4.7 / 5.0b3 Testing Thread

We're all eagerly anticipating the new versions of unRAID (4.7 and 5.0b3) to come out.

 

Several members of the forum have test arrays and are poised to spring into action to test the new version.

 

This thread lays out the tests that the community would like to see performed on a test array to really check this thing out.  If people would "sign up" for some of these tests, and post their results, the community will be able to have greater confidence that all of the functionality of the new version is fully working.  If possible it would be good to get several people doing each test in each version - but some of them require a more complicated setup, so we'll just have to see if anyone signs up for some of the tests.  Please post in this thread referencing the tests you plan to run, and me or one of the other moderators will keep this post updated.  Make sure to specify if you will test on 4.7 or 5.0b3 (emphasis should be on 4.7 initially).

 

Please post other test cases and we will keep this post updated with all of the recommended tests.  This may be slightly overkill as we'll probably be duplicating tests Tom is alreadry running, but having actual users do the testing is still probably a good idea.

 

A - Basic regression

 

A1 - unRAID starts / array starts.  Access syslog using URL.  Nothing bad / unusual (EVERYONE running new versions should do this one!)

- carpet3 (Y), Rajahal, RichardU (Y), lionelhutz (Y), Josh (Y) (4.7)

 

A2 - unRAID GUI displays and works.  Start / Stop / Check / Restart / Shutdown buttons work.

- carpet3 (Y), Rajahal, RichardU (Y), lionelhutz (Y), Josh (Y) (4.7)

 

A3 - Samba shares work.  Can read/write from Windows / MAC.

- carpet3 (Y), Rajahal, RichardU (Y), lionelhutz (Y), Josh (Y) (4.7)

 

A4 - NFS shares work.  Can read/write from Windows / MAC.

- carpet3, RichardU (4.7)

 

A5 - Device page freeplay.  With array stopped, remove disks, move them around, etc. on device page.  DO NOT START TRY AND START ARRAY.  Make sure main screen is working.  Then put everything back and Swap 2 disk slots and start array.  Then stop, swap back, and start array.

- carpet3 (Y), Rajahal, RichardU, lionelhutz (Y), Josh (Y) (4.7)

 

A6 - Run full parity check.  Post prior version and new version MB/sec results (not observed, from syslog!!).

- carpet3 (Y), Rajahal, RichardU (Y),  lionelhutz (Y), Josh (Y) (4.7)

 

A7 - Downgrade to prior version (4.7 -> 4.6, 5.0b3 -> 5.0b2).  Verify array comes up properly in prior version.  (cannot be run if any 4K aligned drives are added to the array)

- carpet3 (Y), Rajahal, RichardU (4.7)

 

B - More advanced regression:

 

B1A - unRAID disk simulation.  Stop array, remove disk, start array.  Confirm unRAID simulates the disk

- carpet3 (Y), Rajahal, RichardU (Y) (4.7)

 

B1B - Trust My Parity Process works.  Then, stop array, put disk back in, and run the "Trust My Parity" procedure.  Verify it works and parity check is clean (maybe a few parity issues at very beginning of disk)

- carpet3 (Y (83 parity errors corrected)), Rajahal, RichardU (4.7)

 

B2 - Add precleared disk. Run preclear (sector 63 partition version) on a new disk.  Add it to the array.  Format it.  Verify partitioning.  Verify functioning.  Run speed tests.

- carpet3 (Y), Rajahal, RichardU (4.7)

 

B3 - Upsize an existing disk (using sector 63 partitions)

- carpet3  (Y (parity disk)), Rajahal, RichardU (4.7)

 

C - Partition Changes / AF Drive Testing

 

C1 - Add 4k aligned disk.  Preclear new (not previously in array) to 4K aligned partitions.  Add to array.  Set 4K align partitions.  Start array.  Format disk.  Verify partitioning.  Verify functioning.  Run speed tests.  

