unRAID Server Release 5.0-rc11 Available

[PeterB]

Of course. I wasn't  questioning anything. I'd noticed the other posts, possibly mirroring my "wouldn't that be nice" optimism that the rebuild would somehow be smart about the last TB and only consider it once a data drive intruded into those sectors.

 

So that would mean that at the point in time when a larger drive is added, not only does the system have to ensure that the new drive is filled with zeroes, but it would also have to ensure that the extended part of the parity drive is all zeroed.

[optiman]

Well I can confirm that a rebuilding a drive from parity works.  I lost a drive so I power everything down, slammed a brand new drive in it's place and fired it back up.  I didn't have time to mess with a preclear run, as I was headed out for the day and wanted to start the rebuild process before I left (I know this isn't recommended but oh well), so I just assigned the new drive to that slot, and started the rebuild.  It took a while, 12 hours or so, but completed with no issues.  All my data was there.  I then ran a parity check, which went fine and with zero errors.  Everything is back to normal.  Not that I was ever really worried about this stuff, but it's nice to know that it works just fine.

[madburg]

You could have exported a Smart report of your new drive before selecting it to be rebuilt, then after rebuild completed, run another smart report (based on how much data was written to the drive say 50% of it capacity) you have some idea of the condition of this new drive. Better than nothing, right.

[Joe L.]

You could have exported a Smart report of your new drive before selecting it to be rebuilt, then after rebuild completed, run another smart report (based on how much data was written to the drive say 50% of it capacity) you have some idea of the condition of this new drive. Better than nothing, right.

You can still do the smart report after... 

 

If you see ANY sectors pending re-allocation, let unRAID rebuild the drive again. (un-assign it, start array with it un-assigned, then stop array once more and re-assign it.)  Otherwise, the reconstruction has written a sector that cannot be read... and the next time you perform a correcting parity sync, if that sector was in use, you will have lost data.

 

[optiman]

You could have exported a Smart report of your new drive before selecting it to be rebuilt, then after rebuild completed, run another smart report (based on how much data was written to the drive say 50% of it capacity) you have some idea of the condition of this new drive. Better than nothing, right.

You can still do the smart report after... 

 

If you see ANY sectors pending re-allocation, let unRAID rebuild the drive again. (un-assign it, start array with it un-assigned, then stop array once more and re-assign it.)  Otherwise, the reconstruction has written a sector that cannot be read... and the next time you perform a correcting parity sync, if that sector was in use, you will have lost data.

 

 

Ok, ran the Short self test under the Health tab, no errors.  Will the short self test show the pening re-allocaiton errors, or do I have to run the extended test?

[barrygordon]

I am attempting to convert from 5.0RC8 to 5.0 RC11 and am having the following issues.

 

I did as the release not instructions stated

 

When I booted the array came up as started, yet the release notes state it should have come up stopped.

 

I do not see the Tower in my PC's network neighborhood although I see all other machines on the LAN I expect to see. The SMB configuration looks correct when I look in settings. Any Thoughts?

 

I checked the drive contents from a PC by entering \\Tower which brought up the share list of directories which all look okay. Contents of the drives seems to also be correct and I can play a movie file from the Tower's movie directory.

 

Do I need to run user permissions as I am migrating from 5.0 RC8 ?

 

I have not yet loaded any plugins, but plan to install Simple features

[madburg]

Yes!

 

Download | Release Notes

 

There are a couple critical bug fixes in here and everyone is encouraged to upgrade to this release.

 

Please note: if you are upgrading from 4.7 or lower, please run the "New Permissions" utility available on the Utils page.  This is necessary in order for the 5.0 security model to work correctly.  If necessary, you will be able to downgrade back to 4.7 with no ill effects from this.  If you are upgrading from an earlier 5.0-beta or 5.0-rc release, you will also need to run the "New Permissions" utility again unless you upgraded to 5.0-rc9 or -9a and did so yesterday.

[Frank1940]

I am attempting to convert from 5.0RC8 to 5.0 RC11 and am having the following issues.

 

I did as the release not instructions stated

 

When I booted the array came up as started, yet the release notes state it should have come up stopped.

 

I do not see the Tower in my PC's network neighborhood although I see all other machines on the LAN I expect to see. The SMB configuration looks correct when I look in settings. Any Thoughts?

 

I checked the drive contents from a PC by entering \\Tower which brought up the share list of directories which all look okay. Contents of the drives seems to also be correct and I can play a movie file from the Tower's movie directory.

