Yes, having node 1 pool0 and node 2 pool1 disks in one stack (and thus in one shelf) is supported. It is not recommended as per TR-3548 (search for "mixed disk pool configurations") but it's supported.   You might need to manually assign any replaced disks though, since the controllers can't always figure out by themselves which controller should get any new disks assigned (especially if you have >1 spare per node)   -Michael ... View more

So what real-world applications do you have that do not work with the current implementation as-is?   Changing something like this would require extensive compatibility testing with operating systems and network stacks going back as far as LAN Manager 3.0 for DOS, so unless you have a very compelling use (and are a very big customer, I guess  ) I doubt that anything will be changed there. ESPECIALLY if the actual use-case turns out to be "I just want the result of this API call to be what Windows version X returns" instead of, say, "my application which is a multi-million-dollar SAP or CATIA installation, doesn't work with the current implementation" 😉 ... View more

If you have any volumes with UNIX security style, then usermapping needs to be configured. Also you need to do "vsever cifs create" (do not confuse it with "vserver active-directory create" which is something different!) to create a machine account in AD (it's not enough to just manually add a machine account into your AD domain). You can check the secd.log (you can get it via http://<netapp node IP>/spi ) for any errors regarding usermapping and/or security. Of course if you have users in LDAP/NIS that you want to map to (instead of, say, just mapping all windows user to one specific UNIX user) then you need to setup LDAP/NIS as name service   But honestly, your partner (the one who sold you the NetApp) should be able to help you with that. Also, it's not often a good idea to use a single SVM for file and block storage at the same time. It's better to separate these into multiple SVMs ... View more

If your data is in any way important for you then I would strongly advise against using such large RAID groups, especially with large SATA disks. You only have two disks of redundancy and the reconstruct of a single 4TB disk may take many days (depending on filer load, I have seen 3TB disks having reconstruct times of 4 or 5 days under heavy load), so when another disk fails in that timeframe you're running completely without protection and every single-block error will lead to corrupt data. Since the difference in capacity is only about 10% I would suggest you use two RAID-DP groups of smaller size (i.e. 11 and 12 disks, with one hot spare)   if you still want to do it, make sure you understand the implications. Type the following to enable the override NetApp> options raid.raid_dp.raidsize.override on   Regards -Michael ... View more

If it's only filedata that you're interested in, then yes, a simple ufsrestore from Solaris should still work (it definitely did work with NDMP dumps from OnTap 7.3.x). BUT you will most certainly lose all non-UNIX permissions (NFSv4 ACLs, CIFS ACLs, etc.) and, for LUNs, you will lose all metadata and only get the (raw) LUN file. Getting data back from a LUN will be a bit difficult but should be possible if you're not afraid of using hex-editors and loop-mounts under Linux 😉   But if you were using SnapMirror to Tape (SM2T) to dump whole volumes, then a NetApp filer will be your only hope of ever recovering data from it (if you don't want to pay Kroll OnTrack for a recovery 😉 However, you can restore these volumes onto any model of NetApp filer that runs at least the same OnTap version as the filer from which the dump was taken, and which supports at least the same sizes for volumes etc. (so no restoring a 20TB volume onto an old FAS2020 but that should be obvious... ) ... View more

Yes, that link seems to be broken or internal (even NetApp partners cannot access it)   To answer your questions, there is only one reason where Aggregate snapshots can help you, and that is if you need to restore a COMPLETE aggregate (with ALL volumes in it) to a previous version, for example if you deleted a volume by accident. There is no way to restore single volumes from an aggregate snapshot. It might also help with recovery if you should ever get a corrupt WAFL (highly unlikely).   The default for NetApp systems has been for quite some time now to disable aggregate snapshots completely. The only reason you might still need them is if you're running MetroCluster (or Local SyncMirror) because then the Snapshots are needed for mirror reconstruction. But that's something that happens behind the scenes and doesn't help you as an admin with anything   So yes, feel free to disable Aggregate snapshot reserve (if you're not on a MetroCluster) ... View more

In that case I would strongly suggest you upgrade to 7.3.7P3 as there are a lot of bugs that have been fixed since 7.3.5*, especially with regard to SMB2 support   You should ask your partner/reseller if you need help with upgrading, even though it's rather easy (download new software, install on filer, do takeover/giveback 2 times) there might be dependencies on other software (SnapManager or other 3rd party tools) that you might need to check, and your partner should be able to help you assess these issues.   You can check if smb2 is enabled by entering NetApp> options cifs.smb2.enable cifs.smb2.enable    on   If it's on, you can disable it with NetApp> options cifs.smb2.enable off   and re-enable it with NetApp> options cifs.smb2.enable on   The settings take effect immediately, and if you have Windows clients that were accessing the filer through SMB2 when you disabled it, these clients might need a reconnect with SMB1, which will happen only after some time (depending on the Windows version and settings it can be up to 15 minutes) or a client reboot (usually faster than just waiting). It should not affect existing sessions but since windows continuously tries to disconnect and reconnect, you should be prepared to expect these long timeouts.   Auditing is a bit more involved, you should probably skim over the relevant documentation (File Protocols Access Guide) and over the basic Windows concepts before trying to implement it (misconfiguration can under some circumstances kill your performance or fill up your root volume with tons of logs)   ... View more

The first Event (ID 563) happens when a file is opened with FILE_DELETE_ON_CLOSE which is usually used for temporary files. Netapp will automatically delete that file when the last open file handle to it has been closed. Note that you (or rather a program) can also use that flag to force deletion of a file that is currently in use by another program (it still needs the delete-permission to the file itself of course, you cannot delete random files that way ) See for example here or here   The second event was introduced with Windows Server 2003 (I think) and is thus not really a "non-standard" event. See here or here for a few details ... View more

