This has become a war and peace effort, please bear with my ramblings, or read other peoples. This post is an all or nothing one with lots of juicy bits in it. It begins around mid May 2003 and ends in late August 2003. I've split it up into roughly month long sections.

May:

Lets see, there was several evenings that involved consumption of large quaitities of alcohol, several very long days at work, a long weekend that I worked, plenty of wet weather, and lots happening at home. Directly followed by me starting to find a new place to live.

Around this time I had a play with Redhat 9, it's nice and no apps in it crashed on me, vastly better than Redhat 8 on previous attempts. I used the SGI XFS install CD to install it too so I got XFS filesystem support (but I'm not using it for root for recovery reasons). So after a play with building a degraded raid 5 array and hot adding the last disk back into the set and other fun stuff (using loop devices), which lasted a few days I proceeded to start rebuilding the kernel rpm's from SGI to include the International Kernel patch (the CryptoApi patch) so I could do loopback encrypted file systems. I didn't have much success with this since Redhat's (and hence SGI's) kernel already have some of the CryptoAPI stuff in there, but not all, so not all the patches applied cleanly and to cut a long story short I couldn't get it working. Actually a virgin 2.4.20 kernel with the international patch (and the loop-jari patch) didn't work with loopback crypto in Redhat 9 either. I wasn't really interested in fishing around the util-linux package to find out why so I read more and found out that not only does Mandrake have XFS, it also may have the loop-aes Crypto Loop support in by default depending on your version. (XFS since 8.0, unknown start of loop-aes). 1950 Meg and 4 hours later I was able to try out Mandrake 9.1 (higher version number than Redhat so it must be better, but I was sleeping so it had to wait until the following evening.

My file server migrated from Windaz 2k server to Linux.

I would have used Gentoo except I don't want the box to be hard to setup again if the need be, not that Gentoo is hard to setup, it's just time consuming, and Gentoo 1.4 final still isn't out at this time. It was going to be a complex setup anyway without further complications there. It was going to end up with a 5 disk raid 5 array, encrypted using XFS. But with Mandrake funnyness, XFS oddballness and general Promise controller dodgyness I ended up using Redhat (9) without encryption. In 6-10 months time I'll probably revisit the encryption idea when it's a bit more stable/documented. [followup below]

I've never really been a Mandrake fan, for no reason really, just didn't get on too well with it on each attempt. But I've heard [reiserfs](http://swfreenet.gateway.net.auSteve Eaton say nothing but good things about it and he uses it on his servers (or did last time he told me about it, well he said on all but one which was Redhat still). I tried it out (9.1), and it's nice, it's polished, but it's not quite as there as Redhat 9 in terms of desktop usage to the off the street person but it's still quite respectible. And for me, it has XFS and it has loop-aes. Me happy again. But I didn't end up using loop-aes or XFS. Since Redhat has xMule, a crossplatform port of eMule, it's nice, I ported my eMulePlus mod to it, and bothered to make a waix only hack in it cause then I can keep running edonkey on my fileserver now it has moved to linux. But once I loaded the waix ip:subnet dump from a real waix only router (don't ask) with 2600 + ip subnets in it, the client intermittently hangs. I ported it back to emule plus on 22 June and so far it is fine. Dunno what to think about that. [followup below]

This post gets vague and unconnected from here on, consider yourself warned of the dangers and rapids ahead.

The weekend just gone (22 June being the sunday) was the ISS trip to rotto. Much fun was had, we even played golf. I felt really good this morning (monday 23) but as soon as I got to work things were back to normal.

Moving soon. ADSL has been cancelled so they don't overcharge me again.

Looks like one of the disks in my raid 5 wasn't up to the duty necessary, or the promise shit controller is more shit that I thought. It's one thing then another, originally I had issues because of a dodgy ide cable, now this. I'm not pleased, fortuately the drive is under 6 weeks old and no data is lost. It's just a pain in the ass to have to rip it out and exchange it for another, also because the reported geometry of the disk is different to the rest (thanks to WD for that one) it is annoying. [followup below]

I wonder some times if it's worth going on or not. My subconscious is on auto pilot most of the time anyway. My Administrator password is "" (without the quotes, it's more secure than most would think). (But of course my Administrator account has no rights =) After all I'm not totally stupid.)

If there's jelly beans behind the rasberrys I'd buy the marshmellows. --Me without enough caffiene talking about the vending machine at work.

