Some information about HP T620 Plus Flexible Thin Client machines for network appliance builds...

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

fohdeesha

Kaini Industries
Nov 20, 2016
2,911
3,437
113
34
fohdeesha.com
Eh, it's whatever limit AMD Jaguar/A50M Hudson M1 Southbridge has for SATA drives (that's the PCH used by the onboard APU).
when was the last time a PCH/SATA controller had an arbitrary drive limit? the sata/SAS1 2TB limit than stemmed from 32-bit block addressing almost 15 years ago?
 
  • Like
Reactions: billc.cn

WANg

Well-Known Member
Jun 10, 2018
1,384
1,050
113
47
New York, NY
when was the last time a PCH/SATA controller had an arbitrary drive limit? the sata/SAS1 2TB limit than stemmed from 32-bit block addressing almost 15 years ago?
Who knows - the APU is nearly 12 years old at this time and it's originally meant for AMD Netbooks so I am sure there's some funky limitation somewhere compared to more modern hardware, like either it's rate limited to SATA2, or there's some CPU cycle hogging issue for DMA, or Spectre mitigation for AMD Jaguar killing their I/O performance. Capacity-wise it's probably not an issue, but cost effectiveness would definitely be a concern.

Technically even pre-LBA28 controllers (silicon deployed prior to 1999) can address 137GB+ drives, it's just that it'll be done using polled IO instead of DMA for anything over the LBA28 limit, so it'll be both slower and more CPU hogging when it does.
 
Last edited:

dennisp

New Member
Apr 1, 2021
19
13
3
Reporting back that stability has improved, I haven't had a crash since doing the upgrades. In addition to switching to mSATA, I removed the m.2 completely from the box, switched from pfSense to OPNsense, and switched from ZFS to UFS. So it is not a perfectly clean test, but things have been stable for about a month now.

I haven't removed my failsafe weekly power cycle yet since it is still remote to me, but so far so good with none of the prior lockups. I did not switch to a RAM disk but that will be a backup option if i see another crash.
Long-term followup on this: I do think that switching to mSATA solved the FreeBSD crashes I was experiencing on m.2. I kept the weekly reboots in place and it ran stable for about a year until I got UFS corruption and it failed to boot after one of the scheduled reboots. The corruption is probably because the reboots were power cuts and not clean shutdowns.

I reinstalled OPNsense, restored, and it's been running fine again with no reboots for about a month now.