Moved

Yeah. I have finally moved to the new place that we bought last October. Even though the move has not yet fully completed (need to file thousands of change address form and move the junks from the garage), but Vivian and I have stayed in the new place for a few nights now. Many of our belongings are still in boxes scattered across the floor, and doing certain things is still not yet convenient, but I imagine it would take a while to get everything sorted out.

In response to my previous blog entry, yes the server box is moved, but it has been a bigger drama than I originally planned. I did not started at 10:00 PM but instead 11:30 PM, as I had to move many other boxes on that day before I could be back working on the server. Taking it off line and cleaning up the dust was the easy bit, but setting it up at the new place was a bit more challenging.

You see. Having your <everything> server at home is different from having it at a co-loc centre. You cannot just dump it on the shelf, power it up, and expect just leave it like that. Since it is at home, it needs to be part of the furniture. Or rather, not to be part of the furnitures. Unfortunately I don't own a house with a dedicated air-conditioned server room that I can neatly place the box on a rack. Instead I was trying to carefully hide that mid-tower behind the desk, and neatly tidy up the wiring so they don't make a mess. After half an hour of fiddling around, everything sits together nicely behind my desk, but at the expense that it would be impossible to take out if the box needs some service, which turns out that I needed it much sooner than expected.

Migration is not that all smooth, and as previously predicted, the hard drive does not feel happy after the move. Here is the warning message that I got from the kernel log every few minutes:

Feb 22 18:16:44 yoda kernel: hda: dma_intr: status=0x51 { DriveReady SeekComplete Error }
Feb 22 18:16:44 yoda kernel: hda: dma_intr: error=0x84 { DriveStatusError BadCRC }
Feb 22 18:16:49 yoda kernel: hda: dma_intr: status=0x51 { DriveReady SeekComplete Error }
Feb 22 18:16:49 yoda kernel: hda: dma_intr: error=0x84 { DriveStatusError BadCRC }
Feb 22 18:16:49 yoda kernel: hda: dma_intr: status=0x51 { DriveReady SeekComplete Error }
Feb 22 18:16:49 yoda kernel: hda: dma_intr: error=0x84 { DriveStatusError BadCRC }
Feb 22 18:17:04 yoda kernel: hda: dma_intr: status=0x51 { DriveReady SeekComplete Error }
Feb 22 18:17:04 yoda kernel: hda: dma_intr: error=0x84 { DriveStatusError BadCRC }

Not looking good! On Sunday morning I actually got a hard crash when ReiserFS cannot write back the journal.

Feb 22 08:35:31 yoda kernel: journal-601, buffer write failed

/dev/hda is the 9 Gb Quantum Fireball that has served me well, and for the past 5 years it has been transported from box to box until it finds its home in my dual Celeron, and serves root and /var partitions. Sounds like the rubbish bin is going to be its next destination.

So last night I have to quickly pull down the server from the net, move all content to another drive on the box (a 60 Gb Seagate), power it down, take out the faulty drive, move the Seagate to /dev/hda, and bring it back on-line again.

Another problem is the swap - there is no swap partition on that drive! And I was too scared to use resize_reiserfs to resize and re-partition that one 60 Gb partition, when I saw the beta software warning message popped up when I tried to run it. In the end, I just set up a 400Mb file as loopback device and use it as swap partition. A dodgy hack done in a hurry I know, but that was the best solution I had at 1:30 AM this morning...

Anyway. Now the server is now moved, and everything seems to be working at this point of time. The new ADSL connection, provided by Swiftel, is much less stable than Internode's bridged connection, as PPPoE dropped out every now and then, and can sometimes take ages to re-connect. I'll talk more about Swiftel later, but I hope the glitches will only be temporary.