July:

The weekend and another few days have passed, I'm not really certain where I'm going to move. Perhaps back near to where I lived before. It's like I've lost the last six months of my life, I can now resume goals that were put on hold when I moved last time. I'm not moving in six months time, thats for sure.

Swiftel have 512/512 (A)DSL for $125 per month, not bad, but no free traffic, though excess is by far the cheapest around.

I got a googlewhack a few weeks back and then discovered it's not really that hard once you try.

Lan on this weekend, oddly enough I'm moving following the lan.

Windows 2000 SP4 is out. I'll gladly apply that as soon as I get rid of Evil spawn of satan from my pc, but I might just bypass it and do not pass go, do not collect $200 and go straight back to Linux.

Parking meters in West Perth aren't cheap; 10c == 4 minutes (or $1.50 per hour).[followup below]

I've moved. 4 days later my GeForce 4 TI card died, I don't think they were related, it just stopped dead, pc wouldn't even post with the card in there. 3 weeks later returned it under warranty (I was 9 months old)

I lanned (walan), then again (exclusive leech lan) and again (Reload 7). I bought an LCD Monitor at Reload, it's sweet.

Reload was ok, I'm being a bit passe about it because I'm exhausted or something. I'm looking forward to a weekend of not going out, it will let me sort out things at home that need dealing with. Also I may have time to edit a little video together for Walan, or to repair my dodgy disk in my raid array.

Moved office's. Now I'm in sardine-ville in Nedlands, but more people started each week I was there, 4 or 5 in 4 weeks.

I lan therefore I am.

August:

Multiple weekends pass.

• Car broke down; RAC is good (well their road side assistance is good, just don't change your address on their website cause they then post your insurance renewal out to the old address ! (I had no insurance at all for 2 months because of this))
• Fixed Raid array
• Fixed Laptop
• Fixed Wireless AP (thanks for the trouble Mr K)
• Burned 50gig to make space to edit the Walan video that I've been meaning to do since following the previous lan (actually the one before, but that's a minor detail)
• Upgrading my PC soon
• Snapgear routers are good, Billion ones are not, thus I'm getting a snapgear one soon

Things are getting under control at home now. I still need to arrange the warranty RMA's for the 3 IBM drives that failed. Rehosting of domain coming soon, once that happens there will (again hopefully) be more frequent updates. I watched End of E again, the DVD has commentary. Wootay.

Upgraded my desktop boxen, Shuttle SB61G2, P4 2.8, 1gig PC3500, Sony DRU 510a. Combined with the LCD and my l33t USB keyboard's it looks rather sexy. Now all I need is my replacement video card because the GF2 Pro doesn't quite cut the cheese in the Midnight Madness 2/pixel fill rate front. As transplanting the whole mainboard didn't agree with WinXP (safe mode BSOD'd on boot) I had to reinstall. (Backup and reinstall). I put Redhat 9, and XP on (XP ONLY because 2000 doesn't detect the hyperthreading CPU properly (according to MSDN)). I ended up disabling Hyperthreading because it made my sound skip in linux. Alsa driver, no OSS driver exists for the Realtek AC97 on this board. So I'll probably put 2000 on shortly.

Had a drive really fail in my raid array. Currently it goes clunk on power up and doesn't work. Removed it, so the array was running in degraded mode for a little while again. Got a replacement drive within 3 days (bought one (on Saturday), RMA's take ages) and when trying to rebuild the array the box was locking at 33% complete. Turned out to be bad sectors on the disk that's been giving me trouble all along. (mumble)

I was on the verge of losing 440gig of data. 1 really failed disk, and 1 disk with an unknown number of bad sectors, after 3 attempts to rebuild the array (using different controllers and ATA cables (40 wire one to force ATA33 in the end)) the disk got kicked from the array and the array went offline. Essentially I had just seen 2 disks fail in a raid 5 array. Then a blast of freedom hit me, I've been slaving to the data whore for too long. A quick search on the net in to how to fix this generally said dude you're so screwed, hope your backups are up to date. Honestly who can afford to backup 450gig of data at home these days, it's expensive enough getting that much space let along keeping backups of it. So I tried to force the (bad sector) disk back into the array (the other disk was of no use since it had only completed a 33% rebuild) and run in degraded mode. To do this I had to forcefully (really force, you'll understand if you ever try this) the recreation of the array (mkraid) over the existing one, hoping the data isn't lost. I guess some eternal good will force was shining on me that night because it didn't hose the disk, nor the volume on it. Actually as long as the array was intact the volume was fairly safe, since I was only doing this rebuild etc with the volume unmounted. I now had access to my data, but as soon as I hit a bad sector the box would lock. I had one new 120gig drive to use to put data on and the rest of my herd of boxen.

