Everything is a Freaking DNS problem

Upgrading MySQL-Cluster

So MySQL Cluster is heading towards disk storage , time to upgrade my development environment in that direction.
After the upgrade I ran into the following problem

mysql> create table blah (i int, c char) engine=ndb;
ERROR 1005 (HY000): Can't create table './test/blah.frm' (errno: 4009)

And the mysql error database doesn't really help :)

perror --ndb 4009
OS error code 4009:  Cluster Failure: Unknown result: Unknown result error

In the ndbd log it showed that the MySQL nodes connect but never really continued to start, strange no more logs and no help on the error.
Took me a while to figure out that MySQL-Max wasn't upgraded . MySQL-Max used to be required for NDB (dunno if it still is) and all of MySQL packages were already at 5.1.6 except for the Max package which deamon was actually running and which was still at 4.1.12 :)

You obviously know what I'll be testing next :)

Fedora Upgrading.

Has it ever occured to anyone that each time you want to do an upgrade of a Fedora Core box. either online or via CD it always fails because your / or /usr/local or whatever filesystem isnt' big enough. Each time you run into the dreadfull "Need 128Mb more free space on /" actually when upgrading from CD it used to just
crash witouth this error. Dunno if that is alraedy fixed.. filed a bug ages ago on that iirc. Anyway apt-get dist-upgrade is running on my iMac now :)

O'Reilly just published a nice article on Building a High-Availability MySQL Cluster

Out of a couple alternatives such as fusedav and wdfs I opted for
WebDAV Linux File System (davfs2) ,
It builds fine on my Fedora Core 4.91 box but when trying to do mount a remote WebDav filesystem I initally was running into a

/sbin/mount.davfs2: CODA kernel version mismatch:
        compiled for version 2, version 3 available

However if you replace the include occurences of coda.h with the actual location of your coda.h file (in /usr/src/kernels/ and then build it again, the error is gone and you can mount webdav stuff.

Also interesting is the /usr/local/etc/davfs2/secrets file where you can put url , username and password combinations so that a mount can be done automatically.

MySQL Cluster gaining disk support rsn according to Mikael Ronstrom: 5.1.6 release crammed with goodies

This can't be true , no really can you imagine us using passwords such as dnet-X , taHdeR, llevoN, esuS , ecruoSneX as passwords. But a company that claims to be a security company uses Sudnilet!‘ ? I really hope that nobody is using eruc-C , erucsA or even nezibU (noisiVteN) as a password.

I suggest they l change their passwords to !Mocagleb

Update: I just got word thas some company really is using nemlod as password

Got a couple of mails from people that badly wanted the presentation I gave at the Novell Tour last week it's here.

Dan Walsh points out in danwalsh: Linux fragmentation - a view from the Security community that we are heading for fragmentation in Linux Security , that's really not a good thing. I can't comment on the technical App Armor side yet but I think he really has some valid points here.. lots of people have been using SELinux and learning it for ages now, there indeed was the need for making SELinux easier to manage however we didn't need a totally ne environment.

Apart from that you might remember my comments on Security and Easy configuration regarding to firewalling. I think it also applies to MAC. If you make stuff to easy to configure you will get Mr Joe Random Newbie who thinks he correctly configured something but actually opened up a whole lot more than he actually wanted to. A security tool really shouldn't be something that can be used by novice users easily, maybe they should realise that the topic is too diffficult for them and get some help from people who actually understand what they are doing.

On the other hand there is the Learning normal behaviour vs Learning Anomalies strategy,
which one do you prefer ?

I`m interrested to see how other distributions will catch on with App Armor, how soon will we see packages for Ubuntu / Fedora / Mandriva etc.. time will tell wether , and usually the best package survives, I`ll keep looking at both sides :)

On a side note, has anyone got statistics on the chances that a proprietary project gone open grows a real community vs an open source project from day 1 ?

The Fosdem 2006 schedule is online.

So this question pops up on irc on how to manage disks for Xen Virtual domains ..

When I answer LVM people always wonder why .. afterall they only have one disk I often ask the following questions : How many partitions can you create on 1 disk ?
When not using LVM can you resize your paritions when you need them ? Do you often run into the
"The kernel still uses the old table. The new table will be used at the next reboot." issue when trying to add partitions ?

With LVM you don't have these problems , and those are only 3 of the LVM advantages that have nothing to do with striping, mirroring or snapshotting
Actually you just create partitions (/var/ /usr/ swap / etc .. in LVM's ) no difference between physical or virtual machine :)

So the next question that popped up a couple of minutes later was , wouldn't that be more confusing and complex than managing just some partitions ?
Wouldn't the overview become more complex ?

Actually the overview will be VERY clear if you use the correct naming conviention :)
Say you don't use LVM .. after six months would you still recall if you used /dev/sda9 for / and /dev/sda10 for /usr and if you did that for vhost or vhost2 .. ?
However say you have a volume group "virtual_machines" and in that volume group you have a logical volume called vhost1.root vhost2.usr
which one would you prefer ?

Ari Jaaksi asks in his Blog: Where is Henry now?

Henry isn't called Henri anymore , he is now called Larry Page, , Sergey Brin or Jimmy Wales (only to name a few) , those are the people that turned Open Source tools into a easy to use platform that EVERYBODY is using Every Day.

PS. Tim O'Reilly already told us that in his Fosdem 2004 Keynote.