Similar Questions

• Move the "Data Service" volume ZFS on OSX Server

  Hi all

  I would like to move all the data service to a ZFS volume on a Mavericks OSX Server. This option can be found here:
  Server.app-> your server-> settings
  and is called "Data Service". As someone has already done this? It works or what I have to do a manual config, whenever the server restarts?

  Hi all

  So I'll answer my own question.  The answer is no, in fact, the app Server does not detect even the zfs volume.  I also try to cheat and to synchronize the/Library/Server volume ZFS and softlink then the original file to the copy of the zfs volume, but App Server would no longer work.

  Then Mac OS x still don't like ZFS much.

  I reformatted the drive with appleRAID, and then I could move everything using the button in the application.  I noticed that this original file is still there.  If it's really just an arrangement written somewhere.  Anyone know where this parameter is written?  Maybe I could get data on the ZFS drive sync and write this setting by hand.

• UFS to ZFS migration / conversion recommendations

  We have hundreds of different systems of UFS filesystem in our production environment and little to no window of opportunity to bring them to a static state to copy the data. Does anyone have any ideas on how I could address the issue? Or has experienced this themselves?

  Thanks in advance for any help, advice, or by pointing and laughing at my doooooom!

  -Kammer

  Hello

  Well what I did at a customer who had about 1 t data and also because downtime stritly minimum.

  I used rsync.

  If you create a zpool and riding in a different directory.

  then, who, with rsync, let synchronize you synchronization for x hours.

  And you this several times at the end of the ERP synchronization and less time.

  Then stop you your application, and then syncing with rsync.

  then remove the old mount point and set the mount with the zfs at the old point

  And start the application.

  In this way you carried out will know you can do a restore if it doesn't

  Hope this helps

  Filip

• ASM is to replace ZFS?

  I know that Oracle I like to think so, but I'm not sure that this new file system is really ready for prime time?

  Let me know your thoughts, or Oracle may be able to answer this question directly.

  Smitty

  Why do you think ASM could replace ZFS? I don't really see any connection between the two. Maybe you can compare ZFS BTRFS, to some extend, but the DSO?

  Information on the storage of the DSO, such as disks and devices managed by ASM, groups are known only by the + ASM instance. An Oracle database instance can connect to an ASM instance to get the necessary metadata for database file extensions. This information is stored in Oracle database APG and Oracle can write to the data files the same way, as it does without ASM.

  ASM is a feature called ADVM/ACFS to allow you to mount an ASM diskgroup as a file system, but it cannot be used for Oracle database files. ASM works at the file level and knows how to optimize Oracle database files, but does not include any similar concept as ZFS. The only way that the operating system can access the database files stored in ASM is using the ASMCMD command line utility, for example, to copy files between the DSO and the operating system.

• backup rman Exadata with zfs

  Our company has recently decided to start backup rman ZFS instead of regular tape backup.

  I would like to know what is the right measures to check if ITS configured right device of ZFS, the implementation of rman backups, etc.

  I'd appreciate someone here can point me to the right document.  I read ZFS Q and A and backup using ZFS best practices. All very unclear and seems none of them provide a digital step step.

  I wonder if Oracle provides an instruction step by step for the DBA.

  9233598-

  Q documents & you examined include the "Oracle ZFS storage pool: FAQ: Exadata RMAN backup with Oracle ZFS Storage Appliance (Doc ID 1354980.1)" Note? This note has a lot of information and is updated regularly and includes information about configuring ZFS and RMAN backups to the device of ZFS.

  Some useful books of whites, an I have used in the past and has been updated again last year: Oracle Exadata with the Sun ZFS Storage Appliance protection: best practices in Configuration and backup and recovery Performance and best practices using Sun ZFS Storage Appliance with Oracle Exadata Database Machine

  These documents will in many details - including providing examples of scripts.

  HTH.

  Thank you

  Kasey

• 5.5 ESXi and ZFS do not play well together?

  A couple of weeks, I've improved some HP DL380G6 of ESXi ESXi 5.1 to 5.5. One of those contains a Solaris 10 VM with a small volume of ZFS with about 120 GB of data and there are regular backups of that provided by using 'zfs send '. With ESXi 5.1 'zfs send' would complete in about an hour. With ESXi 5.5, it is not finished.

