Profile cover photo
Profile photo
Staffan Thomén (duck)
723 followers -
Eternal tinkerer
Eternal tinkerer

723 followers
About
Posts

Post has attachment

Post has attachment
The supermicro 2026T-6RFT+ arrived, I pulled the CPUs out of the DL380 and... I am out of thermal paste.

sad trombone
PhotoPhotoPhoto
11/10/17
3 Photos - View album

Well, it's been an exciting (in a bad way) couple of days.

I have, as my "main" home server was getting on in years (22ish) and was experiencing the occasional random crash, I wanted to get some thing more modern and picked up a fairly cheap HP DL160. This 1U server turned out couldn't handle the poor airflow in my closet so I got another cheap server a DL380 (both gen6).

So far so good, I set it (the DL380) up in my closet as a NetBSD/Xen server which I could partition off to handle the services I needed, and it seemed to work swimmingly.

Last Wednesday something happened with the P410i RAID controller, either one disk in the smaller of the two RAID sets in the system died or the controller went haywire and killed the drive. In any event it started to throwing errors to the console and finally corrupting data on the non-broken disks. I rebooted the system and now one disk was down but that's fine since it's a RAID5, it should work in degraded mode until I can get spare drive, which it did. At first.

After I'd rebooted it that first time, the dom0 started correctly, but the domU:s wouldn't boot because of file corruption, so I cleaned up the filesystems best I could using fsck, fsdb and mild violence (in one case I had to move the files and recreate a filesystem because fsck would say it was all right and then when a certain inode was accessed the VM would panic with dup alloc)

This seemed to sort things out, and I figured I could limp along until new drives arrived. Not so, says HP. On Friday afternoon it again started spewing errors to the console and this time it trashed data on the first drive and then indicated that it was faulty, which seems not to have been true.

I figured that this probably meant that it wasn't just the one drive and HP:s shitty firmware but the SAS controller, so I tried moving the P410 controller from the DL160 (which is just a PCIe 8x card) into the DL380 and that worked, the BIOS found and initialized the volumes and all was gravy. But the first disk was apparently still unbootable and the volume considered "failed".

So I figured no problem, I'll just wipe the volume and rebuild the Xen setup, since no user data was on the first volume and I had backups. This worked, I could boot off USB and install a system on the newly created volume (which included the broken disk #2, which in hindsight might not have been a good idea, although the system happily let me use it).

Then I tried to boot the system from disk, but the RAID controller now complains that disk #2 is dead and the RAID set is failed and hides it. I presume I could coax it to work, but I'm not going to. I've had it with this HP shit.

I have now bought a Supermicro 2026T chassis off ebay which is component-compatible with the RAM, disks and CPUs in the HP and set my mail forwarding VPS to hold mail in the queue for 30 days.

For sale: parts for HP DL160 gen6 and DL380 also gen6. May include deadly curse. Caveat emptor.

Post has attachment
Daily driver
Photo

Post has attachment
Photo

Post has attachment
I ebayed some more drives for the RAID in the HP, 340MB/s write and 424MB/s read... It'll do for spinning rust :-)
Photo

Post has attachment
Photo

Post has attachment
Photo

Post has attachment
Youtube's new algorithm loves mystery biscuits...
Animated Photo

Post has attachment
Replaced my creaking Time pedals with Crankbrothers Eggbeaters 1.

Word is that they've fixed the fragility issues that gave them a bad reputation back in the day. Let's find out if they beat the Time pedals (both of my pedal pairs started acting up after 1 year)

Bike cool factor++ :-)
Photo
Wait while more posts are being loaded