- Rajahal, RichardU (Y), Josh (Y) (4.7)

 

C2 - Remove EARS jumper test.  Remove jumper from existing EARS drive that has already been formatted for unRAID.  Run dd command to zero the MBR.  Preclear to 4K partitions (Joe L., is the dd command going to be required if you are gong to preclear?).  Set to 4K partitions.  Start array.  Format disk.  Verifiy partitioning.  Verify functioning.  Run speed tests.

- Rajahal (4.7)

 

C3 - Add Samsung F4 (with firmware update).  Run dd command to zero the MBR.  Run preclear (sector 64 aligned).  Add disk to array.  Set to 4K aligned partitions.  Start array.  Format disk.  Verifiy partitioning.  Run extensive verification tests (ideas here?).  Run speed tests.

- amp1 (see custom script below), Josh (Y, See Here) (4.7)

 

C4A - Rebuild 63->64.  Simulate failure of a disk (remove from array, start array, stop array).  Run dd command to zero the MBR.  Set new drive to (4k aligned partitions).  Start array.  Verify rebuild works. Josh (Y)

C4B - Rebuild 64->64.  Then simulate failure again, then replace disk with same disk and let it rebuild.  Confirm rebuild worked.

C4C - Rebuild 64->63.  Then simulate failure again, run dd command to zero the MBR.  Set disk to use (sector 63 partitions), and rebuild again. Verify rebuild works.  (If all three tests work, you'll be left back where you started).

- Rajahal (4.7)

 

C5 - Parity swap / disable.  Upsize parity, move parity drive to data drive.  (Both 64K - no preclear signature)

- OPEN

 

D - Add-on testing

 

D1 - Verify unmenu works

- carpet3 (Y), RichardU (Y), lionelhutz (Y), Josh (Y) (4.7)

 

D2 - Verify mymain works

- carpet3 (Y), RichardU (Y), lionelhutz (Y), Josh (Y) (4.7)

 

D3 - Verify powerdown works (press ctrl-alt-del or power button and watch server shut down smoothly)

- carpet3 (Y), RichardU (Y), Josh (Y) (4.7)

 

Dx - ... (need help from others on all of the add on testing that needs to occur)

 

E - Other

 

E1 - Speed test comparisons from 4.6 to new versions for data and cache drives

- carpet3, RichardU, Josh (See Y See HERE) (4.7)

 

E2 - AFP Testing (5.0b3 only)

- OPEN

 

Amp1 Custom Script (new array test w/ EARS and F4)

 

- Upgrade the firmware on the F4s (if you haven't already) (Y)

 

- Prepare flash with 4.7b1. (Y)

 

- Boot unRAID 4.7b1.  Confirm your key is working (all disk slots should be available on the devices page). (Y)

 

- Download newest preclear script. (Y)

 

- Preclear the four disks with the new preclear script (use "-A" option).  Carefully review all output to make sure drives are good.(Y)

 

- Add two of them to the array (1 EARS as parity and 1 F4 as disk1) (Y)

 

- Set them at 4K aligned partition in unRAID (won't matter since preclear will already have done it, but good to set it correctly) (Y)

 

- Start array (parity computes), format disk (Y)

 

- Confirm that drives are using sector 64 offset.  You can use myMain (I just completed an update that displays the partition alignment), or use the comand fdisk -lu /dev/sdX, where sdX is the device name. (Y)

 

- Add some data to the F4. (Y)

 

- Add 2nd F4 to the array (it is precleared so should be quick).  Format it.  Confirm data is still on the existing F4 (i.e., format only formatted the new F4) (Y)

 

- Confirm alignment again. (Y)

 

- Copy a bunch of data and beat the living crap out of the F4s, verifying that they are trustworthy.  (You might want to research situations that have caused data loss with the F4s and the old firmware, and try to do some of those things and make sure you 110% trust them with your most precious data.) (Y)

 

- Run parity check.  Make sure no parity errors. (In Progress)

 

- Stop array, go to devices page and unassign one of the F4s, start the array.  Verify you can access the missing disk (unRAID will be simulating it).