  Try to get a 'zfs send' manually in the command results up to a certain point and then stop. Him 'zfs send' stopped at 8 GB, 20 GB, 40 GB, 60 GB on different series. Whenever it starts to run in misfortune, messages such as these show upward in vmkernel.log:

  (2014 01-07 T 05: 51:16.698Z cpu9:32806) NMP: nmp_ThrottleLogForDevice:2321: Cmd 0x1a (0x412fc3bbc1c0, 0) to dev "mpx.vmhba0:C0:T0:L0' on the way 'vmhba0:C0:T0:L0' failed: H:0 x D:0 x 2 P:0 valid 0 x 0 sense-data: 0 x 5 0x20 0x0. Bill: NO

  (2014 01-07 T 05: 51:16.698Z cpu9:32806) ScsiDeviceIO: 2337: Cmd (0x412fc3bbc1c0) 0x1a, CmdSN 0 x 12711 world 0-dev "mpx.vmhba0:C0:T0:L0" has no sense H:0 x D:0 x 2 P:0 valid 0 x 0 data: 0 x 5 0x20 0x0.

  (2014 01-07 T 05: 56:16.708Z cpu15:32812) NMP: nmp_ThrottleLogForDevice:2321: Cmd 0x1a (0x412fc2626700, 0) to dev "mpx.vmhba0:C0:T0:L0' on the way 'vmhba0:C0:T0:L0' failed: H:0 x D:0 x 2 P:0 valid 0 x 0 sense-data: 0 x 5 0x20 0x0. Bill: NO

  (2014 01-07 T 05: 56:16.708Z cpu15:32812) ScsiDeviceIO: 2337: Cmd (0x412fc2626700) 0x1a, CmdSN 0 x 12716 world 0-dev "mpx.vmhba0:C0:T0:L0" has no sense H:0 x D:0 x 2 P:0 valid 0 x 0 data: 0 x 5 0x20 0x0.

  
  

  When you look at the performance of the system of ESXi, everything seems reasonable until the moment where performance disappears. On Solaris, the system has more than 2 GB of free RAM and makes a constant rate of e/s until this happens and throughput IO falls to zero.

  
  

  Also has anyone seen this behavior?

  If so, how you haven't solved the situation?
  

  The solution to this has nothing to do with the disks, it was networking.

  To solve the problem, I installed vmware tools in the Solaris 10 host, changed the type of e1000g to vmxnet3 NETWORK card and changed the IP configuration on Solaris 10 using the hostname.vmxnet3* instead of files hostname.e1000g* files.

  So, it would seem that there is some corner performance e1000g behavior cases that appear to be inconsistent between the driver in ESXi 5.5 and Solaris 10 that were not already there.

• ZFS and fragmentation

  I don't see Oracle on ZFS often, in fact, I've been called also to meet the first. The database had heavy issues of IO, both by the undersized IOPS capability, but also a lack of performance on safeguards - part reading of it. The ability of the OPS / s has been extended easily by adding more LUNS, so I stayed with very poor bandwidth experienced by RMAN, reading of data files. iostat showed that during a single copy of the data file (cp and dd with 1MiB block), the average block IO size was very small and various wildly. I was afraid of fragmentation, so I started the test.

  I wrote a small C program that initializes a datafile 10 GiB on ZFS and several times the fact

  1-1000 random 8KiB written with random data (content) limits of 8KiB (mimicking a 8KiB database block size)

  2 - a full reading of the end data file at the end in 128 * 8KiB = 1MiB e / s. (imitating the copies of the file of data, backups rman, complete sweep of the table, index full scans)

  3 - goto 1

  It is a data file that gets the random writes and is full analyzed to see the impact of the written word at random on the multiblock read performance. Note that the data file is not cultivated, all scriptures are on existing data.

  Although I expect fragmentation (it must come from somewhere), is dismayed by the results. ZFS sucks really big time in this scenario. Ext3, on which I ran the same tests (on the exact same storage), the timings of reading have been stable (around), ZFS started with 10 ms and went up 35ms for 1 128 * 8Kib IO after 100,000 random writes to the file. It has not reached the end of the trial again - service times continue to increase, so the test takes a very long time. I expect to stop somewhere - the file might finally be completely fragmented and cannot be broken up more.

  I started noticing statements that seem to recognize this behavior in some Oracle whitepapers, as otherwise unexplained tips to copy data files regularly. Indeed, copy the file back it defragments. I cannot tell you all this means downtime.

