Petri Wessman's weblog

System Failure Imminent (have a nice day)

Yesterday Janka noted that the server “had been making funny noises”, and lo and behold, my email queue held a few notices about RAID1 disk failure. Seems that the older disk of the pair (a Hitachi 160G) had crashed hard, all partitions unusable. Still, thanks to the RAID, the system is up and running (in degraded single-disk mode). Put in an order for a new (larger) disk at Verkkokauppa, should be here in a few days – I’m betting that the single newer disk won’t also crash in the meantime. Did do some extra backups, though, just to be paranoid.

So… seems that the software RAID is working as intended and the email notifications are also working. Things could be a lot worse. Having suffered one big server crash in the past and lost quite a bit of data due to the backups also being corrupted, I’m now firmly in the group which refuses to run a server without a RAID disk (and reliable backups).

Lots of things I want to improve, once/if I get some extra cash. I want a separate NAS box to handle backups and other mass storage for our home network (D-Link DNS-323 is my current candidate), and I also want to upgrade the whole server at some point – the current machine is a low-end desktop box with a 1.8GHz single-core Sempron and only 2 gigs of memory. It works, but a multicore machine would make a lot of server operations snappier and more memory would naturally also help. The current machine only has 2 memory slots (low-end desktop, remember) and they are DDR400. I’m not sure if 2G DDR400 modules even exist, and if they do they are almost certainly very expensive – so might as well upgrade the whole machine at some point. No real hurry on that, though.

Published on by Orava, tags , , ,

More specific Maxtor drive specs

Some more info from DataPrey, it seems that the actual “must match” data of the drive I’m looking for is the following:

Model: 33073H3 
Code: YAH814Y0 
PCB code: C9DLA

The other codes are not that important. Pity, since I managed to track down one eBayer with drives that matched everything except that “N,M,B,E” code. Back to the hunt, but it’s looking pretty iffy I have to say.

Published on by Orava, tags , , ,

No luck with data recovery, but...

DataPrey threw in the towel with the disk recovery, saying that they were unable to fix it, sorry.

However, since they had earlier told me that the problem revolved around getting a spare matching disk, I called them and asked for details. Well, turns out that yes, the problem was their inability to find an exact match for the disk, and they had closed the case because they were unable to resolve it in the maximum promised time.

As for the problem, turns out that at the time this disk was made, Maxtor was in the habit of changing the internals of the drives all the time, even within one model. So in addition to needing a Maxtor model 33073H3, it needs to be a very specific instance of that model. (3-4 other things also need to match, in addition to the model number).

Anyway, DataPrey has a long-term “seeking component X” list, so we put this drive there and reopened the case for the time being. They will continue to scan for matches for this drive, and they’ll send me the exact drive details needed so I can hunt for a match on my own, too. The world is a big place, and you never know, I might get lucky on eBay (DataPrey polls eBay too, now and then :).

So for now it looks like the data won’t be coming back. But there’s a small glimmer of hope, still.

As I was writing this I got email from DataPrey with the info, so I’ll jot that down here too. The drive needs to match to:


and it would be nice if it would also match to:


Those codes can be found on the label that is attached to the disk. If anyone has an old Maxtor (30,7 GB) drive that matches to those specs, I’d be very interested in it. I already sent queries to the few matches that I found on eBay.

Published on by Orava, tags , , ,

Disk recovery update

Got a situation update from DataPrey on the disk recovery attempt, with a nice technical explanation of the problem. Turns out the “service zone” of the disk had developed a physical fault, and even though the fault is tiny in size, since the disk “boots” internally from the service zone, any fault in it means that the disk flat-out refuses to work and just bangs the reading heads around. It’s sort of like the BIOS in a PC getting zapped.

This explains the sudden, no-warning crash of the disk – the fault was miniscule, it was just in the worst possible place.

The normal procedure is to hunt down a disk that is exactly the same model as the broken one, and copy the microcode from there. Problem is, this is an old 30G Maxtor disk that wasn’t too common even back then, let alone now, and finding a “donor” disk is proving to be troublesome. They have some leads (4 of them), but they need to find an absolutely exact copy. If they do, recovery should be possible. If they don’t, it will be extremely hard or flat-out impossible.

So we’re still waiting, we should know the status by next week.

Whatever happens, I’m impressed with the professionalism in the feedback I get from DataPrey, these guys at least sound like they know what they are doing.

Added: on the very slim hope that someone might have one lying around, the disk in question is a 30,7 gig (yes, tiny) Maxtor ATA drive, model 33073H3. It’s about 6 years old, and was originally sold with the “Maxtor Diamond Max” label.

Published on by Orava, tags , ,

Drive recovery proceeding

Just got a message from DataPrey, saying that they are hunting down the spare parts for the recovery procedure – they have to find the exact same ones as the disk uses, and since it’s an old disk it could take some days. We’ll see how it goes.

Published on by Orava, tags , ,

New weblog

It… lives!

The new server is chugging along nicely after the post-crash recovery, so I felt it was time for a new blog – I hadn’t updated the old one for about half a year for various reasons. This is now running on Typo 4.0.0, a very nice blog platform that’s written with Ruby On Rails. I’m still playing around with this, and using one of the default themes, but so far I really like what I see. Seems nicer than Wordpress, and Wordpress is not bad at all. Ajax is used to good effect, and the admin interface has some really spiffy Ajax-based drag+drop stuff in it.

I had a new theme layout done for a new blog, but that went with the crashed disk. I have an older backup copy, but I’m waiting to see if we get the crashed disk recovered or not before I proceed on that. I’m hoping to hear from DataPrey this week.

Published on by Orava, tags , , , , , ,

Powered by Publify – Thème Frédéric de Villamil | Photo Glenn