PostNuke

Flexible Content Management System

News

Could PostNuke be the Missing Link?

Because of this, I have found that if I create a soft link to a database file, MYSQL is perfectly happy with it and I end up with two pointers to the same inode (and thus, file). This has the same functional result as duplicating the file and keeping the two copies in sync, however, it is not neccesary to maintain two sets of data..

Want an example? If I would like to maintain one store of quotes in the database for my site named www.spot-on.tv, I would simply add my quotes into that database via the postnuke interface. When I am ready to add these quote to my other site (baybridge) I can do so via the creation of a symbolic link pointing links that use the same filename as the database files in baybridge database to point to elements (in this case, files) within the spot-on database.

The following was performed using:

RedHat 7.2
Kernel 2.4.7-10
mysql 3.23.41
filesystem ext3

Before you start messing around with the guts of mysql, make sure you have a good backup. The command shown below is what I use to back up my entire mysql structure. It is imperative you make a backup.

tar -cvjf /u0/backup/mysql.tbz /var/lib/mysql/*

Now, on with the show.

Inside my /var/lib/mysql directory are subdirectories which hold the actual data and indices for each database:

athena.ebzb.com[root]:/var/lib/mysql-> ls
Rogue_baybridge/ Rogue_ebzb/ Rogue_poorbastards_org/ Rogue_spot_on/
athena.ebzb.com[root]:/var/lib/mysql->

Inside each of these directories are three files for each table. Below, the files which comprise the quotes table are shown.

athena.ebzb.com[root]:/var/lib/mysql/Rogue_spot_on-> ls -l *quo*
-rw-r----- 1 mysql mysql 8606 Feb 13 01:15 nuke_spot_on_quotes.frm
-rw-r----- 1 mysql mysql 17436 Feb 13 01:15 nuke_spot_on_quotes.MYD
-rw-r----- 1 mysql mysql 2048 Feb 13 01:15 nuke_spot_on_quotes.MYI

Now, I look into the baybridge database (relative referencing)

athena.ebzb.com[root]:/var/lib/mysql/Rogue_spot_on-> ls -l ../Rogue_baybridge/*quo* ../Rogue_spot_on/Rogue_spot_on/nuke_spot_on_quotes.frm
-rw-rw---- 1 mysql mysql 8606 Feb 5 03:37 nuke_baybridge_quotes.frm.orig
lrwxrwxrwx 1 root root 54 Feb 13 18:00 nuke_baybridge_quotes.MYD -> ../Rogue_spot_on/Rogue_spot_on/nuke_spot_on_quotes.MYD
-rw-rw---- 1 mysql mysql 0 Feb 5 03:37 nuke_baybridge_quotes.MYD.orig
lrwxrwxrwx 1 root root 54 Feb 13 18:02 nuke_baybridge_quotes.MYI -> ../Rogue_spot_on/Rogue_spot_on/nuke_spot_on_quotes.MYI
-rw-rw---- 1 mysql mysql 1024 Feb 5 03:37 nuke_baybridge_quotes.MYI.orig

You will also note the links have retained the ownership of root, who created them-- but mysql user really needs to have the ownership.


# chown mysql:mysql *quo*

Now-- question is... is this supported? I seriously doubt it. If you want to play in this minefield, be very, very careful and make a complete backup before you touch a thing. The following command takes care of this by backup up all of my mysql databases and stores them on a different filesystem in /u0/backup.

tar -cvjf /u0/backup/mysql.tbz /var/lib/mysql/*

I have this in a cron job to run every day at 4AM, but you must run this immediatly before touching anything.

To review, steps are:

1) Shutdown database
service mysql stop
2) perform backup
tar -cvjf /u0/backup/mysql.tbz /var/lib/mysql/*
3) perform symbolic link procedure
4) Restart database
service mysql start
5) test, test, test.


I have posted this article on my www.poor-bastards.org site.

USE AT YOUR OWN RISK! NIETHER THE POSTNUKE FOLKS NOR I WILL ACCEP RESPONSIBILITY FOR TINKERING GONE BAD. IT WORKS FINE FOR ME, SO THIS ARTICLE IS MERELY AN OBSERVATION OF MY WORK. WHEN YOU PLAY WITH MYSQL COMPONENTS AT THE COMMAND LINE, YOU ARE PLAYING WITH FIRE.

