count(*) in hive and the wc of hadoop file is different.

[#1]

Value differs when I check the count(*) using hive query and when I check the no of lines in the corresponding file. Seelect count(*) from table gives 100 as output where as wc -l of the corresponding file gives 30. My table is in ORC format. Is it how orc behaves?

[#2 · Accepted solution]

@Bala Vignesh N V

Yes, the difference in count is expected. ORC converts the table data into groups of rows called stripes along with auxiliary information in a file footer, default size of stripe is 250 MB. Hence, there will be difference in wc -l on orc file compared to actual numbers of rows in the table.

[#3]

Thanks Sindhu. What happens when we store it as TEXTFILE. The count should remain the same right?

[#4]

Yes, it would be same.