Some days it’d be nice to be a cockroach, life would be simpler that’s for sure. So in a previous post I mentioned the trials of getting an ARC-1280ML working with 16+ drives in JBOD. We’ve decided to go with a pair of the same controller, less ports the ARC-1231ML. Hopefully they’ll fix the limitation at some point. So we’re waiting on that new hardware. Keep in mind there’s nothing really *wrong* with the ARC-1280ML, it’s just not going to do what we need it to do.
Just plugged in a couple new (yes I know discontinued) Intel SBXD132‘s. Naturally, BIOS and Embedded Management (SMP) is code is out of date, 1.06 is current, 1.04 is installed, and I’m not entirely sure about the SMP version numbers. Flash the BIOS uneventfully, then, go to flash the SMP. SMP nearly completes, taking it’s sweet time, then get a mysterious error message in the WebUI along with
20 ERR SERVPROC 02/14/08 16:34:54 (WMN315709449) Flash of Blade System Management Processor (slot 5) from 192.
As our new SysAdmin is fond of saying, No Whammies! The aforementioned afflicted Supermicro PDSMI+ SYS-5015M-T+B chassis is now in one piece slogging through burn-in testing thanks to one VERY helpful comment.
I have to say thanks to mickrussom because he gave me the heads up/answer as to WHY our Supermicro PDSMI+ system wasn’t working. Turns out the Intel 3000 north bridge ONLY supports up to 1066MHZ FSB speeds, despite the specifications on Supermicro‘s site clearly claiming “Supports a Dual-Core Intel® Xeon® 3000 Series processor” — Their emphasis, not mine, without any caveats or mentions of “with FSB Speeds of up to 1066MHZ.
So. This is the story of trying to register dotblag.com.
I’ve used joker.com as my primary registrar for some years now. They’ve been reliable and responsive in the past when I’ve needed manual intervention/help. However. Their CC validation system lately has been, to say the least, persnickety. Wonk. Very wonk. To the point that I have to take my business elsewhere. While processing the registration for dotblag.com, the .com registration was denied but the .
This one is shorter than usual. Things have been busy but I wanted to muse on this, because, thankfully, I didn’t get bit by it.
Linux 2.6.17 to about 2.6.24.1 or so has a nice little bug in the vmsplit syscall, wherein it fails to validate parameters. Thus, one can get root. Well I set about upgrading some machines, and the other Systems Admin here set about upgrading most of the others.
Areca ARC-1280 (and ARC-1280ML’s, same product, different connector) have 24 SATA ports. But only support 16 drives in JBOD mode….or in any mode really, 16 drives exposed to the host is maximum. The controller deadlocks during POST/INIT if you have 17+ drives when it’s set to JBOD mode. So you can have 12 LUNs 2 drives each (mirrors) or some other combination of RAID, but you can’t do JBOD with 24 drives as of the most current (1.
The root name servers now officially have AAAA glue records installed. This is actually a pretty major step towards global IPv6 deployment. There was a few snafu/foul-ups, mostly regarding VeriSign not setting the connection limit high enough on ftp.rs.internic.net so causing a mirror failure.
What does this mean to you, as someone who’s on the internet? Not a damn thing probably since most ISPs and NSPs haven’t begun to widely deploy IPv6.
I haven’t posted about it before but we’ve an ongoing saga in attempting to get together a 1U server for a customer. Normally…this isn’t an issue. Our regular vendor (QSol) though has made some business decisions which means, atleast for the time being, for our lowest level/cheapest 1U machines/dedicated servers they’re a little outside of the price point we’re willing to look at. They did their best to get us something as we have a long standing business relationship with them, and so they sent along a machine.
Trend Sues Barracuda Over Gateway Anti-Spam Techniques
Wow. Just WOW. In a nutshell Trend is suing Barracuda. Why? Because they can. Because the USPTO gave them a BS patent, like they’ve given so many others. Wow. Just. Wow.
If Trend manages to win this it could open up all other AV software, gateway scanning software, and similar products to litigation.
Or, how NOT to “log” errors. Especially on web based stuff. ESPECIALLY.
Output from one of our mail processing frontends. The script runs mailq and collates the output, sorted by recipient domain. Redacted of course to protect the people with the bad idea.
mfe3:~# /opt/bin/top-mailq-to-domains 2947 *REDACTED*.com 158 *REDACTED*.co.uk 117 *REDACTED*.org 116 mailbox.com This is why sending yourself an email for every error on your site is A BAD IDEA. Make a log.
NFS. The very bane of my existence. Well maybe not quite, but it sure makes my life hell. You can’t narrow down problem/runaway areas because there’s no granularity to what tiny statistics gathering it can/does do. There’s no authentication, and little to no access control.
Please, don’t start on NFSv4. Linux can’t even support NFSv3 hardly, and the 2.6 kernel is worse than 2.4. Atleast 2.4 doesn’t completely lose track of locks and locking the way 2.
OK so I’ve long been quite opposed to the use of greylisting. Recently I had a pretty major personal mail server crash and that’s caused me to at least give it a try and seriously rethink that. So when I rebuilt the mail server for wgops.com (yes I know, it’s probably STILL blank) I went ahead and installed a piece of greylisting, and other misc. spam control policy software on the Postfix based mail system.
OK so I’ve often thought about keeping a blog purely for technical crap. So hang on, and lets see how far down the rabbit hole this one goes Alice. Bob is welcome for the ride too.