  On the production server, this issue got so bad that the migration to a new system of different files by copying the files will take much longer that restore backup to disk - backup to disk are written once and are not fragmented. They are lucky, the application requires full scan of the table or index fast full, or perhaps unfortunate scans, because this question were would become impossible to ignore earlier.

  I have observed the fragmentation with all the logbias and detailleerecordsize parameters that are recommended by Oracle for ZFS. ZFS caches could use RAM 14GiB (and moslty made), bigger than the file itself.

  The question is, of course, am I missing something here? Who else has experienced this problem?

  Hi Jan-marten,

  Well, I had a customer business of multi-billion dollar, running Oracle on ZFS (Solaris x 86) infrastructure and it works pretty well. In fact, ZFS introduces a "new level of complexity", but it's worth for some clients (especially the snapshot have for example).

  > S o that I stayed with very poor bandwidth experienced by RMAN read the data file

  Maybe you hit a question I/O timing. I wrote a blog post on a ZFS problem and its synchronization with RMAN I/O behavior: performance (backup) [Oracle] RMAN with synchronous i/o depends on the OS limits

  Unfortunately, you have not provided enough information to confirm this.

  > I have observed the fragmentation with all the logbias and detailleerecordsize parameters that are recommended by Oracle for ZFS.

  How the layout of the ZFS pool looks like? Is the entire database in the same pool? As a first step, you need to separate the log files and data in different ponds. ZFS works with "copy on write".

  How the free ZFS space look like? Depending on the free space of the ZFS pool, you can delay the "ZFS" snood or sometimes let (based on usage of the pool), it disappears completely.

  ZFS snood can be traced with DTrace. For example like this:

  shell> dtrace -qn 'fbt::zio_gang_tree_issue:entry { @[pid]=count(); }' -c "sleep 300"
  

  Concerning

  Stefan

• ZFS SA hand as FRA on a grid Exadata

  Looking to see if ZFS Storage Appliance (we buy a period.) can be used as the FRA for the Exadata instead of the regular + RECO DG.
  The people there went thru' a similar thought process? I read that ODA can be connected to a SA of ZFS for HCC & backup purposes, but I was wondering if there is some official/Metalink document note that says: ok Exadata can have FRA on ZFS.
  
  Obviously, * if * we have our FRA on ZFS, it just takes a simple 'switch' RMAN to copy and avoids the pain in the neck, strip speeds.

  In this type of configuration, you can set two archivelog destinations: an on your + RECO diskgroup, and the other on your ZFSSA.

• PowerEdge 1900 and Poweredge 840 and NO hardware RAID SATA HD, FreeNAS with ZFS

  I searched these forums and the web in general a concise my questions answer, but nothing is done.  I just happen to find someone who is using 'old' technology and it retarget as much as you might think (I am a broke student by the way).

  I bought a Poweredge 1900 Server & server Poweredge 840 ebay about last year.  I've been busy and have yet to work with these two units.  I wanted to use the PE 1900 as a file server and then push a backup of these files on the server off site, the PE840.  I intended to buy four 1 TB or four disks SATA 2 to (WD Black or RED WD) for each system, respectively.

  I've been through a few videos on youtube the other day and discovered FreeNAS.  This seems to be a much better solution than the use of Ubuntu Server, but it also made me think about the hardware Raid with these systems.  If FreeNAS uses ZFS (software raid controller), you want to avoid a hardware raid controller.

  My current assumptions, please correct me if I'm wrong on any of the following:

  I think that each of these systems to support drives up to 2 TB?, I think that I can use the SATA drives and not SAS, four discs in each system drives.

  I believe that I can boot from USB flash drive (a quick, which I intend to install FreeNAS on which they recommend)

  I think I can connect readers to the Perc/Raid Controller controller (& just turn 'off' RAID), or simply directly on the motherboard.  (What woud be the advantages/disadvantages of the use of the card or simply by using the Council)

  I intend on working with what I have right now (cheap and available), then once I get this configuration work that I take it out of the sand/lab sandbox, upgrade the processors and ram up my budget will allow and go live with the installation program (if it is still possible)

  From now on, they are sitting in a corner, begging to consume energy, and fans of polute the soundscape with their cries (PE1900 precisely).  I know that I could pursue other options but I already these two units.

