A co-worker of mine already posted on this but we're coming up with more questions.
We successfully tested the migration of a test vFiler from one node to the other and back. Took 3-4 minutes each way for a vFiler with one aggregate and two "disks" (it's actually got a 3Par SAN on the back end).
Then we tried a vFiler with two aggregates of 14 and 15 "disks." It went from A to B with no trouble, took closer to six minutes.
When moving it back from B to A it blew up. Got the following at the CLI: tungsten-a> vfiler migrate -m nocopy fhdata@tungsten-b tungsten-b's Administrative login: root tungsten-b's Administrative password: Stopping remote vfiler.... vfiler migrate: Aggregate 'fhdata_aggr1' has failed and cannot be brought online. Problem bringing aggr fhdata_aggr1 online: CR_VOL_FAILED vfiler migrate: No aggregate named 'fhdata_aggr2' exists. Couldn't find aggr fhdata_aggr2 in order to bring it online vfiler migrate: No volume named 'fhdata_root' exists. Couldn't find flex-volume fhdata_root in order to bring it online vfiler migrate: No volume named 'fhdata_shared' exists. Couldn't find flex-volume fhdata_shared in order to bring it online vfiler migrate: No volume named 'fhdata_viddshared' exists. Couldn't find flex-volume fhdata_viddshared in order to bring it online Problem bringing vfiler fhdata volumes online. tungsten-a> tungsten-a> vfiler status -r fhdata Vfiler not found: fhdata tungsten-a>
At first there was no sign of the vFiler, aggregates or volumes on either A or B. After a few minutes, part of one aggregate was visible on A. A few minutes after that we could see both aggregates and all disks. Brought the aggregates online and the all the volumes.
The vFiler is still gone, no sign of it on either node.
Does the vfiler migrate command support vFilers with more than one aggregate? Note that all the volumes for the vFiler live on two aggregates and no other volumes are on those aggregates.
While we're on the subject, does anyone know where the configs for a vFiler actually live? Right now my hope is to rebuild the vFiler and reattach the volumes but I'm hoping to at least be able to see what CIFS shares it had, for example. We save a nightly tarball of the complete contents of etc$ so if it's in a config file somewhere I can find it.
On the phone with NetApp support right now but I'm hoping someone has experience with this in the real world. Hope to hear from you.
Randy
Some progress.
I added a CIFS share to the vFiler's root volume at the host node level and can see things like the cirs share config file and exports, etc. Copied off etc for a backup.
Tried recreating the vFiler from FilerView, intending to overwrite config info, but the vFiler wizard squawks at the "create vfiler" step, saying "Vfiler creation failed: Operation not allowed because storage is still in use by the owning vfiler." Looks like even though the vFiler isn't running or visible it's still stuck in there somewhere.
Tried offlining the volumes and got a different error that the etc folder couldn't be reached.
-----Original Message----- From: toasters-bounces@teaparty.net [mailto:toasters-bounces@teaparty.net] On Behalf Of Randy Rue Sent: Tuesday, February 07, 2012 12:14 PM To: toasters@teaparty.net Subject: vfiler migrate
A co-worker of mine already posted on this but we're coming up with more questions.
We successfully tested the migration of a test vFiler from one node to the other and back. Took 3-4 minutes each way for a vFiler with one aggregate and two "disks" (it's actually got a 3Par SAN on the back end).
Then we tried a vFiler with two aggregates of 14 and 15 "disks." It went from A to B with no trouble, took closer to six minutes.
When moving it back from B to A it blew up. Got the following at the CLI: tungsten-a> vfiler migrate -m nocopy fhdata@tungsten-b tungsten-b's Administrative login: root tungsten-b's Administrative password: Stopping remote vfiler.... vfiler migrate: Aggregate 'fhdata_aggr1' has failed and cannot be brought online. Problem bringing aggr fhdata_aggr1 online: CR_VOL_FAILED vfiler migrate: No aggregate named 'fhdata_aggr2' exists. Couldn't find aggr fhdata_aggr2 in order to bring it online vfiler migrate: No volume named 'fhdata_root' exists. Couldn't find flex-volume fhdata_root in order to bring it online vfiler migrate: No volume named 'fhdata_shared' exists. Couldn't find flex-volume fhdata_shared in order to bring it online vfiler migrate: No volume named 'fhdata_viddshared' exists. Couldn't find flex-volume fhdata_viddshared in order to bring it online Problem bringing vfiler fhdata volumes online. tungsten-a> tungsten-a> vfiler status -r fhdata Vfiler not found: fhdata tungsten-a>
At first there was no sign of the vFiler, aggregates or volumes on either A or B. After a few minutes, part of one aggregate was visible on A. A few minutes after that we could see both aggregates and all disks. Brought the aggregates online and the all the volumes.