I proceeded to prune the arrays contents back as much as possible. Absolutely anything I could download easily (waix, ftp etc) got blown away. The I started copying the data off. Bottom line, the array was hosed and had to be rebuilt. But I still needed another drive to put in the array (since now I had 2 I couldn't use). Sadly this was a Saturday night / Sunday morning so it had to wait until Monday. I ended up haggling a bit at the shop on monday and got 2 120's for a good price (better than saturdays price), I guess the guy at the shop was feeling a bit sympathetic/wanted to leech at the next lan, so he helped me. Now I had enough space to back up the remainder of the array so I could rebuild it. Utilising the little spare time I have (so few hours in the evenings) I backed up 350 gig (or so) and then built the new array (same size) using 4 disks in degraded mode, and 2 other disks with data on them.

While backuping up the data I experimented with Crypto-Loop again. My problems before were related to my attempt to use XFS which for some reason I couldn't figure out didn't work at all via Crypto Loop, yet Ext3 and ReiserFS did. Since the file server wasn't being reinstalled I searched for Crypto loop Redhat info. I was sure part of the crypto loop system was present since the time I tried to put it into the SGI XFS Rpm a few months earlier. While browsing redhat's bugzilla I found this bug: [

Losetup and util-linux are not crypto aware.](http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=56698) Reading this pointed me to the answer to my problem. Util-linux in Redhat 9 was not crypto aware. Grabbing their Src rpm and building that yielded me a working set of losetup/mount/util-linux which worked with loopback AES encryption. Taking this to the next level on my new 440 gig array and it worked (with reiserfs).

It was simple once being setup, a simple mount /dev/md0 -o encryption=aes /woot and I was prompted for key size and password. Once modifying modules.conf to make the right crypto modules load on boot it was nice. I benchmarked it and was disappointed. Copying data off a local disk (WD1200JB) onto the degraded, encrypted, reiserfs array I was able to burst to 12meg a second and got up to 17meg a second reading, sustained write was more like 10meg a second. My 1200mhz celeron just didn't have enough power in it. Without encryption (still degraded, still reiserfs) I was able to write 25meg a second and read over 30meg a second. As I'm contemplating putting a gigabit card in the box this just wasn't fast enough, and I'm not upgrading it again (the file server started out as a Dual P3 550, 1gig ram, and is currently a 1200mhz celeron (100mhz fsb) 512meg ram). As before I will revisit it in 6-10 months again.

I copied all the data off the 2 data disks, then hooked up the next set of disks (the other disks lying around my desk, a nearly dead 60gig IBM deathstar, a 13gig quantum, the 40gig external one etc) and all the data off my desktop pc (which has quite a bit of free space since I had only recently reinstalled). Once this was complete, I was able to raidhotadd the first disk into the array, and it worked smoothly, the following morning I raidhotadd'd the other disk, so I now have an online spare disk in my array =) Now all I have to do is put the disks back into the case, I now have 3 newer disks and 3 older disks, so they are fairly well matched (better than before, 4 old, 1 new), the spare disk is a new one.

A few days later the PC is all back together with covers on and running nicely. PCBudget is out of my bad books for the time being, since he replaced the one fawlty disk on the spot with a brand new one (not a refurbished one). Jim says the WD JB (or SE) series drives aren't replaced with reconditioned drives, but with new ones. Another good reason to get the Special Edition JB 8Meg cache drives with the 3 year warranty over the 2 meg cache 1 year warranty BB ones. Also I got a new GF4 for the warranty of my old one. This one runs just as hot though and seems to stutter in 3d graphics. Haven't eliminated bad AGP drivers yet though.

Moved office again. Back in West Perth wasteland, but still as busy as ever, when will it let up. I found more expensive parking too, $4 per hour, maximum daily charge $40 ! (I'm parking there, next to a Rolls) but unlike the owner of the Rolls (or would that be driver) I'm not paying since the boom gates on the lot don't work, I'm l33t or what. As they said in a not so famous flick, Remember, hacking is more than just a crime. It's a survival trait. Speaking of which there's been a lot of worms goind round Windows these last few weeks.

Me to the team: The only thing that keeps us going is the countdown to the next public holiday.