 

- Stop array, go to devices page, and assign the remaining EARS to the removed F4s slot.  Set drive to sector 64 partition alignment.  Start the array. Drive should rebuild onto the new EARS.  Verify EARS restores properly.

 

- Confirm alignment.

 

- Run speed tests of F4 compared to EARS

 

- Use "dd" command to clear MBR on the remaining F4.

 

- Stop array, assign F4 to a new disk slot, make sure to set it to use sector 64 alignment, start array.  unRAID will do its own internal clear.  Will take a while.  When done, allow unRAID to format the disk.  

 

- Confirm alignment.

 

- Write more data to the array.

 

- Final parity check.

I can do: A, B, D E

 

Have a system setup at the moment specifically for testing

I have an idle machine with three 2TB EARS drives waiting for a 4K unRAID install. I can do any test except those requiring a previous version, a MAC, or an F4.

 

I've got two 2 TB WD EARS on the way, so I'll be running some tests with them.  Maybe I'll do one with jumper and one without.

 

I plan on doing the following:

- All of A except A4 (I've never used NFS shares before, so I don't see any reason to start now) and A7 (see comments on A7 below)

- All of B

- All of C except C3 (I don't have any F4s and don't want any just yet)

 

Eventually I may get to D and E, but probably not for a while.

 

I'll try to do the above A-C tests next week some time (assuming 4.7 is out by then).

 

Regarding A7 - I believe that LimeTech has specifically said that it will not work.  No previous versions of unRAID will recognize a disk with the partition starting at 64 as valid.  Therefore, the array would not come up if any AF drives @ sector 64 were present.  Instead it would show missing disks.  If an array has only disks using sector 63 then it should work.

 

Another boon to this thread would be a list of all the linux syntax that performs the various functions and verification that many of these tests require.  I know almost none of it off the top of my head.

Regarding A7 - I believe that LimeTech has specifically said that it will not work.  No previous versions of unRAID will recognize a disk with the partition starting at 64 as valid.  Therefore, the array would not come up if any AF drives @ sector 64 were present.  Instead it would show missing disks.  If an array has only disks using sector 63 then it should work.

 

It will not show up as "missing", it will show up as "unformatted".

 

This is a great "acceptance test procedure", but please bear in mind, it's not "official".

Another boon to this thread would be a list of all the linux syntax that performs the various functions and verification that many of these tests require.  I know almost none of it off the top of my head.

 

How about we look at developing some sort of standardised testing technique for the above (and whatever anyone can think of), would go nicely on the wiki. Could include the order to do things, what commands are to be used etc

 

It would at least give us a base line to work from

Regarding A7 - I believe that LimeTech has specifically said that it will not work.  No previous versions of unRAID will recognize a disk with the partition starting at 64 as valid.

 

Was there any mention of using a drive with the partition starting at 64 in any of the 'A' tests.  Doesn't regression testing imply not using any new functionality - just testing that the old functionality hasn't been broken?

 

Note that the 'B' tests stipulate drives with the partition starting at 63 only.

Regarding A7 - I believe that LimeTech has specifically said that it will not work.  No previous versions of unRAID will recognize a disk with the partition starting at 64 as valid.

 

Was there any mention of using a drive with the partition starting at 64 in any of the 'A' tests.  Doesn't regression testing imply not using any new functionality - just testing that the old functionality hasn't been broken?

 

Note that the 'B' tests stipulate drives with the partition starting at 63 only.

 

Exactly right.  The A suite is basically to take your existing array, upgrade to 4.7, look at your syslog and check out a few basic things, and then downgrade to 4.6.  No new drives added.  Downgrading as part of that should work fine.  In fact nothing in Suite B would prevent downgrading either.  But if you do the C tests you would not be able to easily downgrade.

 

Will update to make it more clear that the downgrade test will not include any drives formatted with the new 4k aligned partitions in the array and update the people's names later tonight when I'll have more time.

