Oracle Scratchpad

August 14, 2009

Tera-bug

Filed under: Infrastructure,Troubleshooting — Jonathan Lewis @ 12:26 pm BST Aug 14,2009

I came across an interesting Oracle error recently: “ORA-15099: disk {name} is larger than maximum size of 2097152 MBs”, which you can find on Metalink under bug 7243305.

The status of this bug is: “Closed, Not a Bug” – but the description points you to bug 6453944 (present in 10.2.0.3, fixed in 11.2) for which the only visible text is:

RELEASE NOTES: ORA-15196(ASM block corruption) occured, if larger than 2TB disk was part of a diskgroup
REDISCOVERY INFORMATION: If you added larger than 2TB size disk to a diskgroup and saw ORA-15196, then you hit this issue.
WORKAROUND: Not to add larger than 2 TB size disk to a diskgroup

The text for bug note 6453944.8 tells you that the patch (to stop you creating diskgroups with oversize LUNs) is in the 10.2.0.4 and 11.1.0.7 patch sets – but the patch only stops you adding new oversize LUNs, it doesn’t help if you’re already using them. So if you’re running a very large system on ASM, you might want to check the sizes of your ASM volumes and take some remedial action if you’ve already got LUNs larger than 2TB.

There are no clear indications of what can go wrong with ASM LUNs larger than 2TB – but one example is wrapping: when Oracle gets to the end of 2TB, it may wrap back to the start of the LUN and overwrite the ASM metadata. When you’re running that big a system and need to do a full ASM disk group restore that’s a serious problem.

 

Useful related notes:

Metalink Doc ID: 370921.1: ASM – Scalability and Limits
Metalink Doc ID: 736891.1: ORA-15196 WITH ASM DISKS LARGER THAN 2TB

7 Comments »

  1. Are they using a modified FAT?
    LOL! :)

    NOTE: Mine is only a joke!

    Comment by lascoltodelvenerdi — August 17, 2009 @ 12:30 pm BST Aug 17,2009 | Reply

  2. I love the way that blogs allow information to spread. Shortly after publishing this note I got an email pointing me in the direction of bug 5448714 on Metalink.

    REDISCOVERY INFORMATION:
    ORA-1455 is received when a backup of a datafile with size > 4Tb is attempted.

    WORKAROUND:
    None

    RELEASE NOTES:
    Rman cannot backup datafiles larger than 4 Tb

    I haven’t managed to reproduce this on my laptop yet – and you may never run into it if you’re using ASM (because you may expose yourself to ORA-15099 first) – but it’s nice to know that it’s fixed in 11.0

    Comment by Jonathan Lewis — August 18, 2009 @ 4:55 pm BST Aug 18,2009 | Reply

  3. I’ll just upgrade the oracle on my netbook to 11 Jonathan, so I don’t exceed the 4TB backup limit of RMAN….

    (being serious, I’ve seen this issue on an ASM system – If I remember correct, more and more sub-2TB discs were added to a volume supporting a tablespace holding the termnial partition of a table that the client ws failing to split…)

    Comment by mwidlake — August 19, 2009 @ 12:06 pm BST Aug 19,2009 | Reply

    • It’s surprising how often this “growing partition” happens. I’ve seen a few places which started with good intentions, but forgot to keep them going when a critical DBA left.

      The symptoms to watch out for: a table with five partitions of 80 million rows, and a one partition of 1.5 billion rows. (YMMV)

      Comment by Jonathan Lewis — August 20, 2009 @ 9:34 pm BST Aug 20,2009 | Reply

      • The “growing terminal partition” has also become a common pattern in my experience. It seems to be a case of someone sets up a sensible intention of terminal partitions being split either before they are used or within a short time of them being used (eg the terminal partition being split each month so that there is a month’s data in both the old and new partition) but someone moves on and the partition splits are ignored as “everthing is working”.

        Until it does not work.

        And then it gets tricky.

        btw, have you come across any real petabyte databases yet? With those, it is going to be very hard not to hit the limits that Oracle Corp are not that good at testing themselves….

        Martin W

        Comment by mwidlake — August 20, 2009 @ 11:31 pm BST Aug 20,2009 | Reply

  4. This harkens back to the good old days of Oracle 5 where if you created a datafile greater than 24m, it would let you, take all day, and then tell you it was not usable when it attempts to format the last few kb. And then there was the “feature” where you could only have 32 datafiles in an entire database….

    Comment by Michael Fontana — January 5, 2010 @ 9:15 pm BST Jan 5,2010 | Reply

  5. The Oracle Education materials imply that the limit is 2 x 10^12 bytes, which it is not.

    Comment by Brian Fitzgerald — September 29, 2012 @ 10:56 pm BST Sep 29,2012 | Reply


RSS feed for comments on this post. TrackBack URI

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

The Rubric Theme. Blog at WordPress.com.

Follow

Get every new post delivered to your Inbox.

Join 3,909 other followers