‘t has been a while since I last blogged, not a huge amount going on really, just getting on with life.
I’ve decided to pickup Horizon again, I’ve pretty much decided on functional PHP and MySQL again. While I’d like to try J2EE/Struts for it, I also want to be able to use it in the majority of places, like my parents home machine and my “router” neon.
I’ve been playing around with lighttpd a fair amount after Andy proded me to a few weeks back, I have to say I’m very impressed with it. I’m not a big fan of FastCGI spawning three copies of PHP without any users, but I suppose it could be worse. I’m tempted to set it up on nebula for ajb007.co.uk and provide a second alias to try it through.
nebula is approaching a point where is despiratly needs a reinstall, however I’m waiting until FreeBSD 6.2 which should hopefully have Xen dom0 support. Then it’s a trip down to telehouse for a quick Xen install, and create a cell which has the current nebula in it, then to create the proper installation from afar. Xen will provide me the ability to reinstall entire cells with very little down time.
xenon has also been having a few issues since it’s MB swap, the Adaptec 2810SA RAID card was being lost contact with during heavy use, disabling SMP seems to have restored the stablitiy, which probably means there’s a race condition somewhere that needs to be found.
I’ve been having issues trying to make neon pass Bonjour traffic between xenon’s vlan and the inner lan vlan. The problem is Bonjours address is in 224.0.0.0/24 which isn’t ment to be routed between subnets… “great”. However I want to use daapd for sharing to iTunes. I’m thinking I’ll need to write a relay program that does simple listening and relaying, and of course IGMP… :/ This is also an issue to stream music to the Airport Express from iTunes, although not an issue using DVD Jon’s streaming program.
Work is good, had a few issues with ResNet over the past few weeks, Cisco’s VMPS protocol is a PIA. We were having an issue where the VMPS server appeared to be loosing data, so after a long rewrite the problem still persisted, the end issue was that the VMPS and *nix kernel was buffering that much for us that the requests became out of date.
It appears that it sends 3-5 requests for the same MAC auth to ensure that one gets through, and thoose request are only valid for 10 seconds, before it does it again with a different ID. Our issue was that this was all to much for the PostgreSQL database to deal with in a timely fashion.
There are three things I can do to VMPSd to solve this, we can check upon reciept of VMPSd packet to see if we already have that ID, upon popping from the queue if the request is over 10 seconds old and upon processing, if it’s just a reconfirmation just bounce back the VLAN provided. This will put off the problem for a while, but there isn’t a huge amount that can be done in the long run, this method is maintainable for 6,000 ResNet users, but I’m not sure how well it will scale campus wide. Saying that, I’m not convinced of the database design the external script is using.
Ahhh well, I shall get on that this week if I have time and make it cleaner again.
Feck all else to talk about really,
D.