 

If you had previously set to Server to autostart the array in the recent rc releases, the setting will cause the array to start itself. 

 

It sounds like your server has not yet managed to get the attention of the 'Master Browser' on the SMB network.  Time may resolved the issue.  (May not either!!!) 

 

In any case, go to settings page, click on 'SMB' and set 'Local Master:'  to "Yes".  This will allow your server to become the Master Browser and this will fix the problem.  (You may have to shutdown all devices (both computers and media servers) on the network that use SMB to force this to happen.)

[madburg]

I wouldn't do that, I would not rely on this linux SMB implementation to be the master browser for all your MS clients.

Up to you at the end.

 

[barrygordon]

madburg,

 

Thanks, I have started user permissions and it has "stalled" on disk 4. Being the impatient SOB that I am I started it again and it stalled again.  I then remembered that disk 4 is my backup disk and has thousands of files where all others have less than 100 files.

 

I assume that the fact that I started it again will not have a negative effect other than maybe making the process take twice as long as it should if both are running

 

Frank,

 

My SMB configuration has local master set to yes. I think the issue is the Dune players on the network.  I believe they are set to be the Browse Master in their configurations (OS level?,  Preferred Master?) . I shut down the Dune player that was the Browse Master according to nbtstat and the other Dune player became the Browse Master and the Tower showed up in the Network neighborhood as the shutdown of the Browse Master caused a new election. Once again my impatience is causing me grief.

 

I have a mod to the GO script that I will install to see if that will force the Tower to become the Browse Master. It alters the OS level to 255 and Preferred master to Yes. 

[Frank1940]

madburg,

 

Thanks, I have started user permissions and it has "stalled" on disk 4. Being the impatient SOB that I am I started it again and it stalled again.  I then remembered that disk 4 is my backup disk and has thousands of files where all others have less than 100 files.

 

I assume that the fact that I started it again will not have a negative effect other than maybe making the process take twice as long as it should if both are running

 

Frank,

 

My SMB configuration has local master set to yes. I think the issue is the Dune players on the network.  I believe they are set to be the Browse Master in their configurations (OS level?,  Preferred Master?) . I shut down the Dune player that was the Browse Master according to nbtstat and the other Dune player became the Browse Master and the Tower showed up in the Network neighborhood as the shutdown of the Browse Master caused a new election. Once again my impatience is causing me grief.

 

I have a mod to the GO script that I will install to see if that will force the Tower to become the Browse Master. It alters the OS level to 255 and Preferred master to Yes.

 

That is why I said to shutdown all of your computers and media devices that use SMB.  (Perhaps, I should have said ALL AT ONCE.        )    With many of us having multiple devices and computers all using SMB on our networks, is that any of these media devices can end up being the Master Browser and you have a real turkeyshoot as to which version of Linux SMB is really being run.  Since most of us also have our UNRAID servers on continuously, having it become the Master Browser usually leads to the fewest problems. 

[barrygordon]

Frank, You are quite correct. It is a turkey shoot and my gun is a blunderbuss. Thanks for all the help, advice. Right now all is fine and after I mod the GO script I will see what happens.

[madburg]

How do you know new permissions stalled?

I used to have issue but as of RC10, Tom changed a few thing and even my largest disks complete now, just have to wait awhile (I mean wait!)

I choose to not do anything while it ran.

[optiman]

You could have exported a Smart report of your new drive before selecting it to be rebuilt, then after rebuild completed, run another smart report (based on how much data was written to the drive say 50% of it capacity) you have some idea of the condition of this new drive. Better than nothing, right.

You can still do the smart report after... 

 

If you see ANY sectors pending re-allocation, let unRAID rebuild the drive again. (un-assign it, start array with it un-assigned, then stop array once more and re-assign it.)  Otherwise, the reconstruction has written a sector that cannot be read... and the next time you perform a correcting parity sync, if that sector was in use, you will have lost data.

 

 

Ok, ran the Short self test under the Health tab, no errors.  Will the short self test show the pening re-allocaiton errors, or do I have to run the extended test?

 

I'm running the Extended self test, taking forever.  After 1 hour, I'm at 20%.  This is a 3tb drive and about 35% full.  I hope there is no issue running this Smart extended test on a drive with my data on it.  I also would still like to know if the re-allocation errors would show up on the short test vs. the extended test.  I ran a Parity check with the write correction box unchecked after the new drive came online and had zero errors.

