Database Reference
In-Depth Information
[initandlisten] allocator: system
[initandlisten] options: { dbpath: "./db" }
[initandlisten] journal dir=./db/journal
[initandlisten] recover : no journal files present, no
recovery needed
[FileAllocator] allocating new datafile ./db/local.ns,
filling with
zeroes...
[FileAllocator] creating directory ./db/_tmp
[FileAllocator] done allocating datafile ./db/
local.ns, size: 16MB, took
[FileAllocator] allocating new datafile ./db/local.0,
filling with
[FileAllocator] done allocating datafile ./db/local.0,
size: 64MB, took
0.327 secs
[initandlisten] command local.$cmd command: { create:
"startup_log",
size: 10485760, capped: true }
ntoreturn:1 keyUpdates:0 reslen:37 470ms
[websvr] admin web console waiting for connections on
port 28017
[initandlisten] waiting for connections on port 27017
Most of the startup information for the server is informational, but there
are some items of note. First, the warning about soft rlimits is important.
Most operating systems (Linux included) have defaults that assume
something much closer to a “consumer” system than a “server” system,
especially a database server. As a result, many of the limits on memory
and open files are set unusably low for production environments. Even
distributions that bill themselves as “server” distributions suffer from this
problem.Fordevelopmentitisprobablynotnecessarytoraisethislimit,but
a production server should set this limit fairly high because a large database
could have thousands of resources in use at any given moment.
The other item to note is that MongoDB pre-allocates some files when it
starts for the first time. These files are simply filled with zeros because
they will be memory mapped by MongoDB and filled with data. If the
size of the data on disk is roughly known, then these empty files can be
Search WWH ::




Custom Search