Browse oracle-l: This Month's Archive • Main Archive Page • Related posts • Previous by Date • Next by Date

Re: 10046/10079 Tracing understanding

From: Tanel Põder <tanel.poder.003@xxxxxxx>
To: "ORACLE-L" <oracle-l@xxxxxxxxxxxxx>
Date: Sat, 6 Aug 2005 00:47:48 +0100

Hi,

One reason for seeing several tiny waits for SQL*Net more data to client events 
followed by one long one and then tiny ones again, might be the OS TCP buffer 
filling up - Oracle records only the time to copy results to OS network buffer 
in SQL*Net toward client events. So as long as the buffer has enough space, we 
complete the send operations in few microseconds, but when the buffer fills up, 
our send is blocked until some packets have been sent to target and ACKs 
received, which normally means wait times in at least milliseconds or even more.

Tanel.

  I can't comment on why the empty packet to the client (unless its an ACK or 
similar), but in your 10046 trace you have a number of very quick more data to 
client waits (like 10-20 milliseconds) and then one rather slow one. I'd be 
intrigued as to what data was in the 'slow' packet. It may be (though I don't 
think it should be a more data wait, but rather a message from wait) an 
indication that indeed the client is doing something slow with that particular 
packet. 

    Any help understanding this would be appreciated. 

    Thanks - Brian

    10046 only:

    WAIT #1: nam='SQL*Net message to client' ela= 11 p1=675562835 p2=1 p3=0


  snip


    WAIT #1: nam='SQL*Net more data to client' ela= 15 p1=675562835 p2=2048 p3=0
    WAIT #1: nam='SQL*Net more data to client' ela= 251155 p1=675562835 p2=2048 
p3=0
    WAIT #1: nam='SQL*Net more data to client' ela= 15 p1=675562835 p2=2048 
p3=0 

    ...


    10046 & 10079 together

    ...

    WAIT #1: nam='SQL*Net more data to client' ela= 14 p1=675562835 p2=2179 p3=0
    SEND: (bp=0xb74209d8, bl=255, isdblink=0)
     (overflow) 1537 bytes sent via fast-path 
    Dump of memory from 0x0ABAD2DB to 0x0ABAD8DC
    ABAD2D0                   FF202020 20202020          [   .    ]
    ABAD2E0 20202020 20202020 20202020 20202020  [                ]
            Repeat 14 times
    ABAD3D0 20202020 20202020 FF202020 20202020  [           .    ] 
    ABAD3E0 20202020 20202020 20202020 20202020  [                ]
            Repeat 14 times
    ABAD4D0 20202020 20202020 FF202020 20202020  [           .    ]
    ABAD4E0 20202020 20202020 20202020 20202020  [                ] 
            Repeat 14 times
    ABAD5D0 20202020 20202020 FF202020 20202020  [           .    ]
    ABAD5E0 20202020 20202020 20202020 20202020  [                ]
            Repeat 14 times
    ABAD6D0 20202020 20202020 FF202020 20202020  [           .    ] 
    ABAD6E0 20202020 20202020 20202020 20202020  [                ]
            Repeat 14 times
    ABAD7D0 20202020 20202020 FF202020 20202020  [           .    ]
    ABAD7E0 20202020 20202020 20202020 20202020  [                ] 
            Repeat 14 times
    ABAD8D0 20202020 20202020 FF202020           [           .]
    Dump of memory from 0xB74209D8 to 0xB7420AD7
    B74209D0                   20202020 20202020          [        ]
    B74209E0 20202020 20202020 20202020 20202020  [                ] 
            Repeat 14 times
    B7420AD0 20202020 20202020                    [        ]
    WAIT #1: nam='SQL*Net more data to client' ela= 13 p1=675562835 p2=1792 p3=0
    SEND: (bp=0xb74211d0, bl=255, isdblink=0)
     (overflow) 1793 bytes sent via fast-path 
    Dump of memory from 0x0ABAD20A to 0x0ABAD90B
    ABAD200                   20FF2020 20202020          [  .     ]
    ABAD210 20202020 20202020 20202020 20202020  [                ]
            Repeat 14 times
    ...

    __________________________________________________
    Do You Yahoo!?
    Tired of spam? Yahoo! Mail has the best spam protection around 
    http://mail.yahoo.com 




  -- 
  Niall Litchfield
  Oracle DBA
  http://www.niall.litchfield.dial.pipex.com

References:

10046/10079 Tracing understanding
From: Brian Wisniewski

Re: 10046/10079 Tracing understanding
From: Niall Litchfield

Re: 10046/10079 Tracing understanding

Other related posts: