Switch to DuckDuckGo Search
   January 3, 2020
< | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31

Toggle Join/Part | bottom
[00:21:13] <jbk> i don't suppose anyone knows offhand of some easy mdb-magic to go from a fd in a proc to a vnode_t ?
[00:22:02] <jbk> (e.g. I want the address of the vnode for fd X in proc Y -- assume file is kept open for a long time)
[00:25:11] <jbk> it looks like i might be able to wander through struct proc->p_user->u_finfo->fi_list... but not sure if there was an easier way i've overlooked
[00:25:39] *** vila <vila!~vila@laubervilliers-659-1-133-94.w80-15.abo.wanadoo.fr> has quit IRC (Quit: Leaving)
[00:33:02] *** vila <vila!~vila@laubervilliers-659-1-133-94.w80-15.abo.wanadoo.fr> has joined #illumos
[00:42:48] <gitomat> [illumos-gate] 12135 ECRC PCIe errors shouldn't be fatal -- Robert Mustacchi <rm at fingolfin dot org>
[00:44:38] <rmustacc> jbk: Try '<pid>::pfiles'
[00:54:12] <jbk> aha! (it wants the proc struct, but much nicer)
[00:54:14] <jbk> thanks
[00:55:43] <jbk> hopefully it'll work :) (going to try to flip the read only flag on a zvol after it's opened to see if that'll induce a zpool failure)
[00:56:02] <jbk> unfortunately, 'zfs set readonly=on' only works before a zvol is first opened
[00:56:37] <jbk> but the zvol strategy routine still checks the flag, so... fingers crossed
[01:22:06] *** Kurlon <Kurlon!~Kurlon@cpe-67-253-141-249.rochester.res.rr.com> has joined #illumos
[02:43:38] *** nde <nde!~nde@c-98-220-146-104.hsd1.in.comcast.net> has joined #illumos
[02:45:31] *** jcea <jcea!~Thunderbi@2001:41d0:1:8a82:7670:6e00:7670:6e00> has quit IRC (Quit: jcea)
[03:27:22] *** X-Scale <X-Scale!~ARM@> has quit IRC (Ping timeout: 268 seconds)
[03:53:46] *** X-Scale <X-Scale!~ARM@> has joined #illumos
[03:57:55] *** zeroSignal <zeroSignal!~sh42@fsf/member/zeroSignal> has quit IRC (Ping timeout: 250 seconds)
[03:59:26] *** mnrmnaugh <mnrmnaugh!~mnrmnaugh@unaffiliated/mnrmnaugh> has quit IRC (Ping timeout: 268 seconds)
[04:03:51] *** mnrmnaugh <mnrmnaugh!~mnrmnaugh@unaffiliated/mnrmnaugh> has joined #illumos
[04:36:26] *** zeroSignal <zeroSignal!~sh42@> has joined #illumos
[04:54:13] *** BOKALDO <BOKALDO!~BOKALDO@> has joined #illumos
[04:58:57] *** kev009 <kev009!~kev009@ip72-222-200-117.ph.ph.cox.net> has quit IRC (Ping timeout: 258 seconds)
[05:06:06] *** kev009 <kev009!~kev009@ip72-222-200-117.ph.ph.cox.net> has joined #illumos
[07:00:41] *** neirac <neirac!~neirac@pc-184-104-160-190.cm.vtr.net> has quit IRC (Read error: Connection reset by peer)
[07:01:00] *** neirac <neirac!~neirac@pc-184-104-160-190.cm.vtr.net> has joined #illumos
[08:16:48] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has quit IRC (Quit: tsoome)
[09:18:33] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has joined #illumos
[09:54:08] *** nde <nde!~nde@c-98-220-146-104.hsd1.in.comcast.net> has quit IRC (Remote host closed the connection)
[10:13:44] *** drscream <drscream!~drscream@coconut.frubar.net> has quit IRC (Ping timeout: 246 seconds)
[10:14:44] *** drscream <drscream!~drscream@coconut.frubar.net> has joined #illumos
[10:48:12] <sjorge> jbk you don't happen to have notes?
[10:48:37] <sjorge> I haven't gotten vanilla LDAP auth to work yet, but I noticed ldapclient (will be tricker to setup using salt than some configs) has support for the RBAC files
[10:48:46] <sjorge> So there must be some schema extension I am missing
[10:57:30] *** man_u <man_u!~manu@89-92-19-81.hfc.dyn.abo.bbox.fr> has joined #illumos
[11:01:19] *** arnoldoree <arnoldoree!~arnoldore@ranoldoree.plus.com> has joined #illumos
[11:03:49] <v_a_b> sjorge I am running an OpenLDAP server in a Solaris 11.3 zone. Clients are S10, S11.3/11.4 and OmniOS CE. So vanilla LDAP auth works.
[11:04:30] <v_a_b> Or did you mean "auto provision with salt doesn't work yet"?
[11:57:31] *** igork1 <igork1!~igork@static.> has joined #illumos
[11:59:50] *** igork <igork!~igork@> has quit IRC (Ping timeout: 265 seconds)
[12:08:34] *** neirac <neirac!~neirac@pc-184-104-160-190.cm.vtr.net> has quit IRC (Remote host closed the connection)
[12:08:59] *** neirac <neirac!~neirac@pc-184-104-160-190.cm.vtr.net> has joined #illumos
[12:36:38] <andyf> jbk - thanks, I will take a look!
[12:59:01] *** elegast <elegast!~elegast@83-161-180-31.mobile.xs4all.nl> has joined #illumos
[13:09:41] <andyf> jbk I'm not sure what you mean about zfd and zlogin -I - is it related to the HUP error, and is this something that needs fixing in gate too?
