Thursday Sep 14, 2006

1 node specjappserver world record - using Containers



I couldn't pass this one up. One of my favorite bloggers puts it in perspective. Grab a Sun Fire x4600, assign a few cores here, a few cores there using containers and pbind and you end up with a world record. How cool is that?

OK, I'm 100% positive a lot more effort went into the benchmark, but you get the point :)


Wednesday Sep 06, 2006

SDN Channel - new developer-centric group blog



Oooohh. Check out the new Sun Developer Network Channel (SDN Channel). Sweet. If you're a developer and want to keep up with how Sun can make your life a more productive one, check it out. Note, this augments the existing Sun Developer Network - the place to go for developing software on/for/using Sun products. Oh yeah, and don't forget about the developer forums. I've found that to be an invaluable resource.

Sigh, the SDN Channel puts my blog to graphical shame. Me-thinks I'll have to start sprucing it up. Talk about raising the bar! Geez.


Friday Aug 18, 2006

Make Xeon faster than Itanium2 using SPARC



Hmm, an interesting technology.

Note this particular quote:

"Weigel asserted that QuickTransit performs well. Sparc-Solaris software generally runs faster on Xeon-Linux using Transitive's software than on machines with Sun's 1.5GHz UltraSparc IV+ chips."

Thanks to the Benchmarking Guru, we have some data showing that UltraSPARC IV+  outperforms Itanium2 again and again.

Therefore, using the transitive property in concert with Transitive's QuickTransit and similar logic, HP and Intel customers can "generally" replace their HP-UX Itanium2 servers with Linux Xeons running SPARC binaries.

TGIF :)


Thursday Aug 17, 2006

Open Source: There can be only one?



I suppose I can pile on here, here, and here. I can understand one vendor taking pot shots at another. It happens all the time. Being in a glass house, I'm not throwing stones. OK, perhaps just one. Besides, we've upgraded.

The most surprising comment came from Mr. Handy:

 "There's room for a proprietary one and an open one. Once one is open, you don't need any more.

To use IDEs as an analogy, NetBeans was around long before Eclipse. IBM didn't feel NetBeans should be the only IDE. They launched a competitive IDE. It gained adoption due to NetBeans shortcomings. To compete, NetBeans improved dramatically, to the point where many think NetBeans is simply a better IDE.

Given the IDE competition, how did the developer (customer) \*not\* win?

Since an open source "monopoly" does not apply to IDEs, why should it apply to open source operating systems? Why is more than one open source operating system, especially standards-based operating systems, not a good thing for customers?

Perhaps "There can be only one" should be left to entertainment. I think a developer community of 13,000+ and those running the 5,000,000+  hosts benefiting from that community would tend to agree.


Wednesday Jun 14, 2006

Happy B'day OpenSolaris



OpenSolaris is one year old. It's amazing to see how far OpenSolaris has come in so little time. Multiple distributions. A growing number of contributors. Many communities. Growing Interest. And more.

Any guess as to what the 2nd birthday will bring?

Wednesday May 31, 2006

SavaJe Update

Looks like the problem with deploying my "Hello World" app to the phone was not the SavaJe platform itself, but a bug in the Mobility Pack 5.5 for CDC. Thanks to Lukas for the comment he left on the last entry on the topic. Taking his advice to install update 1 of the mobility pack addressed the issue. The next step is to do something a bit more rich than "Hello World" label with an exit button. Anyone have some spare time they can lend me? :)

I'm getting great support from the SavaJe support folks on my phone's speakerphone problem. They are really interested in getting to the root cause of the issue. It's not solved yet, but a patch is in the works. I look forward to when I can use it as my primary phone to show off my featureless applications :)

Saturday May 27, 2006

Early retirement thoughts



I'm not planning to retire yet. I've got a ways to go. However, when I do, I know think I know where.           


Wednesday May 24, 2006

zlogin is not an ssh substitute

I'm workin' on a highly available web application configuration on my laptop. So far I have 5 zones on my 2GB RAM laptop:

  • Administration zone: Application Server Admin server and a directory server for LDAP authentication
  • Application Server zone(s): 2 Zones, with application servers running, along with HADB
  • Web Server Zone: A web server running, load balancing across 2 application server zones.
Yes, I could run all of those services in one zone, but hey, what would the Clingan Zone be without lots of zones?

I had been starting most services automatically via rc scripts and inittab, per product instructions. I could run them via SMF if I write my own (or google for them), but I haven't gotten around to it yet.