What OnTap version are you running? I remember there once was a strange bug with SMB2 that manifested itself in a similar way (files created in a subdirectory disappeared and were later found a few directories further up). But that was with newly created files, not with pre-existing files. Try disabling SMB2 on the filer to see if that helps. Note that this requires all CIFS clients that are currently connected via SMB2 to reconnect to the filer.   You can also enable CIFS Auditing on the volume to see who/what does the move operation if it's indeed client-initiated   Also you should upgrade to the most recent OnTap version (8.1.4P9 if you're currently at or before 8.1.x, or 8.2.3P5 if you're already on 8.2) ... View more

Hm. Looks like you have a shelf which assumed a soft FC ID. This is not good. It normally means that you have a dumplicate shelf ID in the same stack. But in this case it's the internal disks which is odd. I would suggest you open a support case for this. You might need to re-seat some IO modules or reboot the filer to clear this inconsistent state ... View more

You should remove the "cluster" license on the remaining node (since you won't be usin it anymore if you transition to a single system) and reboot the filer. That way it will "forget" that it is a HA pair. You might even get a slight performance boost, since the NVRAM can be used completely by that one node instead of having to be shared with the other node. Then, pull the second controller from the chassis for a few centimeters so that it doesn't confuse the surviving node ("I am not a cluster but I keep seeing another controller installed?!?") ... View more

In cDOT is it possible to have volumes in different vServers that have the same name. This must somehow be mapped to the underlying D-Blade (nodeshell). So the first volume you create for any vServer with a certain name usually gets this name 1:1 mapped to the D-Blade, but as soon as you create another volume with the same name (but in a different vServer) the system has to rename that second volume.   But generally you really really shouldn't look at the volume names on the D-Blade (and NEVER EVER change anything volume- or aggregate-related on the nodeshell!) as the cluster makes certain it does everything correctly. Eventually the possibility to see volumes on the nodeshell will go away anyway since it is not really needed anymore ... View more

What the others said, please do an uprade sooner rather than later. 8.0.2 is rather old and since it's an unpatched release contains a few annoying bugs... your NetApp partner should be able to tell you more about it and help you upgrade your system.   The hardware in question isn't, by chance, a FAS3140? We had similar problems with a few 3140's once which were only fixed by upgrading everything to a more recent version (OnTap, BIOS, disk/shelf firmware, etc.) ... View more

do a "disk show -a" on both nodes, wait a few minutes, then do it again and compare the results. if they don't match, compare the output of "disk_list" on both nodes (you need to be in diag mode for this). This shows the lowest physical level of the disks. If they don't show up there then it's most probably a hardware issue of some sort. If the disks are there then it's simply a mismatch of what the node(s) think the disks are. This can probably be resolved via "disk assign", "disk remove_ownership", etc. ... View more

Depends on the context. It could mean a disk with a foreign aggregate on it. Or a LUN which can be (or is being) imported via FLI on an 8.3 system   Or, generally any LUN visible to the system through an FC initiator with which it doesn't know what to do ... View more

Yes, but they are not actively used. And if one fails or is removed, the system should automatically elect another one as replacement (otherwise every failed mailbox disk would crash the filer in the same way) so if you give the system enough time between the "remove ownership" commands it should work without causing a panic ... View more

Does the filer even boot? When you connect to the SP and type "system console", do you get into the bootloader? if so, try "autoboot" and post the results here. If you get no response at all (i.e. only the "press ctrl-d to return to SP" or whatever it says) and it doesn't react to ctrl-c then it's probably busted. But if you get a bootloader prompt or even some boot messages that could help a lot in diagnosing.   You could also try the serial console instead of the SP, in case it's just the SP that is busted. Try connecting via serial and powering up the system. You should see boot messages from the BIOS and then get a bootloader prompt ... View more

You probably need to remove it from WINS manually (if you have WINS configured), which is a bit of a PITA since you have to delete the domain's entry on all WINS servers at the same time and wait for the DCs to repopulate it with the correct list of DCs.   On the AD side your AD admin should have used dcpromo.exe (or the new, fancy GUI variant thereof) to remove the DC from the AD ... View more

It will only work when the domain is right, because the NetApp is joined to exactly one domain and can only validate credentials from that domain. So if you joined DOM1 then the user DOM1\foo would be mapped to unix user foo. If a different user (with the same name) from a different domain tries to connect, say, DOM2\foo, he would get an "access denied" since the filer has no means of checking his credentials (the filer knows nothing about domain DOM2, and even if it knew, since it is not joined to DOM2 it could not check the user's credentials) ... View more

Under what user account are you running the batch file? If it runs as LOCALSYSTEM it has different security credentials and probably no access rights to the filer. Does the MKDIR work if you try it (as regular logged in user) from a command prompt? Does CIFS work at all on the filer or do you always get that error? It could be that the clock is out of sync on the NetApp (Kerberos allows for maximum of 300 seconds clock skew). It might also help to reset the password on the filer's domain account by using "cifs changefilerpwd" or reconnect the DCs via "cifs resetdc". If all else fails you can still do "cifs terminate; cifs setup" to re-add the filer to the AD ... View more

Can't you just "cf disable" (or rather "storage failover modify...") before removing disk ownership from those 2 mailbox disks, and re-enable it again afterwards? ... View more

VSM destinations will only be converted to 64bit after you break them. ... View more