COULD I POSSIBLY MAKE IT ANY CLEARER... STILL SOMEONE WILL GET THIER PANTIES BUNCH OVER A PLAY SESSION GONE BAD.


Hosting For PostNuke

Footnote: 1

searching and postnuke

For example, the search should cover the personals module, the downloads, the articles, the faq, the wiki module, sections module, astrology module, chinese astrology module, and whatever other modules you may be running...

It should truly spider your own site, write it to a database and be searchable by keyword...

Also, it should allow for weighted results and or paid results to where you have a little more control over results...

even google takes cash for results nowadays...


SIDE NOTES:

I was in a major car wreck jan 15th and am still in a wheelchair... Leg broken in several places, foot broken, knee rebuilt, etc...

My development has been slowed down in favor of recovery... But I'm still here...





Proposal for module identification

Multiple implementations of roughly the same functionality and identical modulenames of roughly different functionality cannot be prevented as the PN community grows. It' s not a bad thing either. However , a few problems will occur in the (not so) long run if we don't think about it.

Wouldn't it be nice if every PN module would be assigned some sort of unique identification which can be used for several (some more useful than others) purposes:


  • is a particular module installed? (module interoperability checks)

  • module name independent permission components

  • with the help of a central repository of identified modules all sorts of information can be distributed about that module (updates, security risks, monitoring etc.)



I'm sure more usefull examples are not hard to imagine.

To start of a design discussion this is an attempt at a short description of the concept I have in mind. I will use a typical scenario in the new situation.

--begin of utopia--
I need some PN module functionality foobar and a quick check on the pn-module-repository gives me the impression that no validated foobar functionality is available. After some forum posting here and there I decide to develop the foobar functionality myself. Let's begin ...
....little timewarp...
Ok, The foobar functionality is complete. I return to the repository and apply for an identification for this module after describing what it does, who I am and some other usefull information. I get a temporary identification for the foobar module and I'm asked to submit my module so it can enter the validation process which checks the minimal requirements automatically so foobar will run on a specified PN version. Everything checks out fine and the identification is made final.

The identification assigned to foobar can now be used by other modules to identify my module. For example to check whether it is installed and if not, connecting to the repository and present information on where to get it.

After some new development on foobar I update the information in the repository. All users which have indicated this are sent an email with a notification that the module has changed.
-- end of utiopia--

The assigning of unique identification (numbers) to modules can be fully automated if desired to burden only machines with the administration of the sytem. Site managers will get somewhat more freedom in their module naming as the only fixed item in the module will be the unique identification on which all other information can be based. The numbering system can be as simple as a sequence-number or somewhat more friendly. As long as it is guaranteed to be unique and it allows for module identity checks (to prevent wannabee modules to masquerade as a certain module) the system will work out fine.

Disappointed with some module authors

[rant]
Each time this individual offered the enhanced version to the postnuke community, mysterious anonymous post would appear, being very critial of this authors work. Obviously the anonymous post was the original author.

I am very dissapointed with the original authors having seen that. This goes against the very spirit of open source.

If you are not emotionally prepared to have someone make changes to, and re-release your module, don't release the code in the first place.

Especially, when you as the original author release your module under the gpl license. Any changes that are made to the module by individuals, must be available for download.
[/rant]

I am currently working on a couple of projects that are changes to some existing modules. ie I want my website members (none yet site not up) to be able to have a personal calendar, and address book. Several different modules are available, but I haven't found any that work for that scenario. When I'm done, my intention is to release the changes to the postnuke community.

After seeing the grief given to another developer, I seriously question whether I will release my changes as widely as I would have wanted.

Sorry for the long rant.

-Bill Cavalieri
bill@cavalieri.net

Url to a message?

I was not happy with the way the link to "parent" messages worked and I have modified a line in modules/NS-Comments/index.php as follows (the only line containing the word "_PARENT":

echo " | "._PARENT."";

Which seems to work wonderfully when I move up the hierarky: if you click on "parent" it displays the parent messages and the children thread. The problem is when I get to the first child of a message that is the direct reply to an article. If I click on the link to his "parent" I would like to get the full article and the thread of all responses, but I only get (perhaps correctly) the messages and the thread and not the article story. What is most annoying is that this page does not have a style but just a white background. Any idea why?
First Page Previous Page Page 119 / 277 (1181 - 1190 of 2763 Total) Next Page Last Page