Thank you guys, and bjp especially. This is a great idea and will give us mortal users a lot of confidence in the new releases before they're out of beta, and beyond.

I've updated the OP.  There is no problem haviing multiple people do each test.  4-5 each would be ideal.  I understand some of the more esoteric ones will be harder to find people with drives to support the test, but hope we'll get at least 2 for all tests.

 

Please post your intentions.  Section C is pretty big - don't feel you have to sign up for all.  Just doing one of the tests would be a big help.

 

I added a test for swap disable.

 

NOTE:  Only people with test arrays should participate, except perhaps Suite A - and then only after several other users have tried it out on test arrays.

 

Shouldn't we also test AFP functionality for 5.0b3?

Agreed.. AFP will required substantial testing in order to iron out any quirks and depending on the version of netatalk that is finally used, there may be some issues with Snow Leopard.

 

 

I don't own a Mac so I can't test AFP

Okay -

 

Added RichardU to perform some tests.

 

Added AFP testing. I know nothing of AFP, some if someone wants to create more detailed test cases, post them and I will add them.

 

Need more volunteers!!  Please be specific on tests you will do.

Ok - 4.7b1 is out!

 

Testers - please try to at least complete the Suite A and D tests quickly.  I know some of the Suite B and C will take longer.

 

Post your results.

 

Even if you didn't volunteer, if you run some of these tests please post your results in this thread.  It will help others evaluating when is the right time to upgrade.

I'm new to Unraid, have two of the SuperMicro CSE-933T-R760B NAS Network Storage Servers that were recently being sold on eBay, both destined to be Unraid boxes.  I've got Pro licenses for both boxes, but so far only one is being used for Unraid.  The second box I've just been using for updating firmware at the moment.  I've been slowly moving files over from 10 F4 Samsung Spinpoint drives in two 8-esata boxes connected to a Mac Mini.  I'm at a good point to do some testing, and I've got the weekend clear to do some testing.

 

What I've got right now to put into the testing:

* one fresh from eBay SuperMicro CSE-933T-R760B NAS Network Storage Server.

* two 2TB Samsung F4 SpinPoint drives with updates firmware.  They were previously GUID formatted, but I just switch them to unformatted MBR drives.  No format, partitions, or data on them right now, not precleared in any way.

* two 2TB EARS drives, still in the seled antistatic bags, no jumpers on them.

* four Macs in the house to test AFS (if I get talked into running 5.0b3).

* a second, in-operation Unraid server that I can use to compare things (same model SuperMicro NAS server, running 4.6, pro license, two more 2TB Samsung F4s, used as a file server for a Plex media-player front-end on the Living Room Mac Mini).

 

I've also been doing Linux since version 0.12 and know networking well.

 

Let me know what testing holes you have and I'll get going.  I'm guessing that my unused server and 2 unused F4s and two never-used, never-jumpered EARS drives would allow some comparison testing if desired.

 

I'm new to Unraid, have two of the SuperMicro CSE-933T-R760B NAS Network Storage Servers that were recently being sold on eBay, both destined to be Unraid boxes.  I've got Pro licenses for both boxes, but so far only one is being used for Unraid.  The second box I've just been using for updating firmware at the moment.  I've been slowly moving files over from 10 F4 Samsung Spinpoint drives in two 8-esata boxes connected to a Mac Mini.  I'm at a good point to do some testing, and I've got the weekend clear to do some testing.

 

What I've got right now to put into the testing:

* one fresh from eBay SuperMicro CSE-933T-R760B NAS Network Storage Server.

* two 2TB Samsung F4 SpinPoint drives with updates firmware.  They were previously GUID formatted, but I just switch them to unformatted MBR drives.  No format, partitions, or data on them right now, not precleared in any way.

* two 2TB EARS drives, still in the seled antistatic bags, no jumpers on them.

* four Macs in the house to test AFS (if I get talked into running 5.0b3).

* a second, in-operation Unraid server that I can use to compare things (same model SuperMicro NAS server, running 4.6, pro license, two more 2TB Samsung F4s, used as a file server for a Plex media-player front-end on the Living Room Mac Mini).

