Announcement

Announcement Module
Collapse
No announcement yet.

Critical: May be a bug in pt-table-checksum need little help in analyzing the data

Page Title Module
Move Remove Collapse
X
Conversation Detail Module
Collapse
  • Filter
  • Time
  • Show
Clear All
new posts

  • Critical: May be a bug in pt-table-checksum need little help in analyzing the data

    Am experimenting with pt-table-checksum for checking integrity on prod servers am getting ambiguous results,

    this is from one of the tables checksum result on stdout
    TS ERRORS DIFFS ROWS CHUNKS SKIPPED TIME TABLE

    05-30T17:08:11 0 1 0 1 0 0.013 xxx.xxx_test

    ---------------------------------------------------------------------------------------------------------------------------------------------
    the result in the percona.checksums table says there is no diff for table on master as well on the slave, but there is difference in the number of rows of data in the table.

    +-----+----------+-------+------------+-------------+----------------+----------------+----------+----------+------------+------------+---------------------+
    | db | tbl | chunk | chunk_time | chunk_index | lower_boundary | upper_boundary | this_crc | this_cnt | master_crc | master_cnt | ts |
    +-----+----------+-------+------------+-------------+----------------+----------------+----------+----------+------------+------------+---------------------+
    | xxx | xxx_test | 1 | 0.001505 | NULL | NULL | NULL | 0 | 0 | 0 | 0 | 2014-05-30 17:08:11 |
    +-----+----------+-------+------------+-------------+----------------+----------------+----------+----------+------------+------------+---------------------+

    I tried different combinations and all the time the result showed as if there is no difference in the number of rows
    master_cnt this_cnt actual master_cnt actual this_cnt
    0 0 0 2
    2 2 2 3
    3 3 3 2

    something like that how ever i try the result in the table for both master_cnt and this_cnt remained the same value as that of master_cnt, how ever the row count in the slave is. But in the STDOUT result set the Diff to '1' indicating the data is different in that table.

    Need help urgently, some one help.

    Thanks
    DL


  • #2
    The outputs in a better way:
    STDOUT
    TS :05-30T19:32:02
    ERRORS :0
    DIFFS :1
    ROWS :11
    CHUNKS :1
    SKIPPED :0
    TIME :0.012
    TABLE :xxx.xxx_test

    checksums table:

    db : xxx
    tbl : xxx_test
    chunk : 1
    chunk_time : 0.001468
    chunk_index : NULL
    lower_boundary : NULL
    upper_boundary : NULL
    this_crc : ee8960c
    this_cnt : 11(actual row count is 5)
    master_crc : ee8960c
    master_cnt : 11
    ts : 2014-05-30 19:32:02

    Rows on master are : 11 and Rows on Slave : 5 still if you have any queries on my Question please feel free to ping me..
    Thanks

    Comment


    • #3
      Hi, are you looking at the percona.checksum table on the master? On the master, the this_crc and master_crc would always agree with each other since they are the same thing, the same for this_cnt and master_cnt.

      Please check percona.checksum on the slave server to see the difference between the slave and the master.

      Here is the documentation that explains how the tool works: http://www.percona.com/doc/percona-t...-checksum.html

      Comment

      Working...
      X