Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[fordem]

Fw 1.06 includes a new feature which supposedly automatically rebuilds the RAID array.

When you go into the RAID menu page and enable it, you get a warning saying

Please note that enabling Auto_Rebuild will automatically erase all data on a newly inserted hard disk drive without confirmation in order to rebuild the RAID array

Are you sure you want to enable Auto-Rebuild

Although this may sound like hot-swapping is now supported - it apparently is not - hot unplugging a drive to simulate a failure produces the expected amber LED and email, hot plugging a replacement drive gets a pink/purple/white LED and the the new drive is not recognized in the admin pages.

Restarting the unit it appears to boot up and just sit there with all the normal blue LEDs - log in to the web admin page and it prompts you to format the new drive - isn't that what it is supposed to do "without confirmation"?

Am I missing something here?

[fordem]

OK - for the curious - here's what it's all about.

If you have Auto-Rebuild enabled - when you reboot the unit after it formats the drive, it will rebuild the array without prompting you - if you have Auto-Rebuild disabled, it will not rebuild the array until you manually initiate the rebuild.

In hindsight - kind of obvious based on the wording of the warning  but how practical is this?

I believe what was needed was a way to manually initiate a rebuild, but, was there a need to not do it automatically?

Where I'm coming from is this - you can skip the formatting of the newly installed drive, but once you format it for use as RAID1 is there really anything else you're going to one to do other than rebuild the array - what do you think?

[STR3T]

I was wondering about that when I saw it mentioned, too.  Thanks for the note.

Made me think re: it's purpose:

What if you had RAID1, but you have a spare HD laying around but different
manuf. or size.  You can plug in and not reinitiate RAID1 but JBOD.  Then you could move files over to the 2nd HD manually and have immediate backup while waiting on a 2nd HD order to matchup w/ RAID1 again...just a thought on this.

[thor79]

I think it's just an option for control freaks. Making it auto rebuild might freak those people out...so presenting the option to disable auto rebuild satisfies those people. IMO Having options is always the better way to go.

[fordem]

I was wondering about that when I saw it mentioned, too.  Thanks for the note.

Made me think re: it's purpose:

What if you had RAID1, but you have a spare HD laying around but different
manuf. or size.  You can plug in and not reinitiate RAID1 but JBOD.  Then you could move files over to the 2nd HD manually and have immediate backup while waiting on a 2nd HD order to matchup w/ RAID1 again...just a thought on this.

Interesting concept, but I don't think it can be done - I tell you what - I have a freshly built RAID1 config right now, with no live data on it, let me put your idea to the test.

[STR3T]

Did we crash you hard w/ the experiment fordem?  Seriously, since you didn't think it would occur in that manner...what do you think are the possible solutions to adding a 2nd unmatched HD into a prior RAID1 config?

[fordem]

I have tried it yet - yesterday was a busy day here.

To answer your question - what happens when you add a second 'unmatched' drive to a prior RAID1 configuration will depend on the nature of the "unmatching" - typically a larger drive is no problem, and, on the DNS-323, what will happen - if you allow it - is that the unit will simply partition the drive so that it has a "data" partition of a size matching the existing drive, and the remaining space is left unutilized.

If I recall correctly - pre 1.03 firmwares have three partitions per drive, and 1.04 and later have four (assuming that the entire drive as one volume - if you have a RAID + JBOD configuration the numbers go up by one), so if you had a 1.03 RAID1 configuration using the entire volume of say a pair of 80GB drives, each drive would have three partions - a drive fails and you replace it with say a 160 GB drive, that drive would be formatted with four partitions, three identical in size to the ones on the remaining 80GB drive and a fourth one containing all the remaining space.

The above I have already done with the pre 1.06 firmwares - so - with this new option - the drive is going to be formatted, but not synchronized - which is essentially just a block copy from one data partition to the other.

I've never tried replacing the failed RAID disk with a smaller one - pre 1.06 it was essentially an exercise doomed to failure - but - let's consider your theory.

First - I suspect you're not using the term JBOD in the same sense as D-Link does - D-Link considers JBOD to be on large logical volume containing the storage space of all the available physical volumes, and I think you're using it to mean individual volumes.

Second - what I expect to happen is that the partitioning logic (partitioning and formatting are not the same) in the DNS-323 will simply refuse to work, given the fact that it cannot create a partition of the required size - but - we shall see.

[fordem]

As I expected (or should that be suspected ?) - a smaller drive produces the message "The replacement hard drive does not have enough capacity to support the RAID1 volume.  Please insert a hard drive with more capacity."

Now - here's an interesting one for you ...

Second go-round - I inserted a drive with "more capacity", skipped the format, went and deliberately disabled Auto-Rebuild.  Restarted, got the prompt to format, clicked next, saw the screen prompting for "Auto-Rebuild" which was disabled and clicked next - the unit formatted the drive successfully and upon the restart - promptly proceeded to "auto rebuild" - I checked the status page, it is rebuilding, I can see it from the drive activity LEDs (that's what first got my attention) and when I look in the Tools/RAID page - Auto Rebuild IS disabled. 

[STR3T]

Ok, so...so much for my theory on AutoBuild feature. 

Any ideas then what it is  there for?  Just in case you want the extra click to confirm you want it to do what it's going to do anyway?  Funny.

[fordem]

After numerous reformats - I haven't quite figured out what this is supposed to do.

Several times my DNS-323 automatically rebuilt the array even though "Auto-Rebuild" was disabled, and this last time it didn't, but it doesn't prompt me to rebuild - even worse it says "completed" for sync time remaining, and I KNOW it hasn't done any rebuilding.

