Fault: Sun Java Enterprise System Calendar chews an entire CPU core and returns an HTTP 401 to all requests.

Diagnostic criteria: An httpd process will be hammering identical logs so fast they’ll be rotating before you can read them. Look for something like:

[23/Jul/2008:21:40:41 +1000] HOSTNAME cshttpd[PID]: Account Notice: Closed session 6kNPKqo+K9I opened by USERNAME from [IP]

Root cause: The session database is cactus. Murder it.

Fix: Stop all the calendar components with stop-cal and clear out the contents of /var/opt/sun/comms/calendar/SUNWics5/data/http/session - they’ll be re-created when you run start-cal to bring it all back up.

Sun released their new Convergence integrated Mail, Calendar and IM web application yesterday - and it’s pretty damn impressive. There is a little work to getting it rolling if you’re not upgrading to the latest JES Mail & Calendar at the same time, but it works reasonably well and is possibly the finest webmail/cal out there - certainly a lot nicer than Outlook Web Access.

More information from Sun (with screenshot) here.

There’s been a lot of banging on about Sun’s future and direction moving forward of late. This tosser manages to join the group of completely uninformed plonkers throwing their tuppence in.

Here’s my take on it. This is pretty obvious, and based solely on what my customers ask for, demand, want and hope for.

• Sun should stick to selling quality hardware.
• Sun should keep Solaris going strong, because Solaris is a strong driver for hardware sales.
• Sun should keep selling service and support to go with Solaris and hardware.
• Sun should keep the hell away from complicated software offerings, because the market just isn’t biting.

Simple. The market wants to buy Sun hardware and run Solaris on it. Even if they’re a Linux shop, most places will at least trial Solaris with their new hardware. Every customer I have seen trial Solaris on Sun hardware (that’s x86, too, X4150s, for example) keep using Solaris. Even when they bought the hardware to run Linux in the first place.

What I’m *NOT* seeing is good traction on Sun’s software portfolio - the trials aren’t turning into sales that way Sun would hope. That’s not to say Sun should ditch their software portfolio, because much of it is very good. It’s to say that Sun should hold on software company acquisitions (MySQL was silly) and focus on delivering what the customers are happy to pay for. Build nicer hardware and the customers will buy it. Make Solaris better and even more people will use it. Not because it’s Open Source or anything nebulous like that, but because Solaris Just Works. Even Oracle still recommend large SPARC/Solaris deployments for critical databases, and they have a public Linux agenda.

Solaris/SPARC is seen as The Standard in the enterprise space. That does a huge amount to keep selling hardware. Citrix is seen as The Standard in it’s market, which is why selling SSGD is so hard. Sun need to keep reinforcing the core message and the core product and the core customer demand.

As everyone probably knows by now, a new DNS cache poisoning vulnerability has been announced recently. Just a reminder that you should probably be getting your resolvers updated.

Information is available:

• http://www.isc.org/index.pl?/sw/bind/index.php
• http://www.doxpara.com/ (includes a little test thingy)
• http://www.kb.cert.org/vuls/id/800113

Packages with an updated BIND release (9.5.0-P1) available for Solaris here:

• SPARC
• i386

Sun have patched their included BIND.

• SPARC - 119783-06
• i386 - 119784-06

Thanks to David Purdue for looking up the patch numbers.

This probably works with PINE as well.

You’ll want to edit .pinerc in your home directory.

Edit the smtp-server entry to point to your outbound SMTP server: smtp-server=your.smtp.relay.address

Edit the inbox-path entry to something like: inbox-path={your.mail.server.address/your_imap_account_username}INBOX

Edit the rsh-open-timeout entry and set it to zero: rsh-open-timeout=0

I’ve built packages for the replacement for PINE from UW. Alpine has removed some source constraints (important to some people, I guess), has some new features and otherwise seems to be an incremental upgrade on PINE. These packages were built on Solaris 10, so they should be fine on OpenSolaris / Nevada (Well, SXCE/SXDE) as well.

Alpine for SPARC (PCOWalpine)

Alpine for x86 (PCOWalpine)

Please let me know if there are any problems with these builds. Click here for more information about Alpine.

UPDATE: Packages rev’d to include -passfile by popular demand. Grab ‘em again. Read the README file.

Wah! Acro won’t print by default on Secure Global Desktop.

Never fear.

Change the print command to “lp -c -d tta_printer”

Fixed!

I don’t think I ever want to see a D1000 again. I have SCSI errors burnt into my retinas.

Slides from tonight’s MSOSUG presentation on Solaris 9 Containers.

Get it here: http://www2.purplecow.org/static/msosug-sol9cweb1.pdf

When using Ximian (Well, Ximian/Sun by this point) Evolution on Solaris to connect to Exchange, most things are pretty simple.

However, I found the “Global Address List” functionality wasn’t working for me. Seemed wierd.

The solution: Make sure your use a DOMAIN\USERNAME login, not just USERNAME. Everything else work work with just the short name, but the code that runs the GAL type functionality seems to demand the whole shebang.

Read the rest of this entry »