Everything had been working fine. I had installed and run the web server successfully all via zlogin and by launching the RC script. Wait a second, why was I starting the web server by hand? I selected "launch at startup" on the installer screen. Hmm, my web server rc script wasn't being called, because it existed. So I ran zlogin, as I usually do, to get to my zone of choice. Turns out zlogin can be a bad idea as my pseudo-default login method. Here's why.

svcs -x showed that sysidtool was running and that a boatload of services were waiting on it. Huh? Lemme think. After creating the zone, didn't I log in with "zlogin -C" and finish configuring the zone? Obviously not. What was interesting is that I've gone on for a while running the Java Enterprise System Web server in a zone that hadn't even been configured.

Note, the interface was plumbed and up by the global zone upon zone boot, and I had copied a host file to all zones, so host names were resolving.

You may ask why I would take the trouble to "su root" to even be able to run zlogin? Why not just ssh in the first place? Well, since this was run on my laptop, I don't want to autoboot zones. I boot them when I need them.

That means root. That means zlogin. Until now.

Tuesday Mar 07, 2006

What does Matisse4MyEclipse mean for SWT (retry)?

I've been thinking about this a bit and now I'm on to thinking out loud. There is no doubt that Matisse is a pretty hot GUI builder for cross-platform applications. There is no doubt that Sun has marketed the hell out of Matisse (myself included) because it is a darn good piece of functionality.

I'm a NetBeans user, pretty far from SWT and close to Swing, so I definitely have my biases. I know some pros and cons to SWT. That's about it. I haven't written a line of SWT code. My cards are on the table. Now to the point.

What does Mattise4MyEclipse mean for SWT? Will it have no effect to SWT adoption because the design center of SWT (native components) appeal to some developers? Will this negatively impact SWT adoption? Will Matisse4MyEcplise drive more desktop adoption of Java and grow the pie (and SWT with it)? I must say that I am not really seeing much discussion of the topic. That speaks volumes in and of itself. Perhaps it's just too early. Except for James who, as usual, is off to an early start.

Hats off to Genuitec for the port.

Update: A problem with Writely's posting of this entry requires me to delete and repost the entry. Could be user error :) Thanks. The Management.

What does Matisse4MyEclipse mean for SWT?

I've been thinking about this a bit and now I'm on to thinking out loud. There is no doubt that Matisse is a pretty hot GUI builder for cross-platform applications. There is no doubt that Sun has marketed the hell out of Matisse (myself included) because it is a darn good piece of functionality.

I'm a NetBeans user, pretty far from SWT and close to Swing, so I definitely have my biases. I know some pros and cons to SWT. That's about it. I haven't written a line of SWT code. My cards are on the table. Now to the point.

What does Mattise4MyEclipse mean for SWT? Will it have no effect to SWT adoption because the design center of SWT (native components) appeal to some developers? Will this negatively impact SWT adoption? Will Matisse4MyEcplise drive more desktop adoption of Java and grow the pie (and SWT with it)? I must say that I am not really seeing much discussion of the topic. That speaks volumes in and of itself. Perhaps it's just too early. Except for James who, as usual, is off to an early start.

Hats off to Genuitec for the port.

Update: A problem with Writely's posting of this entry requires me to delete and repost the entry. Could be user error :) Thanks. The Management.

What does Matisse4MyEclipse mean for SWT?

I've been thinking about this a bit and now I'm on to thinking out loud. There is no doubt that Matisse is a pretty hot GUI builder for cross-platform applications. There is no doubt that Sun has marketed the hell out of Matisse (myself included) because it is a darn good piece of functionality.

I'm a NetBeans user, pretty far from SWT and close to Swing, so I definitely have my biases. I know some pros and cons to SWT. That's about it. I haven't written a line of SWT code. My cards are on the table. Now to the point.

What does Mattise4MyEclipse mean for SWT? Will it have no effect to SWT adoption because the design center of SWT (native components) appeal to some developers? Will this negatively impact SWT adoption? Will Matisse4MyEcplise drive more desktop adoption of Java and grow the pie (and SWT with it)? I must say that I am not really seeing much discussion of the topic. That speaks volumes in and of itself. Perhaps it's just too early. Except for James who, as usual, is off to an early start.

Hats off to Genuitec for the port.

Update: A problem with Writely's posting of this entry requires me to delete and repost the entry. Could be user error :) Thanks. The Management.

What does Matisse4MyEclipse mean for SWT?

I've been thinking about this a bit and now I'm on to thinking out loud. There is no doubt that Matisse is a pretty hot GUI builder for cross-platform applications. There is no doubt that Sun has marketed the hell out of Matisse (myself included) because it is a darn good piece of functionality.