 

I've also been doing Linux since version 0.12 and know networking well.

 

Let me know what testing holes you have and I'll get going.  I'm guessing that my unused server and 2 unused F4s and two never-used, never-jumpered EARS drives would allow some comparison testing if desired.

 

 

You asked so I'll give you a meaty set of tasks.  Feel free to shorten it based on time you have available.  

 

- Upgrade the firmware on the F4s (if you haven't already)

 

- Prepare flash with 4.7b1.

 

- Boot unRAID 4.7b1.  Confirm your key is working (all disk slots should be available on the devices page).

 

- Download newest preclear script.

 

- Preclear the four disks with the new preclear script (use "-A" option).  Carefully review all output to make sure drives are good.

 

- Add two of them to the array (1 EARS as parity and 1 F4 as disk1)

 

- Set them at 4K aligned partition in unRAID (won't matter since preclear will already have done it, but good to set it correctly)

 

- Start array (parity computes), format disk

 

- Confirm that drives are using sector 64 offset.  You can use myMain (I just completed an update that displays the partition alignment), or use the comand fdisk -lu /dev/sdX, where sdX is the device name.

 

- Add some data to the F4.

 

- Add 2nd F4 to the array (it is precleared so should be quick).  Format it.  Confirm data is still on the existing F4 (i.e., format only formatted the new F4)

 

- Confirm alignment again.

 

- Copy a bunch of data and beat the living crap out of the F4s, verifying that they are trustworthy.  (You might want to research situations that have caused data loss with the F4s and the old firmware, and try to do some of those things and make sure you 110% trust them with your most precious data.)

 

- Run parity check.  Make sure no parity errors.

 

- Stop array, go to devices page and unassign one of the F4s, start the array.  Verify you can access the missing disk (unRAID will be simulating it).

 

- Stop array, go to devices page, and assign the remaining EARS to the removed F4s slot.  Set drive to sector 64 partition alignment.  Start the array. Drive should rebuild onto the new EARS.  Verify EARS restores properly.

 

- Confirm alignment.

 

- Use "dd" command to clear MBR on the remaining F4.

 

- Stop array, assign F4 to a new disk slot, make sure to set it to use sector 64 alignment, start array.  unRAID will do its own internal clear.  Will take a while.  When done, allow unRAID to format the disk.  

 

- Confirm alignment.

 

- Write more data to the array.

 

- Final parity check.

 

Let us know how it all goes along the way.

 

When the time comes for 5.0b3, you can transition to that and test AFP.

 

I would advise that you do not fully trust the array with the only copy of your data.  But after you're done with these tests, you'll probably trust this array more than your other one!

 

Thanks amp1!

Here are some initial 4.7 tests.

A1 - Starts fine. The first unmounting seemed to take forever. The next starts and stops were quick though.

A2 - Seems OK. The 4k setting stuck and didn't mess anything up.

A3 - Shares still exist. Can read and write to them. The speeds seem about the same. My first write was slow to start for some reason.

A5 - The disk management seems OK. FYI, you can not swap 2 disks on the devices page and start the array.

D1 - Works fine. Sickbeard, Couchpotato and SAB all worked too.

D2 - Seems OK. One odd quirk I noticed is the cache drive appears as NEW when the array is stopped. I will have to check this on an older version.

 

Peter

Here are some initial 4.7 tests.

A1 - Starts fine. The first unmounting seemed to take forever. The next starts and stops were quick though.

A2 - Seems OK. The 4k setting stuck and didn't mess anything up.

A3 - Shares still exist. Can read and write to them. The speeds seem about the same. My first write was slow to start for some reason.

A5 - The disk management seems OK. FYI, you can not swap 2 disks on the devices page and start the array.

D1 - Works fine. Sickbeard, Couchpotato and SAB all worked too.

D2 - Seems OK. One odd quirk I noticed is the cache drive appears as NEW when the array is stopped. I will have to check this on an older version.

