 

entries in your log ?

 

Also your log is not complete, it creates a new file once the syslog reaches a certain size.

 

Looking through your syslog I find rows containing

 

Aug 14 19:35:10 Tower emhttp: get_filesystem_status: statfs: /mnt/user/test Transport endpoint is not connected

 

after which emhttp decides to stop SMB, but this might be the point where you stopped it...

 

The ioctl errors were discussed in http://lime-technology.com/forum/index.php?topic=21201.0 and were plugin/simplefeatures related.

 

I would suggest you get that problem sorted first...

[lovingHDTV]

Imagine that, I just read this a few hours ago and just hit the exact same thing on my system.  The syslog says:

Aug 15 13:51:14 Tower unmenu[1759]: df: `/mnt/user': Transport endpoint is not connected

 

I just enabled my IBM M1015 with a single drive on it, forced all the disks to sleep.  I can access the disk directly via \\tower\disk4, but all shares are gone, even from the web page.

 

I don't use simplefeatures.

 

Stopping, restarting array, or SMB does not get shares back.  A reboot does.

 

dave

 

Ok, I've been voting to get this FINAL for a while now because I've had no issues.  That changed today.  My shares dissapeared so I went to my unraid server via web gui and I could see my shares, but my popcorn hour couldn't find the share and my pc also said the share did not exist.  So I decided to pick the movie share and just hit Apply under the section where you select who has access.  This used to be a good trip to get the share stuff going again in the older versions.  This time I got an error saying the share had been deleted!  WTF!  So went back to the share list and they were all gone.  I instantly just stopped the array and did a reboot.  After the reboot all shares were back and everything seems normal.  I have no idea why the server would just stop sharing the shares and why the web gui was reporting no shares exists.  My server had been running a couple days and I've not had this issue since before 4.7 Final and certinaly not with v5.

 

The good thing is my shares were really not deleted and the reboot fixed it.  the question is WHY....

 

syslog attached

 

I hope it helps idenfity what happened.

 

Thanks!

[optiman]

You did notice the hundreds of

Aug 14 14:48:20 Tower kernel: scsi_verify_blk_ioctl: 32 callbacks suppressed
Aug 14 14:48:20 Tower kernel: hdparm: sending ioctl 2285 to a partition!
Aug 14 14:48:27 Tower last message repeated 5 times

 

entries in your log ?

 

Ok what do they mean??

 

 

Also your log is not complete, it creates a new file once the syslog reaches a certain size.

 

Looking through your syslog I find rows containing

 

Aug 14 19:35:10 Tower emhttp: get_filesystem_status: statfs: /mnt/user/test Transport endpoint is not connected

 

 

Yes, this is when I stopped and rebooted.

 

 

after which emhttp decides to stop SMB, but this might be the point where you stopped it...

 

The ioctl errors were discussed in http://lime-technology.com/forum/index.php?topic=21201.0 and were plugin/simplefeatures related.

 

I would suggest you get that problem sorted first...

 

Ok, I will look into that.  thanks.

 

 

[Ice_Black]

It been more than a month and no rc7 or final.

[jaybee]

It been more than a month and no rc7 or final.

[pantner]

It been more than a month and no rc7 or final.

 

i expect Tom is doing more research into the linux kernal issues plauging LSI users. While there is a version with a fix on it, who knows what other effects that fix will have or if it has been implimented in the correct way.

 

There has been some major progress recently and when it is ready, Tom will release RC7 or the final.

[optiman]

You did notice the hundreds of

Aug 14 14:48:20 Tower kernel: scsi_verify_blk_ioctl: 32 callbacks suppressed
Aug 14 14:48:20 Tower kernel: hdparm: sending ioctl 2285 to a partition!
Aug 14 14:48:27 Tower last message repeated 5 times

 

entries in your log ?

 

Ok what do they mean??

 

Can anyone tell me what these entries mean and what I should do about it?