About aborzenkov

aborzenkov · ‎2012-01-29

There is no documented way to switch off disk shelves. They (unfortunately) remain powered on even if system shuts down as result of overheating.

aborzenkov · ‎2012-01-28

Please show aggr status -r output.

aborzenkov · ‎2012-01-28

Locally attached SAS has to just copy data in memory. Externally attached arrays are accessed over multilayer protocols and additional devices (FC/LAN switches), which introduce additional overhead. So each single IO will likely have higher latency for external array than for locally attached one. If your benchmark is single threaded, each IO has to wait for completion of previous one. Thus you achieve lower throughput (for a single thread) than in locally attached case. As you asked for published results - look e.g. at http://www.vmware.com/pdf/vsp_4_vmxnet3_perf.pdf. It shows appr. 8000 IOPS for 10GbE network. Assuming 512B IO size you get just 4MB/s; with 4KB request size - 32MB/s.

aborzenkov · ‎2012-01-28

You can ignore them. ha-config-check apparently was not updated to handle modern filers, including cluster interconnect.

aborzenkov · ‎2012-01-28

If all connected systems are using multipath connection and correctly configured for failover, you just do cf takeover/cf giveback on each head. Steps are provided in Data ONTAP upgrade guide as well as ASUP upgrade adviser. Most settings to ensure that LUNs remain online should have been taken on connected hosts, not on filer.

aborzenkov · ‎2012-01-28

You should have added that it applies to Data ONTAP 8.x only.

aborzenkov · ‎2012-01-28

I am quite sure that the PassMark test with its default settings is running only one thread. Which mostly explains the results.

aborzenkov · ‎2012-01-28

Are you able to create any directory under /etc at all? Do you have CIFS license? What is security style of root volume?

aborzenkov · ‎2012-01-27

It is filer\C$\etc\sshd\root not filer\C$\etc$\sshd\root.

aborzenkov · ‎2012-01-27

Without knowing - concurrency (how many threads were running in parallel) - queue depth for each thread (number of outstanding requests) - IO request size - read/write ratio for each test it is hard to even start guessing ...

aborzenkov · ‎2012-01-27

I believe disk encryption requires special drives and all disks in a filer must support it. So it probably a bit more limited than “anything that can support 8.0”.

aborzenkov · ‎2012-01-27

I am not sure AZCS applies to arbitrary disk types (at least, for now). I got impression that currently it applies only to new high density shelf.

aborzenkov · ‎2012-01-26

If you have RAID_DP and option raid.background_disk_fw_update.enable is on (default), disk firmware update starts automatically as soon as you install new firmware file and happens in background.

aborzenkov · ‎2012-01-26

Add your E-Mail address to autosupport.partner.to option. Addresses in autosupport.to receive only notifications about serious problems in 7.3.x and above.

aborzenkov · ‎2012-01-25

IIRC 8.1 will obsolete free licenses. I wonder what is the situation there.

aborzenkov · ‎2012-01-25

You can't. You will have to use some external processing to extract information, change it and replace the line again. And yes, if you changed just /etc/exports, you will need to run "exportfs -a" (or just for this entry) to activate it.

aborzenkov · ‎2012-01-25

But he's going to use SnapMirror anyway, so if it is possible to resume it later for each volume from existing snapshots, it could be interesting option. Otherwise the biggest downside is need to copy all the data just to create volumes. I think this is where power shell toolkit could be quite useful. Not being PS expert, I assume it is something like $source = connect-NaController -Name source $target = connect-NaController -Name target get-NaVol -controller $source | New-NaVol -controller $target -Name $_.Name $targetaggr -Size $_.TotalSize ... Thanks to http://communities.netapp.com/docs/DOC-6293 for explaining how to work with multiple controllers. Message was edited by: Andrey Borzenkov

aborzenkov · ‎2012-01-23

As long as nobody complaints, that it is slow ...

aborzenkov · ‎2012-01-20

Most likely you need to setup user mapping from Windows user on your machine to Unix user on NetApp. I am not familiar with Windows NFS implementation so I do not know how to do it.

aborzenkov · ‎2012-01-20

Oh, BTW, this problem was unrelated to single file restore, it was the whole backup that was being restored. To speed up single file restore the only option is to use DAR, but it has to be enabled during backup and you have to use specific utilities to restore (not all of NetWorker restore tools support DAR – check administration guide). --- With best regards Andrey Borzenkov Senior system engineer Service operations

aborzenkov · ‎2012-01-20

Sorry, no. I never got around to retesting it myself and customer never complained about slow recovery after that either (I do not say it became faster, just that nobody complained …) It is extremely hard to troubleshoot such issues simply because you do not perform multi-terabyte recover every day and when you do, you need to complete task and have no time to troubleshoot ☹

aborzenkov · ‎2012-01-19

Adding environment variable to profile is universal fix. Environment is property of current process inherited by its child processes. Your login session has no parent/child relationship to rc scripts and so has no way to inherit environment from them.

aborzenkov · ‎2012-01-18

If you do not have space for snapshot, how can you do snapmirror? In any case, you can't clone LUN to different volume. You can create FlexClone of volume itself, delete unneeded LUNs in each and then split. To finally free space you will need to remove snapshots with deleted LUNs. You will need flexclone license for it. But as already answered, you can move LUN into qtree and use QSM. This won't give you more space in volume though; snapshot is done on volume level anyway.

aborzenkov · ‎2012-01-17

please show output of "aggr show_space" from both filers

aborzenkov · ‎2012-01-17

You can't move volumes this way. You can move files between different volumes, but not volumes themselves.

Re: Shut down the FAS systems (with disk sheves) using script

Re: I have a sparing problem where a 1TB SATA spare will not spare out a failed 750GB SATA disk on a...

Re: What is the expected disk performance?

Re: upgrade ontap software ndu on active/active cluster

Re: upgrade ontap software ndu on active/active cluster

Re: passwordless ssh to filer

Re: What is the expected disk performance?

Re: passwordless ssh to filer

Re: passwordless ssh to filer

Re: What is the expected disk performance?

Re: Newbie: NetApp Encryption

Re: Aggregates with AZCS checksums in 8.1

Re: updating firmware on disk prior to ontap upgrade

Re: How to automatically send weekly status?

Re: [netapp1: raid.root.unmirrored:error]: Root volume is not mirrored. A takeover of this filer may...

Re: adding entries to a /etc/export file

Re: copy 120 volumes from one filer to another

Re: When would be time to upgrade controllers?

Re: can't connect to NFS share

Re: Extremely slow NDMP recover with NetWorker

Re: Extremely slow NDMP recover with NetWorker

Re: SnapDrive process failure in Linux

Re: Lun moving

Re: Why is total space size on aggr between controllers are different with same amount of disks?

Re: Cut & Paste Volume?