  I really want to utilze FreeNAS its serving capacity file and ZFS for its reliability.  Is it possible with this camera?, or should I junk them and move on?

  Thank you for your answer!

  I've done some research more and since FreeNAS is built off of FreeBSD supported hardware is the same.  I couldn't find a list for FreeNAS specifically so I looked to FreeBSD.  I was still confused about the material and decided to go forward with the poweredge 840, for two reaasons, there already 8 GB of ram and it is not as heavy to move on (from 1900). I shoot the raid controller (for simplicity) and I plugged in three 250 GB SATA drives and a 40 GB SSD.  I had to set the bios for the usb key is recognized and also due to adapt to be considered as a hard drive, then had to define the sequence of boot to the usb drive, acting as a hard drive.  Once all that has been said and done, I wrote my FreeNAS image on my USB and booted up.

  It took maybe 10 minutes and I was up and running (once I had configured the hardware and bios), I created a RaidZ (as proposed FreeNAS) and with my three 250 GB drives, it produced 453 GB of usable disk space.  The 40 GB SSD is used as L2ARC (from my cache of understanding), I tried the installer with and without the 40 GB SSD.  The difference was impressive, just opening the photos or streaming video on my laptop Win 8.1 (Dell 17r) is almost flawless, as if the files were stored locally.

  I was impressed by how quick and easy to use FreeNAS, simple and fast network shares can be configured with permissions that works quite simply not in less than 5 minutes.  I don't have Windows Server down, but generally configuration active directory takes a little more effort.

  I am satisfied for the moment and continues to experiment with the Poweredge 840 until I commit to a processor upgrade.  Once I acquire 16 GB memory for my Poweredge 1900 I will perform with her, as it has only 1 GB now, crazy.

  Thank you for your help.

• Can I install Oracle 10 g on "Zettabyte File System" ZFS in Solaris?

  Hi all, I really need your help.
  Can I install Oracle 10 g on "Zettabyte File System" ZFS in Solaris?
  Or it must be installed in the NFS file system? I mean I read things on ZFS, but not sure if Oracle suport it.
  Thank you for your opinion and knowlage.
  Eugene

  It is supported, details here (search for Oracle):

  http://www.solarisinternals.com/wiki/index.php/ZFS_Best_Practices_Guide

  Werner

• ZFS or ASM? upgrade 9i to 10g the new Solaris 10 OS, Sun Fire x 4500

  I need to move/upgrade oracle 9i db to 10g of old sun server to a new server, which is the Sun Fire x 4500 running on Solaris 10. It has no external table, all the disks are internal. My question is which file system is better? ZFS or ASM? From HIS perspective, I like zfs better because it is very easy to manage. But metalink says that he does not support zfs. Even with the DSO when you configure the ASM instance you still need to relay on mirror/raid its oracle home operating systems software. Anyone have experience or test it before?
  
  Thank you.
  
  Joe

  Hello

  ASM is a better option.

• Where can I find the latest research on Solaris 10 zfs and without?

  I know Arul and Christian Bilien did a lot to write about storage related to Oracle technology. Where are the latest discoveries? Of course, there are some exotic configurations that can be implemented to run the optimizer, but is there a set of "best practices" that usually work for 'most people'? Are there common advice for people using Solaris 10 and zfs on SAN hardware (IE, CEM)? Double-stripe must be configured with meticulous care, or it works "well enough" just by taking some rough assumptions?
  
  Thank you very much!

  Hello

  I have a few links that I used:

  http://www.solarisinternals.com/wiki/index.php/ZFS_Best_Practices_Guide

  http://www.solarisinternals.com/wiki/index.php/ZFS_for_Databases

  These are not really new, then you may have encountered them already.

  The following ZFS blog list:

  http://www.OpenSolaris.org/OS/community/ZFS/blogs/

  Yet once, it seems not be huge activity on blogs featured here.

  Jason.

  --
  http://jarneil.WordPress.com

• The network for ZFS supports "virtual storage device.

  Hello

  I'll put up a 5.1 ESXI machine to test the performance / proof of concept with a "storage device" in a virtual machine on a host computer.

  Right now I use a version of solaris (community edition of nexenta) to test.

  With success, I was able to set up the machine with through disk controllers and add a vswitch initial with a configured vmkernel port and a group of ports for virtual machines. -J' was able to set up comments solaris vm with two adapters e1000 in the vswitch, to be used for the management and the other configured with an address within the same private network that the port on the management... network vmkernel and things seem to work well.

