About aborzenkov

aborzenkov · ‎2011-08-18

Release notes for 7.3.6 do not say anything about 62xx support, so it looks like error on download page indeed.

aborzenkov · ‎2011-08-17

If I read this KB correctly, LUN is not necessarily corrupt. So may be deleting could still be avoided. It actually suggests creating snapshot of current state and then doing lun cloning from it.

aborzenkov · ‎2011-08-17

Well, it just tells about "rate of change of data", without defining how "data change" is actually calculated. And how it is calculated is exactly my question

aborzenkov · ‎2011-08-17

Some things that come in mind - Check for errors on SAN ports filer/tapes are connected to. - Try to increase blocking factor. Max supported size is 256. - Try to use non-compressing device node. LTO5 in compressed mode has quite high minimal data speed requirements (I believe, around 95MB/s); this is barely what your system is capable of. It could fall back to start/stop leading to poor performance. You could also try to test dump to tape directly as the first step (using different blocking factor and/or device nodes). Just to eliminate any possible NetWorker quirk.

aborzenkov · ‎2011-08-17

It could also be possible to do lun clone (or flex clone) which is less destructive.

aborzenkov · ‎2011-08-17

Have you checked whether your tape drive is supported by your Data ONTAP version?

aborzenkov · ‎2011-08-17

I do not think there is clear definition of "snap delta" anywhere. Specifically, right now I had a case of delta between two VSM snapshots of 290GB, but transferred size was somewhere around 600GB. The difference is too large to be explained by metadata. So far my only explanation is that "snap delta" shows size of changed data, while VSM of course transfers both changed and new data. It is not the first time I noticed somewhat great discrepancy between estimation using snap delta and real amount of transfereed data. So - what exactly is counted in "snap delta"?

aborzenkov · ‎2011-08-17

I have FC's for the ESH2 modules, but I'm not sure if the 940 will again, see the ESH2 modules at the beginning of the loop followed by ESH modules? It will. You may need to check loop speed settings in case you have 1Gb/s controllers. for some reason termination was set to "on" Sure it was termination and not shelf speed? Speed mismatch will cause shelf to drop off loop.

aborzenkov · ‎2011-08-17

aggr options aggr_name

aborzenkov · ‎2011-08-16

Is volume vol_dc03 online? “vol status”?

aborzenkov · ‎2011-08-16

Please show result of “lun show -v” on filer.

aborzenkov · ‎2011-08-16

I can find no mount command in filer view orin my terminal window You mount volume on host, not on filer. On filer you share it (CIFS) or export it (NFS) How do you create a LUN form file? Just specify file path when creating a LUN. Beware, LUN content will be whatever garbage happens to be in file right now But in the worst case you may be able to delete it again using "lun destroy" lun create -f <file_path> -t <ostype> [ -o noreserve ] [ -e space_alloc ] <lun_path>

aborzenkov · ‎2011-08-16

Actually it is possible to create LUN from a file. May be worth a try before finally removing it

aborzenkov · ‎2011-08-16

It is too low speed, but there could be many factors contributing. For a start, try to estimate theoretically possible backup speed by dumping to null (use the same volume/directory you try to backup normally): https://kb.netapp.com/support/index?page=content&id=1011894. Watch how long each phase of dump takes.

aborzenkov · ‎2011-08-15

Well … there will obviously be some interruption on this port before link is established on new speed. On the total, if you have multiple paths from hosts to filer and use multipathing software, it should be transparent.

aborzenkov · ‎2011-08-15

Start with stracing cp command to find out what system call exactly returns EPERM. I suspect it is attempt to (explicitly) change file mask/permissions, because file is obviously created.

aborzenkov · ‎2011-08-15

It is true that NDMP rules out host based (software) encryption. But BE supports hardware based encryption; HP LTO5 does offer encryption support and should be supported by BE. It also is expected to work with NDMP - in this case SCSI pass through is used to upload keys into tape drive. So it is quite possible that everything that is needed to use encryption is already in place. Of course, it means that keys are managed by BE and only BE will be able to restore data. I do not know whether (hardware based) encryption is separately licensed feature of BE.

aborzenkov · ‎2011-08-11

You could try to set target adapters in FAS3140 to 2Gb/s to verify that it really has any impact on this: fcp config speed 2

aborzenkov · ‎2011-08-10

According to https://kb.netapp.com/support/index?page=content&id=2013323 and https://kb.netapp.com/support/index?page=content&id=3010858 there is 90 seconds grace period for filer to come back in active/active configuration. If this time is not exceeded, you may want to open case with support to investigate. But really, why would you want to reboot in active/active? Normal way is to use cf takeover/cf giveback.

aborzenkov · ‎2011-08-09

This explains why used + avail do not match total. As for why %used and Used do not match … well, I guess (total – avail)/total simply provides better approximation. Avail can be less than (total – used) under many conditions. But to get definite answer you need to look at the sources after all ☺ In general I guess “avail” is the only thing that finally matters.

aborzenkov · ‎2011-08-09

OK, so two possibilities - Some space is reserved for snapshots (look at “df -r”) - If volume is thin provisioned, “available” reflects free space in contained aggregate. What “df -Ah aggr-for-volrh2” says?

aborzenkov · ‎2011-08-09

Show output of "df -h volname"

aborzenkov · ‎2011-08-09

Well ... OracleVM is basically a Xen server with some extra management layer on top and SnapProtect appears to support Xen. And it never says "Citrix XenServer" but rather "Xen Server" which appears like they do support generic Xen. So it is quite probable that it will work. Not that all features are available (e.g. snapshots are apparently supported for Vmware/Hyper-V only). I guess you could ask for a trial version and test.

aborzenkov · ‎2011-08-03

If you run space reclamation after creating snapshot, all your freed space is locked in snapshot. show output of "df -h" and "df -rh" for volume in question as well as "lun show /your/lun/on/this/volume"

aborzenkov · ‎2011-08-03

To restore individual file from LUN snapshot you would need to mount it (usually it involves creating a writable copy, LUN clone or FlexClone) and copy file from mounted snapshot to your active LUN. If you have SnapDrive installed on your host(s), mounting is handled by SnapDrive, so the only potential problem is to find snapshot that contains required file version ☺ Another potential issue is permissions to perform this operation. You may want to look at something like SnapProtect that keeps catalog of files in snapshot and makes it very easy to browse for different file versions.

DOT 7.3.6 available for FAS6280 & FAS6240

Re: Create/destroy LUN

What exactly "snap delta" counts?

Re: netapp ndmp backup via networker 12-14MB/s

Re: Create/destroy LUN

Re: netapp ndmp backup via networker 12-14MB/s

What exactly "snap delta" counts?

Re: FAS3020 Repourpose, adding shelf from DECOMM FAS940

Re: Show raid group size

Re: Create/destroy LUN

Re: Create/destroy LUN

Re: Create/destroy LUN

Create/destroy LUN

Re: netapp ndmp backup via networker 12-14MB/s

Re: Cannot run blades at 4Gb since upgrading to new filer heads

Permission denied on NTFS volume shared by NFS

NDMP and Encryption

Re: Cannot run blades at 4Gb since upgrading to new filer heads

Takeover starts when reboot is done.

Re: Discrepancy between %Used and Used Capacity values

Re: Discrepancy between %Used and Used Capacity values

Re: Discrepancy between %Used and Used Capacity values

SnapProtect for OracleVM?

Volume and LUN size mismatch

Re: managing snapshots