FreeBSD 7.0 with SSD Crashes

[map7]

I'm running FreeBSD 7.0 on brand new hardware. The hardware has been going fine with a standard hard drive.

I'm trying to use the new OCZ Core Series V1 64GB SSD as my main drive.
The motherboard is a ASUS p5n-em hdmi and I'm using the SATA 2 controller which comes on that motherboard (nvidia).

I've noticed that when the computer is under heavy load (disk intensive) I start to get errors then it freezes.

I've managed to freeze the computer twice now. The first time I couldn't get the errors as it froze in the GUI. The second time I ran the same tests, but put the screen onto ttyv0 so that I could see the errors when it froze.

The errors I was getting are:

g_vfs_done():ad4s1f[WRITE(offset-145440768, length=131072)]error = 6
g_pvafnsi_cd:o nien(i)t:iada4tse1_fw[rWiRt[eT_Ei(noofdfesbelto=ck5_1u43f7s28:6 209l1r2e, aldeyn gsttha=rt1e3d1
 72c)pu]ierdr =o r 1=
 6
g_vfs_done():ad4s1f[WRITE(offset-51438649344, length=131072)]error = 6
g_vfs_done():ad4s1f[WRITE(offset-145620992, length=131072)]error = 6
g_vfs_done():ad4s1f[WRITE(offset-32372391936, length=16384)]error = 6
unknown: TIMEOUT - WRITE_DMA retrying (1 retry left) LBA=8398437

The computer is running on a UPS.
The tests which were performed were:
- load up GUI (fluxbox)
- load some apps (firefox, xmms, abiword, xpdf, mrxvt, gkrellm)
- do a 'portupgrade -Rf php5' & 'portupgrade -Rf net/samba3' as this is what I was originally trying to recompile.
- Run two bonnie++ session at the same time.

These tests were all run at once to really work out the disk drive. The applications were mainly loaded so that swap space would be used (I'm sure there is a better way of doing this, perhaps creating a large memory drive so that it forces the HD to use swap or something like that, please suggest).

Nothing special has been done for the SSD, yet. I was concerned that it was freezing the system and have been discourage from doing any optimisations for the SSD if it's just going to freeze in the end.

[michaelrmgreen]

Are you making any progress? I'm thinking of going this route so I'm interested.

[map7]

The crashing was due to faulty memory and had nothing to do with the SSD in the end. After long tests on the memory (using memtest86) it started failing under heavy load.

I haven't returned to the SSD yet as I would have to bring the server down to use it.

[michaelrmgreen]

Aha, thats interesting, how long did you have to run memtest86 before it showed a fault? What brand of memory was it?

[map7]

I don't know exactly how long it was, 1hr or 2 i suppose. I just left it going and came back to it later with a heap of errors.

The memory was Kingston and I replaced it for the same brand/model. Kingston is pretty good. I usually use kingston or corsair ram in my machines without problems. So I think this was just a one off.