  I wanted to test if there is no improvement in performance using the vmxnet3 adapter and have a completely 'virtual network' dedicated solely to the nfs traffic.

  So I created an additional vswitch without adapters attached to it, with an another vmkernel port and a second virtual machine portgroup and added a vmxnet3 adapter to my "vm storage" to go on this vswitch.

  Before I do, I was wondering if VMCI or anything else already going to do what I want (i.e. ESXi allow communicate with the computer in virtual storage on nfs with speeds above that the limitation of the e1000 adapter would provide by default)

  I say this because my initial test with iozone gave me numbers that seemed beyond 1 GB...

  So, to summarize, the e1000 would be made generally better than 1 Gbit of networking 'local' (in the same host); and (assuming that the handles of the tcp/ip stack of the guest OS UI VMXNET3 well - IE drivers, configuration, the efficiency of the OS.)  the VMXNET3 adapter would be even better.

  It can even better if the uplink to the physical network is faster.  The rate of 1GBit 'link' basically means nothing.

  If someone could point me to a 'best practice' guide or the discussion about setting up isolated networks 'private host' for best performance with NFS virtualized shared storage, I'd appreciate it. My plan is to implement three hosts, with the majority of local data stores to the same host, with little nfs traffic going to each box...

  Well, there's really only 1 way to set up a network of 'host-only', so I don't think you will find a guide BP.

  Also the terminology escapes me a bit, because 'virtual private network', 'private network' have other meanings in other contexts. Don't know what to call a network which has not set the physical network adapters and is localized to a single ESXi host...

  Most called it a network invited only.

• All appearing as the same volume in ESXi Solaris iSCSI/ZFS volumes

  Currently, I'm sharing via Solaris iSCSI disks. I googled around on this problem and came across many discussions around June on a flaw in the implementation which seems to match what I see.

  I had created originally only one volume, and it worked fine. After that I created several others (using shareiscsi = on) I scanned again and saw that all targets have been found, but they all seemed to be clones of the original goal, and it wouldn't let me add as VMFS volumes.

  I tried to create a new target with multiple LUNS via iscsitadm, but it showed as having a single LUN - once again, the same clone.

  I have another target on the same server that is connected to a thin Win2008 area. He can see all the different targets for what they are.

  From what I understand, the problem concerned the GUID is the same, but I see that the GUIDS are different on all targets.

  According to all the posts I read, this problem has been resolved in the OpenSolaris some time ago, but it is perhaps still in the standard Solaris 10 kernel?

  Help would be appreciated. Thank you very much!

  I think you'll like better than Solaris, OpenSolaris, and it is supported by Sun with a contract.

  It works and it will be always every month before more recent improvements than Solaris only.

  http://blog.laspina.ca/

• format hard drive for windows and larger files, TV supports

  Hello

  pls suggest me with the solution to the following problem.

  I need to know what format of hard drive works with mac, windows, & TV too, even for large files and quality bluray movies

  FAT32 only supports files up to a maximum size of 4 GB, which would be certainly much too small for Blu - Ray quality media.

  exFAT supports files of unlimited size effectively, Mac and Windows can read and write to exFAT to format external drives but neither can boot from such a unit. If a TV or other system supports exFAT drives, you need to check yourself formatted.

  NTFS supports files of unlimited size indeed, Mac can only read NTFS, it cannot format or write to NTFS, unless you get an add-on. Of course Windows supports NTFS. Again, you will need to check if your TV etc supports NTFS. If you want to add full support for NTFS to a Mac, I recommend this - https://www.paragon-software.com/home/ntfs-mac/

  HFS + is the current format of Apple and new supports files of unlimited size effectively. Mac can of course read, write and format disks. Does not support the Windows as a standard HFS +, but still, you can buy an add-on for Windows Add this feature. It is extremely unlikely that a TV would support HFS +. If you want to add full support for HFS + for Windows, take a look at this - http://www.mediafour.com/software/macdrive/

  There are also other formats of disc for example ZFS and BTRFS, Ext3, Ext4 and soon SATF, but in this case, they are not relevant.

  Note: as indicated above, in many cases, I said 'actually unlimited', it is of course a limit, but one so great that he will strike you not for individual files, including Blu - Ray or even files UltraHD (4 k).