12:38:39 hello, can't get mellanox connectx-6 40GBE Card to around 5 to 6 Gbits/sec any hints ? /kernel/drv/mlxcx.conf already 'tuned' 12:39:40 mean mellanox Card only gives 5 to 6 Gbits/sec 14:45:30 @jbk --> piadm isn't smart enough to do that yet, but it is only smart enough to check on creation time. 14:55:48 danmcd: ahh ok.. i had been using a pre-piadm hacked up 'boot from local disk' setup and finally wanted to make the switch 14:56:11 turns out i had left space for the EFI partition on the l2arc/slog device anyway.. so just created it 14:56:21 Cool. 14:57:23 also have another system (that's slightly remote) in the same boat, and wanted to try conversion on my local machine first 14:59:04 esp since i goofed the bootloader config when updating the PI, and it was a bit tedious trying to recall all the things to set to get it to load, so want to convert that one too :) 14:59:47 since that one was apparently old enough that the .4th file didn't have the set-platform word(? I think that's the right term) 16:14:27 has anyone attempted an openindiana install on x86 mac mini hardware? 17:22:58 [illumos-gate] 15190 libofmt: comparison of integers of different signs -- Toomas Soome 17:36:07 [illumos-gate] 15193 libdladm: sizeof-pointer-memaccess -- Toomas Soome 17:43:52 [illumos-gate] 15197 libdladm: enumeration values not handled in switch -- Toomas Soome 18:22:33 Dammit Exchange (or Mail.app)... 19:31:18 [illumos-gate] 9293 bootadm: empty line between boot entries in menu.lst cause bootadm list-menu to segfault -- Toomas Soome 19:34:55 [illumos-gate] 15201 bootadm list-menu segfaults for non-existent datasets -- Toomas Soome 20:29:27 does anything actually use the reserved partition on zfs disks? libefi has some comments, but I can't seem to find anything in the kernel that actually touches it 20:29:41 I believe tsoome does for boot-y purposes 20:29:57 nono 20:30:10 oh, is that a separate area? 20:30:27 reserved partition is used to store the fabricated device id 20:30:50 on MBR+VTOC layout, the alternate sectors are used for the same purpose 20:31:00 Yeah the devid goes in there when we make one up 20:31:25 Whether we make one up or not though is generally I think left up to the disk driver itself 20:31:36 yep 20:31:56 Probably that should be a bit more common-code-ish 20:34:40 the code for it is in usr/src/common/devid 20:35:17 Well, by that I mean: it feels like in the cases where we don't have a devid for the disk, maybe cmlb should just make one up uniformly and put it in the right place 20:35:25 rather than leaving it up to the driver 20:35:53 Or at least we should do it uniformly 20:36:04 e.g., maybe blkdev should be doing it