Right now I've had enough of this - tomorrow, maybe, I'll take another shot at it, or maybe I'll just reformat again for RAID1 and put the unit back in service.

[fordem]

Ok - a new day - I'm less fatigued, less frustrated - I put my spare disk in a Windows system to delete the partitions for another "go-round" and notice that I have three partitions -2 x ~500MB, 1 x 7 GB and ~230 GB of empty space - this is definitely not correct.

Check the DNS-323 and the status page size I have a 7GB RAID1 volume with a pair of 250GB disks - what did it do with the rest of my space??

I have seen this before with earlier firmwares - so I reformat, first as separate volumes and then as RAID1 and the problem is resolved - at least for now.

I "fail" a disk, install a clean replacement, and once again - it automatically rebuilds - even though "Auto-Rebuild" is disabled.

Here is what happens - you replace a failed drive, you power up the unit and login and you are propmted to format the drive (only if the drive is large enough - if it's too small you'll get a message telling you that), on completion of the format, you'll get a prompt to restart the unit and on the restart, when you login, you'll be taken to the TOOLS/RAID page which will tell you that the RAID volume is synchronizing.

The reason why I didn't see this the last time around is that with the 7GB RAID volume, it only took a few minutes to synchronize.

Based on the experience documented in this thread - I draw the following conclusions ....

1) the wording of the messages is misleading - the DNS_323 will automatically resynch the array, without prompting regardless of the setting chosen for "AutoRebuild" - you CANNOT disable "AutoRebuild", presumably you can trigger a manual rebuild.

2) D-Link's partition & format script logic is still seriously flawed, and the new feature doesn't fix anything - it simply provides a way to trigger a manual resync, and confuses the end user with the misleading messages.

Come on D-Link - I'm sure there's enough space in flash memory to include a "manual partition page" for advanced users - show the disks be make, model, serial number and capacity, and let the user choose which one to format, let the user select the partition size.

Add a telnet server - to let the user directly manipulate the process.

Add a disk utility that runs on the PC and controls the partition/format process - it can't be that difficult to do.

[fordem]

OK - after many hours, formatting & reformatting, synching & resynching, I think I have figured this out.

First - the manual rebuild button is not always available, you cannot rebuild the array, just because you want to (OK - you can - but not by just clicking the button), the button's default state is disabled, and it will only be enabled, if & when, the unit detects a degraded array.

Please be aware that whilst a failed disk will result in a degraded array, there are other causes for the array to become degraded, and the unit does NOT necessarily indicate a degraded array through the LEDs or by sending an email - it is entirely possible to have a degraded array and not know about it until you notice that one of the disk activity LEDs is not flashing or you check the status screen.  D-Link - it might be an idea to add degraded array to the list of email alerts.  Whatever logic is used to check the disk space status can at the same time check the array status and send the appropriate alert message (and maybe flash the LED amber or something)

Manual rebuild mode first - this is the default state, and this is how things used to be in pre 1.06 firmwares - if the unit detects a degraded array, it does nothing other than display this condition on the status page.  The rebuild button is active, and if you click on it, the unit will rebuild the array, copying from the disk it sees as valid, to the disk it thinks is not.

Auto rebuild mode - this is the new option in this firmware - if the unit detects a degraded array and thinks the disks are both good, it will automatically, without any prompting, rebuild the array, copying from the disk it sees as valid, to the disk it thinks is not.

In case you're curious as to why I point out that the array is rebuilt, copying from one disk to the other - there is a reason - I have seen other devices rebuild the array from the disk with the older data, overwriting the current data, so I KNOW it is possible, and without a thorough investigation of the logic on which this unit makes it's determination, I would assume that this unit can also do the same thing.

[D-Link Multimedia]

Ok, so...so much for my theory on AutoBuild feature. 

Any ideas then what it is  there for?  Just in case you want the extra click to confirm you want it to do what it's going to do anyway?  Funny.

It is mostly for users that do not want to access the web gui or do NOT have access to the web gui but still need to perform the task of keeping the raid array up. If you could simply shut down, replace the failed drive and start it up without having to log into to anywhere, click any buttons would you use it? Or would you rather do the same steps but add the logging in, clicking of buttons and watching the sync timer count away. Your choice, thats why its an option .

[fordem]

It is mostly for users that do not want to access the web gui or do NOT have access to the web gui but still need to perform the task of keeping the raid array up. If you could simply shut down, replace the failed drive and start it up without having to log into to anywhere, click any buttons would you use it? Or would you rather do the same steps but add the logging in, clicking of buttons and watching the sync timer count away. Your choice, thats why its an option .

I [exploitive deleted] to be the bearer of bad news - but - if that was the intent, it does not work.

Based on my tests, documented above, the unit will NOT, repeat NOT, format a new drive without prompting you first.  It will resyncronize the array without prompting you IF autorebuild is enabled AND IF an out of synch array is detected AND IF the hard drive that the array is rebuilding to is recognized as having the correct partition structure AND has a serial number matching the one stored in flash.

As you can see there are quite a few conditions that need to be met - I did test with a drive with the correct partition structure and a different serial number, it prompted me to format and I skipped the format and successfully did a manual resynch, but on the next bootup, I was prompted again to reformat the "new drive".

Any time you swap a drive on this unit, expect it to prompt you to format it (or to swap left to right), because it tracks the drives by the serial number - you can skip the format, but, it will prompt you on the next boot up.

Bottom line - you cannot replace a failed drive and have the array rebuilt automatically - you MUST login and agree to at least the formatting of the new drive.

By the way - the expletive that was deleted was h.a.t.e. - and whilst I'm about it - what the heck is an exploitive?