The vFiler is still gone, no sign of it on either node.
Does the vfiler migrate command support vFilers with more than one aggregate? Note that all the volumes for the vFiler live on two aggregates and no other volumes are on those aggregates.
While we're on the subject, does anyone know where the configs for a vFiler actually live? Right now my hope is to rebuild the vFiler and reattach the volumes but I'm hoping to at least be able to see what CIFS shares it had, for example. We save a nightly tarball of the complete contents of etc$ so if it's in a config file somewhere I can find it.
On the phone with NetApp support right now but I'm hoping someone has experience with this in the real world. Hope to hear from you.
Randy
_______________________________________________ Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
Sounds like a bug...did support find an issue and file a burt? The vfiler itself keeps the configuration of shares in /etc like a physical controller, but the physical controller uses the registry in vfiler0 which isn't modifiable (without support). You could try vfiler create -r to recreate the vfiler with the existing root volume (then you don't have to reconfigure settings either), but sounds like something might be stuck in the registry if it doesn't report the vfiler exists but errors when creating with the exists error.
What did support say? Best to run with them after seeing the messages reported..
-----Original Message----- From: toasters-bounces@teaparty.net [mailto:toasters-bounces@teaparty.net] On Behalf Of Randy Rue Sent: Tuesday, February 07, 2012 12:38 PM To: toasters@teaparty.net Subject: RE: vfiler migrate
Some progress.
I added a CIFS share to the vFiler's root volume at the host node level and can see things like the cirs share config file and exports, etc. Copied off etc for a backup.
Tried recreating the vFiler from FilerView, intending to overwrite config info, but the vFiler wizard squawks at the "create vfiler" step, saying "Vfiler creation failed: Operation not allowed because storage is still in use by the owning vfiler." Looks like even though the vFiler isn't running or visible it's still stuck in there somewhere.
Tried offlining the volumes and got a different error that the etc folder couldn't be reached.
-----Original Message----- From: toasters-bounces@teaparty.net [mailto:toasters-bounces@teaparty.net] On Behalf Of Randy Rue Sent: Tuesday, February 07, 2012 12:14 PM To: toasters@teaparty.net Subject: vfiler migrate
A co-worker of mine already posted on this but we're coming up with more questions.
We successfully tested the migration of a test vFiler from one node to the other and back. Took 3-4 minutes each way for a vFiler with one aggregate and two "disks" (it's actually got a 3Par SAN on the back end).
Then we tried a vFiler with two aggregates of 14 and 15 "disks." It went from A to B with no trouble, took closer to six minutes.
When moving it back from B to A it blew up. Got the following at the CLI: tungsten-a> vfiler migrate -m nocopy fhdata@tungsten-b tungsten-b's Administrative login: root tungsten-b's Administrative password: Stopping remote vfiler.... vfiler migrate: Aggregate 'fhdata_aggr1' has failed and cannot be brought online. Problem bringing aggr fhdata_aggr1 online: CR_VOL_FAILED vfiler migrate: No aggregate named 'fhdata_aggr2' exists. Couldn't find aggr fhdata_aggr2 in order to bring it online vfiler migrate: No volume named 'fhdata_root' exists. Couldn't find flex-volume fhdata_root in order to bring it online vfiler migrate: No volume named 'fhdata_shared' exists. Couldn't find flex-volume fhdata_shared in order to bring it online vfiler migrate: No volume named 'fhdata_viddshared' exists. Couldn't find flex-volume fhdata_viddshared in order to bring it online Problem bringing vfiler fhdata volumes online. tungsten-a> tungsten-a> vfiler status -r fhdata Vfiler not found: fhdata tungsten-a>
At first there was no sign of the vFiler, aggregates or volumes on either A or B. After a few minutes, part of one aggregate was visible on A. A few minutes after that we could see both aggregates and all disks. Brought the aggregates online and the all the volumes.
The vFiler is still gone, no sign of it on either node.
Does the vfiler migrate command support vFilers with more than one aggregate? Note that all the volumes for the vFiler live on two aggregates and no other volumes are on those aggregates.
While we're on the subject, does anyone know where the configs for a vFiler actually live? Right now my hope is to rebuild the vFiler and reattach the volumes but I'm hoping to at least be able to see what CIFS shares it had, for example. We save a nightly tarball of the complete contents of etc$ so if it's in a config file somewhere I can find it.
On the phone with NetApp support right now but I'm hoping someone has experience with this in the real world. Hope to hear from you.
Randy
_______________________________________________ Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters _______________________________________________ Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters