00:12:26 rmustacc: zfs doesn't have lfs's "segment cleaner" which would empty out a metaslab for fresh sequential writing. 00:15:50 alanc: reads mainly like "whoever ported zfs to linux didn't understand linux's user namespaces" 00:22:13 jbk: I think you have the metaslab spacemap thing (which indeed uses less space to keep track of free/allocated when the slab is nearly full) confused with intent logs (which use blocks that aren't marked allocated in the spacemaps..) 00:23:24 yeah it could be the spacemaps i'm think of... it's been a _long_ time since I've had to look that deeply :) 00:24:09 intent logs seem like they'd work well since you're really just appending 00:24:27 in other news I gave Hans's NVME host memory buffer patch (https://code.illumos.org/c/illumos-gate/+/4929) a spin on a test machine where it would used and .. I guess it works? nothing blew up, things seem to be performant. 00:24:28 → CODE REVIEW 4929: 18307 nvme: want support for Host Memory Buffer feature (NEW) | https://www.illumos.org/issues/18307 00:26:13 I don't know nvme-level things well enough to seriously review it. 00:26:48 jbk: until you start reusing the intent log blocks when the transactions commit, and then you're scraping shingles off the roof and reshingling.. 02:30:57 sommerfeld: That's good news. How much memory did the controller want? 12:16:08 I started to work on #18324. I would initially disable the superfluous code with #ifdef, or should I remove it immediately? 14:32:23 I'd just remove the code to create the old NOTE segment entirely. 15:35:44 illumos#18324 15:57:02 hrm.. 15:57:16 it's a debug build, but bge seems to periodically freeze for a few secs at a time... 15:58:09 curious if anyone else has seen this... 16:00:17 I haven't noticed this personally. What bge part? 5719/5720? 16:00:40 5720 16:01:13 it's possible it might be a victim, but i've got the iperf running on the ice nic on the same box, and it's going fine when those pauses happen 16:02:18 i've not really looked at the driver much (yet).. 16:08:44 rmustacc: nvmeadm get-features reports " Host Memory Buffer Size: 65536KiB" 16:12:07 (it also reports: Temp. Sensor 1 Over Temp. Threshold: 65262C and Temp. Sensor 1 Under Temp. Threshold: -273C which looks suspiciously like 0K to 65535K) 16:12:20 (in this case K=Kelvin) 16:15:04 it identifies itself as a "MSI M450 1TB V1" 16:16:35 per nvmeadm identify-controller, it also has Host Memory Buffer Preferred Size: 65536KiB and Host Memory Buffer Minimum Size: 65536KiB so it's 64MB or go home. 16:45:42 pasted the relevant bits into https://www.illumos.org/issues/18307 16:45:43 → FEATURE 18307: nvme: want support for Host Memory Buffer feature (New) | https://code.illumos.org/c/illumos-gate/+/4929 18:04:04 [illumos-gate] 18352 i2c ioctl tests race device tree reconfiguration -- Andy Fiddaman 18:04:04 [illumos-gate] 18327 zen topo module fails enumeration when hardware threads have no logical CPU -- Andy Fiddaman 21:19:38 [illumos-gate] 18305 fix ktest_get_fn(9F) typedef example -- Robert Mustacchi