03:03:25 nice 04:31:09 hmm interesting... vmware's NVMe emulation says it supports dataset management, but if you issue a dataset mgmt command, it returns an invalid cmd error that causes a panic 04:31:23 (i.e. don't try to trim a VMware NVMe zpool :P) 09:46:22 Ouch, you should definitely open a bug for that if you haven't already. 09:53:39 oooofff 13:13:42 yeah, we probably should perhaps be a bit smarter about handling invalid commands 13:14:25 though the NVMe spec refers to some parameter limits that (at least in current versions) don't appear to actually be documented (at least trying to search through the PDFs) 13:14:58 as I was wondering if maybe our 'can we do TRIM on NVMe' needs additional checks beyond the 'support dataset management' bit 13:23:51 jbk for a 20h/00h INVALID COMMAND OPERATION CODE on vioscsi I need to check actually sd driver or where to find out which cmd was sent ? 13:54:21 probably 14:22:17 i'm extremely sleep deprived today, so not exactly thinking straight 14:24:15 jbk no problem, seems is really an issue with the hypervisor that they are using in oci, I just tested openbsd 7.2 and it crashes when loading vioscsi, it used to work 16:29:58 Does someone have a system with mpt_sas sitting around that they could help test something on? 16:49:26 rmustacc: i have one 16:51:16 Woodstock: Are you in a place that you could run something on it and try a disk pull? 16:53:59 rmustacc: yes, that should work 16:54:19 I'll e-mail you details? 16:54:23 sure 17:01:18 I also have two boxes using mpt_sas I believe if you need additional data points 17:05:10 Yeah, that will.