[barrygordon]

MAdburg,

 

Stalled was a bad choice of words. What I perceived as stalled was the browser interface display. It moved nicely from disk 1 to disk 2 to disk 3 to disk 4 and just remained there. I suspect I may have multiple copies of user permissions running as I restarted at least one, and doing a browser refresh probably generated another copy as it would have sent the start request again. There are 8 disks with each disk except disk 4 having less than 500 files.  Disk 4 has thousands!

 

I assume I can just stop the array, shutdown and when i come back up restart user permissions and be patient!

 

Thoughts?

[madburg]

I believe I saw there was logic in the code so it does not run again if found to be running, but cant verify that at the moment.

 

Yes you could reboot and start from scratch. You want it to complete once successfully.

 

It theory, as I understand it is as long as you or some plugin don't mess (set) permission outside the defaults you should not need to ever run it again. And if you discover that say a plugin misbehaves and was corrected by a dev, you would want to rerun to set things straight.

 

[barrygordon]

Madburg,  Thanks.  I think that closes this out.

[ufopinball]

Stalled was a bad choice of words. What I perceived as stalled was the browser interface display. It moved nicely from disk 1 to disk 2 to disk 3 to disk 4 and just remained there. I suspect I may have multiple copies of user permissions running as I restarted at least one, and doing a browser refresh probably generated another copy as it would have sent the start request again. There are 8 disks with each disk except disk 4 having less than 500 files.  Disk 4 has thousands!

 

I had similar problems as some of my disks store a decade's worth of pictures, as opposed to hundreds of blu-ray rips.

 

I was told you can run the permissions script manually from a telnet/console session, and that you can specify a particular disk or even a directory-at-a-time:

 

newperms /mnt/disk1

 

newperms /mnt/disk1/Documents

 

For (possibly) more info, see the tail end of this thread:

 

http://lime-technology.com/forum/index.php?topic=22459.0

[barrygordon]

I guess it is not closed out. From what I can see it seems to be the same issue that was happening in the referenced thread. What is the consequence of newperms not running against all of the disks?

 

From what I can tell all is okay on my system, but perhaps not. Do the more knowledgeable suggest I run newperms against all of my disks individually and do the one it seems to be failing on last?

 

That should not be an issue to do since my tower has a keyboard and monitor. a ps command will tell me what is running. 

 

When I ran it using 5.0 RC8a I do not recall a problem, but perhaps disk 4 had significantly fewer files.

 

Any advice, information, suggestions appreciated. When I ran it using 5.0 RC8a I do not recall a problem, but perhaps disk 4 had significantly fewer files.

[madburg]

Problem at the moment is you stated stalled and now stating same problem. I personally am not understanding exactly what it is that is happening. Is there an error in the syslog? Or is it skipping the drive? What is occurring for you exactly is an unknown. Not saying ur not having an issue just what exactly is happening, cause at the moment it seems to be a lack of patience for it to complete.

 

You can run one disk at a time if you wish, and if u go that route, if you start with the problem drive or leave it for last is really up to you. It doesn't work better one way or the other.

[barrygordon]

my plan is to run it in the standard way (start it from the web interface) to night while I am asleep and let it run all night.

 

I will post tomorrow.  Thanks

[optiman]

You could have exported a Smart report of your new drive before selecting it to be rebuilt, then after rebuild completed, run another smart report (based on how much data was written to the drive say 50% of it capacity) you have some idea of the condition of this new drive. Better than nothing, right.

You can still do the smart report after... 

 

If you see ANY sectors pending re-allocation, let unRAID rebuild the drive again. (un-assign it, start array with it un-assigned, then stop array once more and re-assign it.)  Otherwise, the reconstruction has written a sector that cannot be read... and the next time you perform a correcting parity sync, if that sector was in use, you will have lost data.

 

 

Ok, ran the Short self test under the Health tab, no errors.  Will the short self test show the pening re-allocaiton errors, or do I have to run the extended test?

 

I'm running the Extended self test, taking forever.  After 1 hour, I'm at 20%.  This is a 3tb drive and about 35% full.  I hope there is no issue running this Smart extended test on a drive with my data on it.  I also would still like to know if the re-allocation errors would show up on the short test vs. the extended test.  I ran a Parity check with the write correction box unchecked after the new drive came online and had zero errors.