I'm a NetBeans user, pretty far from SWT and close to Swing, so I definitely have my biases. I know some pros and cons to SWT. That's about it. I haven't written a line of SWT code. My cards are on the table. Now to the point.

What does Mattise4MyEclipse mean for SWT? Will it have no effect to SWT adoption because the design center of SWT (native components) appeal to some developers? Will this negatively impact SWT adoption? Will Matisse4MyEcplise drive more desktop adoption of Java and grow the pie (and SWT with it)? I must say that I am not really seeing much discussion of the topic. That speaks volumes in and of itself. Perhaps it's just too early. Except for James who, as usual, is off to an early start.

Hats off to Genuitec for the port.

Update: A problem with Writely's posting of this entry requires me to delete and repost the entry. Could be user error :) Thanks. The Management.

Tuesday Dec 13, 2005

The Solaris 10 University Challenge

Dan "The Man" Price has just informed us that in his recent travels and discussions with conference attendees, he has uncovered that many of our customers (and prospective customers) do not know about the Solaris 10 University Challenge. There are some rather large prizes involved, like $5000US and an Ultra 20 workstation per individual on the winning team, and $100,000US worth of goods for the University those individuals attend.

The worst thing about this challenge is that I am not eligible. What's up with that? It used to be that working at Sun meant I got some serious perks like lots of free stuff. Free hardware (to use, not keep). Free software. Now, our customers get the software for the same price I do, and they get the opportunity to get free gear and make money off of it. As a non-Sun employee, you get an opportunity, and as a Sun employee I get action items. As a consolation prize, I got a "Duke For President" bumper sticker from Mary. That was a "who you know" thing. Yeah, like your envious :) Oh, wait a second .... I do get the paycheck thing. Never mind. I don't mind sharing our free goodies.

Oh crap, I'm off topic again. Anyway, take a look at the challenge and get yourself, and your University, some major gear. Check out Adam's and Dan's blog entries as well.

Thursday Dec 08, 2005

Create a zone using ZFS in 1 second (or less)

Following up on yesterday's blog entry, I decided to let my curiosity get the best of me. I created a script to create a zone using the steps I outlined in that blog. Using the "date" command, I timed it to 1 second (or so).

Below is the script output and the script, which was done in a hurry and does a lousy job of error checking (use at your own risk). It can be a framework from something more formal, though.

Here is the output from my test:


root@sicilian /var/tmp> ./createzfszone.bash
\*\*\*\*\* WARNING \*\*\*\*\* WARNING \*\*\*\*\* WARNING \*\*\*\*\* WARNING \*\*\*\*\*
The zone configured by this script is not supported
Because this script is a hack, it may not work in future solaris versions

The script requires pre-requisite steps to be taken:
1) Create a zone on a zfs mountpoint using normal zone creation steps
2) Create a snapshot of the zone \*before\* booting it

Example:
# zonecfg -z myzone -f myzone.template
# zoneadm -z myzone install
# zfs snapshot  @initial
#
## where  is the zfs dataset where the zone is installed
Running zfs list ...
NAME                   USED  AVAIL  REFER  MOUNTPOINT
usbpool               4.09G  32.9G     8K  /usbpool
usbpool/jesclone      25.6M  32.9G  3.55G  /zones2
usbpool/zones         4.06G  32.9G  2.84G  /zones
usbpool/zones@initial   323M      -  3.54G  -
usbpool/zones@PostAppserver  6.81M      -  3.74G  -

Enter zfs basline dataset
usbpool/zones@initial

Enter zfs dataset mountpoint
/zonestest

Enter zone directory (should be under the ZFS mountpoint)
/zonestest/jes

Enter network interface for zone [e.g. hme0]
e1000g0

Enter an IP address for the zone
192.168.200.5

Enter dataset name for zfs clone
usbpool/zonestest

Enter the name of the new zone
zonestest
Thu Dec  8 09:05:44 PST 2005
Thu Dec  8 09:05:45 PST 2005
root@sicilian /var/tmp> zoneadm -z zonestest boot
zoneadm: zone 'zonestest': WARNING: e1000g0:1: no matching subnet found in netmasks(4) for 192.168.200.5; using default of 255.255.255.0.
root@sicilian /var/tmp>


Here is the script:

