NETGEAR is aware of a growing number of phone and online scams. To learn how to stay safe click here.
Forum Discussion
Sandshark
Mar 24, 2019Sensei
Destroying EDA volume switched form XRAID and now I can't switch back.
I have a 516 with two external eSATA enclosures, an EDA500 and another brand 5-bay one the NAS identifies as an EDA500. Something happened while the EDA500 volume was in scrub. The NAS became unresp...
StephenB
Mar 25, 2019Guru - Experienced User
You aren't supposed to be able to use XRAID with the expansion units, so the real puzzle here is how you were able to use XRAID before.
There's a lot of risk in having a single volume span two or three chassis, so you really shouldn't do that.
Sandshark
Mar 25, 2019Sensei
I don't have volumes spanning chasses (I agree, bad idea -- I would have lost more with this event if I did). And so long as I didn't do that (which it actually wouldn't let me do in XRAID mode, anyway) or have multiple volumes in a single chassis, the NAS was perfectly happy with XRAID being enabled and actually expanding the volumes. This appeared to be normal behavior. My 4200V2, on the other hand, would not allow two volumes in the single chassis with XRAID when I ran some experiments before putting it in service as my backup device. Note that trying to switch to XRAID is not complaining that I have multiple volumes, as it did with two volumes in the 4200V2, it's complaining the volumes are expanded. And I know I did once have to disable XRAID in order to get the "lost" volume of the EDA restored into the active arrays, and I was able to swap it back later. But apparently, that was before the expansions. It was actually a situation very much like where I am now, except I think the EDA volume was intact then and isn't now.
I started with 4x6TB + 2 * 4TB in the 516 and 4*3TB in the EDA, so I started with two RAID groups. I moved the 4TB's to the EDA (filling the last slot and swapping out one 3TB), replacing them with 6TB's in the 516, thus creating a second layer in the EDA as well. And then I later replaced the remaining EDA 3TB's with 4TB's. Each time, XRAID did the expected "magic" and expanded the array. This was all a while back, probably on OS around 6.7.something. I later put the 3TB's into the other brand eSATA chassis and created yet another volume in XRAID on it. That was definately in OS 6.8.something or better.
Now, maybe with the EDA500 being twlighted, that support is no longer there in 6.9+., or at least tested (and unknowingly broken). Or, as you say, maybe it was an unitentional feature that unfortunately went away But, wow, it sure worked well for somehting unplanned. It would be a shame if the capability is lost, but it sure would be nice to know if it's worth me trying to get the unit back into XRAID or if it's a fool's errand.
I'm trying an experiment using a 312 I bought used for who-knows-why (and now, I guess I know why). I put it in FlexRaid mode, hooked to it the EDA containing three spare drives, and I'm currently syncing a RAID5 volume. The existing 312 volume is not expanded. When it's done, I'm going to see if I can switch to XRAID or not.. It's all data I don't care about, so a perfect test set-up. I also plan to export and -re-import it to see if I can plan to rebuild the array and restore the data using the 312 and move it to the 516. Unfortunately, I need the EDA "back", so I don't have an infinite time to play with it. But if there is anything else you'd like to see tried with this combo, let me know soon and I'll try to work it in.
- SandsharkMar 27, 2019Sensei
So, here is progress thus far:
I created a new volume on the EDA500 while in FlexRAID mode and then successfully changed to XRAID. I put a few files on it and then exported that volume and booted with the EDA disconnected, and the unit had "forgotten" the volume. I then re-connected the EDA with power off and booted, and the volume was imported, with the unit remaining in XRAID mode. As is typical with an import, I had to go in and set share access again, as it had been cleared by the export/import.
When I did that, md0 (the OS partition) began resync as a RAID 1 as well. The log didn't show the import or start of resync, just the export and later completion of resync. But TOP showed it was syncing md0 and the drive access LEDs were blinking like crazy. That only took a few minutes, and then all activity stopped on the EDA, all LEDs were solid, and everything worked just fine. So, it appears it was the expanded volume causing the problem switching to XRAID, just as the unit reported. But I'm not sure if it was the expanded volume or a corrupt volume that was preventing import of the original EDA volume.
I'm next going to re-create the original volume and restore the files to it using the 312. The reported CPU use during the creation of my test volume never topped 20%, so it's the eSATA speed that's the bottleneck and I don't expect things to go slower on the 312 compared to the 516 since the 312 will be doing nothing else.
What remains to be seen is if, after exporting the other eSATA volume, I can switch the 516 back to XRAID with just the one (expanded, two RAID group) volume so I can then import both eSATA volumes while in XRAID mode or if I'll have to destroy and re-create the data volume with the eda volumes exported to get back to XRAID. I'm not going to try that till the EDA volume is re-created and files are restored to it, so I don't have two file restorations in progress simutaneously if destroying the data volume becomes necessary.
It's going to take a few days before I'm ready for the next step. Resyncing a 5x4TB volume on an EDA500 and then restoring some 8TB of data will be time consuming.
- SandsharkMar 30, 2019Sensei
And, the saga continues:
I successfully re-created my eda1 volume and restored data to it using the 312. So, I was ready to do the same with the other eSATA chassis if it became necessary. So, I exported the eda2 volume that was still connected to the 516, disconnected the eSATA chassis, and re-booted. But, I still could change back to XRAID. The system still complained that I cannot because I have expanded volumes. Yes, the data volume (just the one, in spite of the fact that the error message says plural) is expanded, consisting of two RAID groups, but it is XRAID compatible, because XRAID expanded it. So, I'm not sure if it really is strictly because XRAID was turned off and on with an expanded volume or because XRAID was automatically turned off by the Destroy that I cannot go back. I swear I did turn it off and on before with an expanded volume. But I'm not completely sure, so maybe it's just always been that way, maybe something changed in an OS update, or maybe it was the Destroy automatically doing it that created the issue.
So, I left the NAS in FlexRAID mode and successfully imported both eda volumes (one at a time). But, if I ever need to expand any volume, I'm stuck with the FlexRAID limitations until I decide to bite the bullet and either destroy and re-create the data volume before I do any further expansion. I'm not even sure that will do it, maybe I have to factory default.
The warning that told me it was going to switch to FlexRAID when I Destroyed the volume said it was doing so in order that I could re-create it. But why does it assume I want to re-create the volume at all, that I'm not just removing it because I no longer need it? If I were Destroying the primary/only volume, then I can see expecting it to be re-created. Have I discovered a bug, or at least a not completely thought out implementation?
- HopchenMar 30, 2019Prodigy
Hey Sandshark
It is an interesting one but I have fairly limited experience with EDA chasis' - they just weren't that wide-spread. Anyway, one thing that keep popping into my mind is: why are you set on XRAID? Surely Flex-RAID is the way to here? I have never actually tested how XRAID behaves with EDAs. My guess is... funky :)
I don't think XRAID should ever be part of the equation when involving EDAs. You mention that: "But I want the expandability of XRAID back.". What is it that Flex-RAID can't give you here? Is it the auto expandability you desire?
FYI, md0 will be across all disks I believe so that sync of md0 is probably perfectly normal. I reckon, just go Flex-RAID and save the headaches?
Related Content
NETGEAR Academy
Boost your skills with the Netgear Academy - Get trained, certified and stay ahead with the latest Netgear technology!
Join Us!