Depending on the Instance your using, the client may hit HDFS first to get
some instance information. Then either instance will hit Zookeeper to get
the root tablet location. It will then scan that to find the appropriate
!METADATA tablet, but will aggressively cache everything. The client will
then scan (with heavy caching again) the !METADATA tablets necessary to
find the tablet(s) for the range of writes. And then the writes will occur
to appropriate tablets discovered in the !METADATA scan.
Your order of operations for the write seems mostly correct. I think it's
response from logger followed by insert into in memory map and then
response to client. Eric knows this better than anyone though, so I'm going
to defer in case I'm inaccurate.
The scripts start loggers on all of the slave nodes. You can arbitrarily
start/stop loggers, but we highly recommend doing 1:1 between them and
tservers because the method for assigning tserver->logger does make
assumptions depending on that 1:1 correlation.
On Wed, Oct 10, 2012 at 3:47 PM, Hider, Sandy <[EMAIL PROTECTED]>wrote:
> Hi all,
> I want to verify some of the details with Accumulo Writes and Loggers.
> How does the client find the correct tablet server to write to? Does the
> client query the master once and receive the ranges associated with each
> tablet server? Is this list in zookeeper?
> With write-ahead loggers, when a client makes a write request, to a tablet
> The tablet server writes the data to a write-ahead logger which is
> stored on a different slave
> Then acknowledges the client
> Then writes adds data to its in memory table.
> Do I have this order correct?
> Is the number of loggers always fixed, can this be changed with server
> Thanks in advance,