#!/bin/bash

TEMPLATE=/tmp/zone.template
TMPINDEX=/tmp/index


################################
# Print user warning
################################

user_warning() {
echo "\*\*\*\*\* WARNING \*\*\*\*\* WARNING \*\*\*\*\* WARNING \*\*\*\*\* WARNING \*\*\*\*\*"
echo "The zone configured by this script is not supported"
echo "Because this script is a hack, it may not work in future solaris versions"echo
echo "The script requires pre-requisite steps to be taken:"
echo "1) Create a zone on a zfs mountpoint using normal zone creation steps"
echo "2) Create a snapshot of the zone \*before\* booting it"
echo
echo "Example:"
echo "# zonecfg -z myzone -f myzone.template"
echo "# zoneadm -z myzone install"
echo "# zfs snapshot ${zone_dataset} ${zone_dataset}@initial"
echo "#"
echo "## where ${zone_dataset} is the zfs dataset where the zone is installed"
}

################################
# Select zfs dataset to snapshot
################################

get_user_options() {
echo "Running zfs list ..."

ret_val=1

while [ ${ret_val} -ne 0 ];
do
   zfs list
   echo
   echo
   echo Enter zfs basline dataset
   read DATASET

   zfs list | grep "${DATASET}" > /dev/null 2>&1
   ret_val=$?

   if [ ${ret_val} -ne 0 ];
   then
      echo Invalid dataset name
   fi

done

echo
echo "Enter zfs dataset mountpoint"
read ZFS_MOUNT

if [ ! -d ${ZFS_MOUNT} ];
then
   echo "Directory does not exist. Please create the directory"
   echo "or select a new directory"

   exit 1
fi

echo
echo "Enter zone directory (should be under the ZFS mountpoint)"
read ZONE_PATH

echo
echo "Enter network interface for zone [e.g. hme0] "
read INTERFACE_NAME

interfaces=`/usr/sbin/ifconfig -a`
echo $interfaces | grep ${INTERFACE_NAME} > /dev/null 2>&1
ret_val=$?

if [ ${ret_val} -ne 0 ];
then
    echo "${INTERFACE_NAME} is an invalid interface"
    exit 1
fi

echo
echo "Enter an IP address for the zone"
read IP_ADDRESS

echo
echo "Enter dataset name for zfs clone"
read CLONE_NAME

echo
echo "Enter the name of the new zone"
read ZONE_NAME

return
}

###################################
# Create the zone template
# (input to zonecfg)
###################################

create_zone_template() {
#
# Create the zone template
#
cat > ${TEMPLATE} << EOF
create -b
set zonepath=${ZONE_PATH}
set autoboot=false

add net
set physical=${INTERFACE_NAME}
set address=${IP_ADDRESS}
end

verify
commit
EOF
}

#########################################
# Get options from the user
#########################################

user_warning
get_user_options

date

#########################################
# Create the zone template
#########################################

create_zone_template

zonecfg -z ${ZONE_NAME} -f ${TEMPLATE}

zfs clone ${DATASET} ${CLONE_NAME}
zfs set mountpoint=${ZFS_MOUNT} ${CLONE_NAME}
cp /etc/zones/index ${TMPINDEX}
cat ${TMPINDEX} | sed "s/configured/installed/" > /etc/zones/index

rm ${TEMPLATE}

date
The End.

Wednesday Nov 23, 2005

Getting to Postgres through the Java Integration Suite

To be more in line with Sun's database direction, I am updating part of a demo to utilize Postgres instead of MySQL.

The conversion of the database was seamless, I simply re-ran my "create table" and "insert" scripts to re-populate the database with dummy data. Sorry, nothing of substance to share with you on the database conversion side. I'm a Postgres newbie, and it sure felt like it. RTFM. That's what I did. Nope, no substance to that thought either.

On the Java Integration Suite (JIS) side of the equation, the conversion was seamless. No code modified. I just updated about 9 properties (driver name, connect string, etc) across 2 property sheets. That was it. OK, I fat-fingered the driver name and it took me a bit of time to debug. After I figured that out, it was pretty darn smooth. This was all done on my laptop (the test system).

I am only partially done converting the real demo. Postgres is installed with the data imported. I still have to update the JDBC properties on the test system. Once the live upgrade to Nevada build 27a is done on my laptop, I'll boot the petri dish. eDesigner, the JIS development environment, requires Windows. I've told every ex-SeeBeyond employee about this, and I am sure the message has been heard. They just have quite a bit on their plate, so I'll patiently wait. I bet there is a way to update the properties from the command line, but I have yet to read that FM.

FYI, I downloaded Postgres from Blastwave. Supposedly the Sun Update Center either has Postgres or will have it soon. I don't use the Sun Update Center simply because I run Nevada. If anyone knows if the Sun Update Center works with Nevada, let me know. I haven't tried yet.

About

John Clingan

Search

Archives
« July 2014
SunMonTueWedThuFriSat
  
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
  
       
Today