 

self extended test completed after about 20 hours, but found no errors.  So the rebuild worked just fine.

[barrygordon]

I let new perms run overnight with me asleep. It did all 8drives with no problem. It took one and a half hours to complete, so my impatience was the only issue

[madburg]

I let new perms run overnight with me asleep. It did all 8drives with no problem. It took one and a half hours to complete, so my impatience was the only issue

tada! 

[htpcnewbie]

Hi - I upgraded my tower from 5rc8a to rc11 today. Couple of days ago I started having trouble with emhttp doing segmentation fault and search showed few other users have seen that issue. I was hoping this release will fix it but I have been too optimistic. Here is the problem on the new rc11 server:

 

[*]After moving /boot/config/plugins to plugins.bak, my remote computer can load //tower main page. All the MBR settings look okay and the shares look fine, accessible through a remote Windows computer.

[*]I moved my plugin folders back and restarted the machine in the hopes of starting an overnight permissions settings. Now my emhttp seems to have crashed although I do not see any indication of that in /var/log/syslog.

[*]I can bring up my "simple  features" (//tower:8080) web page but nothing else works - No //tower or shares visible in Windows.

I can login through putty to the server.

[*]My hunch is something in the plugins is causing the emttp to crash. The server's uptime was 37 days before I had to restart it. Short of disabling each plugin one by one and restarting the server, is there a way to determine what is causing this behavior? I have copied some relevant output:

 

root@Tower:/var/log# free -m

            total      used      free    shared    buffers    cached

Mem:          4052      1568      2483          0        233      1158

-/+ buffers/cache:        175      3876

Swap:          511          0        511

 

 

root@Tower:/var/log# tail -10 /var/log/syslog

Mar  1 20:51:30 Tower cache_dirs: ----------------------------------------------

Mar  1 20:51:30 Tower cache_dirs: cache_dirs process ID 28258 started, To terminate it, type: cache_dirs -q

Mar  1 20:51:37 Tower sudo:    root : TTY=console ; PWD=/tmp ; USER=myuser; COMMAND=/usr/bin/python /usr/local/couchpotato_v2/CouchPotato.py --daemon --config_file /boot/config/plugins/couchpotato_v2/settings.conf --pid_file /var/run/couchpotato_v2/couchpotato_v2.pid

Mar  1 20:52:20 Tower init: Re-reading inittab

Mar  1 20:52:20 Tower unraid-swapfile[29622]: Initiating unRAID swap-file.

Mar  1 20:52:28 Tower kernel: Adding 524284k swap on /mnt/cache/.unraid.swapfile.  Priority:-1 extents:112 across:2567208k

Mar  1 20:53:01 Tower kernel: NTFS driver 2.1.30 [Flags: R/W MODULE].

 

 

root@Tower:/boot/config/plugins# pwd

/boot/config/plugins

 

 

root@Tower:/boot/config/plugins# ls -1

Snap-5.0-1.plg*

cache_dirs/

cache_dirs-0.7-1jr.plg*

couchpotato_v2/

couchpotato_v2_unplugged.plg*

images/

inst_plugins.sh*

logitechmediaserver-7.7.2-plugin_v1.3.plg*

sabnzbd/

sabnzbd-0.4c-i468-bw.plg*

sabnzbd_unplugged.plg*

sabnzbd_unplugged.plg.old*

sickbeard/

sickbeard_unplugged.plg*

sickbeard_unplugged.plg.old*

simpleFeatures/

simpleFeatures.active.streams-1.0.5-noarch-1.plg*

simpleFeatures.activity.monitor-1.0.5-noarch-1.plg*

simpleFeatures.cache.dirs-1.0.5-noarch-1.plg*

simpleFeatures.core.webGUI-1.0.5-noarch-1.plg*

simpleFeatures.disk.health-1.0.5-noarch-1.plg*

simpleFeatures.dns.server-1.0.5-noarch-1.plg*

simpleFeatures.email.notify-1.0.5-noarch-1.plg*

simpleFeatures.log.viewer-1.0.5-noarch-1.plg*

simpleFeatures.s3.sleep-1.0.5-noarch-1.plg*

simpleFeatures.system.info-1.0.5-noarch-1.plg*

simpleFeatures.system.stats-1.0.5-noarch-1.plg*

simpleFeatures.web.server-1.0.5-noarch-1.plg*

slimserver/

snap/

utserver/

utserver.plg*