 

Peter

 

lionelhutz -

 

Great job!  Thanks.  I will update the OP when I have time.  Could you go ahead and report the problem that you can't swap 2 disk slots in the announcements thread for 4.7b1?  I am not sure when this defect crept in, but that is something that is supposed to work. 

 

I will look into the "new" issue on the cache disk in myMain.

 

Let us know what else you discover.

 

Thanks!

AFAIK, you never could swap 2 devices, or for that matter even just move a drive to a new slot without doing a "Restore" click or command line initconfig. I think it's mentioned as possible in some documentation but I have no idea where.

 

I will post anything else if/when I try some other stuff. I'll likely just leave 4.7b1 running on my server for now.

 

The MyMsin thing has probably always been there, just no-one uses it with the array in a stopped state so it's never been noticed or just ignored if it was noticed. Likely because that disk isn't a protected array disk. With the amazing things MyMain does one little new flag isn't a big deal.

 

Peter

 

You are supposed to be able to exchange 2 disks.  I just tried it with 5.0b2 and it works fine. If you swap 2 disks (e.g., disk2 with disk6), and go back to the main page, the status will say "Two or more disks are wrong.", but next to the start button it says "Start will just record the new disk positions and bring the array on-line. We recommend you start a Parity-Check afterwards just to be safe."  And below the start button is a check box which will enable starting the array.

 

Tom explains the feature in this thread:

 

Reorder disks

 

Does this not work in 4.7b1?

You asked so I'll give you a meaty set of tasks.  Feel free to shorten it based on time you have available.  

 

- Upgrade the firmware on the F4s (if you haven't already)

 

Unless I run into problems, I should be able to get through all of the testing steps you suggest, though it may take more than just this weekend.

 

All 10 Samsung HD204UI/Z4 drives have the firmware update (so I don't get confused as to which are ready to go).  2 of them ready for preclearing on Unraid 4.7b1.  (6 of the drives still on a Mac, 2 in my Unraid 4.6 box.)

 

2 WD20EARS-00MVWB0 mounted in drive carriers and running firmware 51.0AB51 .  These 2 drives will be precleared on Unraid 4.7b1.  My understanding is that I shouldn't need to get new firmware for these, but I should probably grab wdidle3, boot into DOS and disable automatic head-parking with "wdidle3 /s0" so that Unraid can handle spinning down the drives.

 

Question: Should I do the wdidle3 disable auto-park change on the 2 WD20EARS drives before I start?

 

Let us know how it all goes along the way.

I'll keep you up to date.

 

When the time comes for 5.0b3, you can transition to that and test AFP.

As long as that's not too far off, I'll be glad to do 5.0b3 testing, but at some point I would like to move the drives off of the Mac and keep my backups on my second Unraid box (the one I'll be doing this 4.7b1 testing on).

 

I would advise that you do not fully trust the array with the only copy of your data.  But after you're done with these tests, you'll probably trust this array more than your other one!

I hear that.  I don't even trust Unraid 4.6 with all of my files yet -- copies of everything on my 4.6 box are still on drives attached to my Mac Mini disk server.  I really try to keep two copies of everything on at least two different servers.

Question: Should I do the wdidle3 disable auto-park change on the 2 WD20EARS drives before I start?

 

This is completely up to you.  If you don't then you load cycle count will tend to increase pretty fast, as I understand it.  You might do it on one of your EARS and not on the other and see how the LCCs compare in the end?  Ultimately I think it would be smart to run them that on both drives, and no real biggie to just do it at the start.

OK, I double checked and it does work. I was confused about it not being possible. I must have tried to move slots or something before and saw that didn't work.

 

Peter

OK, I double checked and it does work. I was confused about it not being possible. I must have tried to move slots or something before and saw that didn't work.

 

Peter

If swapping between ports currently assigned to the unRAID server, it will work.  I don't think it will if you try to use a different disk controller port than was currently assigned in the array.