accumulo-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Mike Miller <mmil...@apache.org>
Subject Re: accumulo monitor issue
Date Thu, 03 Oct 2019 21:56:11 GMT
Some tips from Ed:
The key is what shows up in the metadata - if you look at the System Metadata tables section
in the Accumulo user’s manual, the files entries have a form of row_id, filename, file_size,
row_count.

The monitor is reflecting what the the metadata table is showing.  In the case of a bulk import,
there will not be a row count estimate, so it will not show in the monitor.

If you entered a row via the shell and then did a flush (or a flush occurred for any reason)
that will create a file entry that includes the row estimate, soi it will show up in the monitor.

If you perform a full compaction (compact -w -t  <TABLENAME>), when it finishes, all
of the rows will have been processed and written to “new” files, and those file entries
will contain the row count values and should show up in the monitor.

Overall, the value shown in the monitor is a estimate and may or may not reflect the number
of rows available in a table via a scan (things like TTL, bulk imports, values written to
memory but not flushed to a file)… Right after a compaction, the value should be close,
but again, just an estimate.

On 2019/10/03 17:17:52, marathiboy <Sligade@fbi.gov> wrote: 
> Thanks Mike,
> 
> 
>  If my bulk inserted data with visibility labels is not showing up (but a
> ne=
>  wly inserted row from the shell does show up on monitor). All the data
> show=
>  s up correctly under hdfs  /accumulo/tablesxxxxx rfiles & rfile-info
> --dump=
>   shows everything nicely.
> 
>  This is happening consistently, so trying understand how it works and how
> c=
>  an I diagnose this?
> 
> 
>  Thanks in advance
> 
> 
> 
> --
> Sent from: http://apache-accumulo.1065345.n5.nabble.com/Users-f2.html
> 

Mime
View raw message