Fabian wrote:
> Update of /cvsroot/monetdb/sql/src/backends/monet5/merovingian
> In directory 23jxhf1.ch3.sourceforge.com:/tmp/cvs-serv13466
>
> Added Files:
> Tag: Aug2009
> ChangeLog.Aug2009
> Removed Files:
> Tag: Aug2009
> ChangeLog
> Log Message:
> switch to branch specific ChangeLog (please don't remove ChangeLog in HEAD)
I took care of the not removing part.
--
Sjoerd Mullender
I have created a new branch which is to be used for the upcoming Nov2009
release. This branch is called "Nov2009".
There are now two "stable" branches, the Aug2009 and Nov2009 branch.
The Aug2009 branch will be used for the Aug2009-SP2 release which is
scheduled for October. The Nov2009 branch will be used for the Nov2009
feature release in November (and the Nov2009-SP1 and Nov2009-SP2
releases in December and January).
The reason for making the Nov2009 branch now, and not waiting until
after the Aug2009-SP2 release is that this way there is more time to
stabilize the code for the release.
There are a number of new features that are planned for the November
release. Some of these features are not completed yet, and all of these
features have to be properly tested and fixed if they're broken. The
not-yet-completed features *must* be completed before the end of
October, or else they *will* be removed from the release.
New features for the November 2009 release are:
- new SQL import from CSV files
- new code for large joins
- monetdb/merovingian distribution
- database replication
- native PHP client
- others?
Features that will definitely not be part of the November 2009 release
include:
- Stratos's slicing and histogram stuff
- more?
These will be removed from the Nov2009 branch but of course kept in the
development trunk.
All developers *must* now decide where each change they want to make to
the code belongs.
If a bug occurs in the Aug2009 branch, the bug *must* be fixed in that
branch. If a bug does not occur in the Aug2009 branch, but it does
occur in the Nov2009 branch, it *must* be fixed in the Nov2009 branch
(this includes fixes to complete the above-mentioned features). All
other changes *must* go to the development trunk.
Propagation will be done from the Aug2009 branch to the Nov2009 branch,
and from the Nov2009 branch to the development trunk.
--
Sjoerd Mullender
This message only applies to the current development branch of MonetDB,
and is informational only, mainly addressed to others working with the
current development branch.
monetdb[1] goes abroad
After quite some hacking to move code back and forth between components,
we now got the first version of merovingian[2] and monetdb[1] that can
do remote control. I will give some small examples here so everyone
that is interested can get up to speed fast, and/or get an idea on the
possibilities.
Basically, all functionality that used to be implemented in monetdb, is
now implemented in merovingian. That means, for instance creation of a
database is performed by merovingian. A positive side-effect of this is
that the discovery aspect of databases is now real-time, e.g.:
% monetdb release dumptest
taken database out of maintenance mode: dumptest
2009-09-25 11:54:10 MSG control[8590]: (local): released database 'dumptest'
2009-09-25 11:54:10 MSG discovery[8590]: new database mapi:monetdb://pegasus.ins.cwi.nl:50000/dumptest (ttl=660s)
We can see that when database is released again using monetdb release,
it is immediately announced in the network, making it available for use.
Locking a database has the oposite effect, of course.
Next, the "(local)" string indicates that this control command was
received over the local UNIX socket, as was the case in previous
releases of merovingian. In addition to this local access, monetdb now
also can connect over a TCP socket, which needs not to be remote, e.g.:
% monetdb -hjarl.ins.cwi.nl -Pxxxxxx status
name state uptime health last crash
testdbatjenniesmachine stopped 100%, 4m -
2009-09-25 11:56:19 MSG control[11449]: pegasus.ins.cwi.nl:49652: served status list
Here we can see that the merovingian at jarl reports it served a status
list to pegasus, which is the host I performed the monetdb status at.
Note the -Pxxxxxxxx in there, which is the passphrase of the remote
merovingian. This passphrase is by default a random string, but can be
set to something else by editing the .merovingian_pass file in the
dbfarm and restarting merovingian.
The way to find other merovingians in the vincinity is by using monetdb
discover. Its output now includes how to reach each merovingian, if
remote control is enabled for them. Remote control over TCP can be
disabled in monetdb5.conf by unsetting mero_controlport. Control
entries can be recognised in discover's output by the trailing asterisk,
which one can also glob for:
% monetdb discover "*\*"
location
pegasus.ins.cwi.nl:50001*
volund.ins.cwi.nl:50001*
jarl.ins.cwi.nl:64001*
This command just returned all merovingians that have remote control
enabled and reports their hostname and port to connect to.
The remote control protocol is very simple, and I'm planning to make at
least a Java library available to perform these commands, such that
monetdb[1] no longer needs to be the only application to perform
commands with merovingian.
[1] http://homepages.cwi.nl/~fabian/MonetDB/Man%20Pages.html/monetdb.html
[2] http://homepages.cwi.nl/~fabian/MonetDB/Man%20Pages.html/merovingian.html
Hello
I have installed MonedDB Sql (32 bit) under Windows (xp), but cannot run it as embedded server.
In MonetDB documentation it is mentioned that embedded monetDb engine is available.
How can I get embedded server installation for windows?
Best regards
Andzej
_________________________________________________________________
Daugiau nei pranešimai – išbandykite likusias „Windows Live™“ dalis.
http://www.microsoft.com/windows/windowslive/
Hi all,
I use BenchmarkFactory and ODBC to test the monentdb performance. But it
seems the default option of monentdb is autocommit, so some test scripts
seems not work.
How should i change the autocommit option (not only use "set autocommit 0"
and just change it in the range of one transaction )?
Does anyone know the details(that'are tools, scripts, monetdb enviroment) of
TPC-H test in monentdb website?
Many thanks~
Regards
Yan
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512
It seems to be that there is a new mechanism that basically doesn't
allow mclient to connect directly anymore. Is this behavior intentional?
The workaround seems to be -umonetdb with monetdb as password.
Stefan
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.11 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
iEYEAREKAAYFAkqgSCkACgkQYH1+F2Rqwn3bGgCeIIj438dKmOp1lasT7sGbgdgE
Ow0An1BBzIVLAr7cuLRhsOcexCP287GY
=qRmk
-----END PGP SIGNATURE-----