[13:21:27] *** mnowak_ <mnowak_!~mnowak_@> has quit IRC (Quit: Leaving)
[13:25:06] *** mnowak_ <mnowak_!~mnowak_@> has joined #illumos
[13:50:10] *** igork1 is now known as igork
[14:10:23] <neirac> rmustacc hi, I have a Intel(r) Xeon(r) CPU E3-1225 v5 @ 3.30GHz, but according to kstat only supports 1 cstate and 1 frequency https://pastebin.com/tF8HwRaP ; but according to this doc it should support more than 1 c-state. I'm running SmartOS on KVM with host-passthrough configuration.
[14:32:36] *** ngchk1 <ngchk1!~ngchk1@> has quit IRC (Remote host closed the connection)
[14:32:46] *** ngchk1 <ngchk1!~ngchk1@b2b-92-50-91-166.unitymedia.biz> has joined #illumos
[14:36:47] *** ngchk1_ <ngchk1_!~ngchk1@b2b-92-50-91-166.unitymedia.biz> has joined #illumos
[14:36:55] *** ngchk1 <ngchk1!~ngchk1@b2b-92-50-91-166.unitymedia.biz> has quit IRC (Read error: Connection reset by peer)
[14:38:33] *** gh34 <gh34!~textual@cpe-184-58-181-106.wi.res.rr.com> has joined #illumos
[14:44:41] *** BOKALDO <BOKALDO!~BOKALDO@> has quit IRC (Quit: Leaving)
[14:45:08] <tomww> you might not see the SmartOS host entering power save states. the cpu might switch off unused parts, but keept the clock high.
[14:47:19] <tomww> it is just not practical on a host with multiple VMs to reduce the clock speed.
[14:47:49] *** elegast_ <elegast_!~elegast@83-161-171-156.mobile.xs4all.nl> has joined #illumos
[14:49:17] *** elegast <elegast!~elegast@83-161-180-31.mobile.xs4all.nl> has quit IRC (Ping timeout: 258 seconds)
[14:54:01] *** yomisei <yomisei!~void@ip4d16bc07.dynamic.kabel-deutschland.de> has quit IRC (Quit: How to defend yourself against an attacker armed with a mathematician)
[14:56:45] *** yomisei <yomisei!~void@ip4d16bc07.dynamic.kabel-deutschland.de> has joined #illumos
[14:57:59] <neirac> tomww thanks, that's right. but I was checking the output of kstat -m cpu_info I think that should report more than one c-states that are available for the cpu right?.
[14:59:14] <dsockwell> have you tried these measurements on bare metal?
[14:59:49] <dsockwell> smartos is a live USB distribution, just don't let it make a zpool with your hard drives
[15:08:23] <rmustacc> neirac: I believe this is a side effect of SmartOS's setting to disable deep c states.
[15:08:31] <rmustacc> via /etc/system.
[15:08:31] *** Kurlon <Kurlon!~Kurlon@cpe-67-253-141-249.rochester.res.rr.com> has quit IRC (Ping timeout: 268 seconds)
[15:08:51] <rmustacc> Which does have unfortunate consequences with turbo boost as well.
[15:33:19] <neirac> rmustacc: oh, ok is fine then, thank you.
[15:36:56] <rmustacc> Well, maybe you'll want to file an issue to have folks at Joyent re-evaluate it. Given the lack of turbo-boost that can create a confusing thing, especially with the modern reliance on turbo for perf for folks.
[15:38:44] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has quit IRC (Ping timeout: 268 seconds)
[15:41:41] <neirac> rmustacc I was working on tickless so that's what I was checking c-states and processor groups, I'll report the issue for turbo boost, thanks again.
[15:44:58] *** Kurlon <Kurlon!~Kurlon@bidd-pub-03.gwi.net> has joined #illumos
[15:47:12] <rmustacc> OK.
[15:48:42] <sensille> oh, c-states again? i had to issues, had to set it in bios to 'os controlled', after that more than one c-state showed up. also had to re-enable deep-c states, so that turbo boost actually gets used
[15:51:05] *** Kruppt <Kruppt!~Kruppt@> has joined #illumos
[15:53:21] <neirac> sensille thanks a lot
[16:04:02] *** BOKALDO <BOKALDO!~BOKALDO@> has joined #illumos
[16:12:32] <jbk> andyf: I guess the zfd and zlogin -I stuff were things added for lx, so they're not (yet) in illumos-gate, but appear to be in illumos-omnios (as part of the lx side pull)
[16:13:12] <jbk> unfortunately, the info I have at the moment (people still out, etc) is sparse, and the man pages don't really go into the 'why'
[16:13:46] <jbk> other than on triton at least, they're used with the docker stuff (at least)
[16:14:30] <jbk> (and that when the zone would shut down, zlogin wasn't exiting like it shuold have -- though checking for POLLHUP on the zone's stdout/stderr fds appears to have fixed that)
[16:14:42] <andyf> You originally said 'not in illumos-joyent' - which confused me
[16:15:02] <andyf> but it was around NYE :D
[16:16:12] <jbk> oh.. i mean illumos-gate :P sorry
[16:16:16] <jbk> err meant
[16:16:38] <andyf> I thought for a minute that we had side pulled something, and then that was improved/removed/reverted in smartos
[16:16:53] *** mnrmnaugh <mnrmnaugh!~mnrmnaugh@unaffiliated/mnrmnaugh> has quit IRC (Remote host closed the connection)
[16:16:58] <andyf> The HUP thing looks sane
[16:17:54] <jbk> there's an unrelated bug (i.e. it's present even before your zlogin fix) where after zlogin -I exits, the pty is horribly messed up, but that at least doesn't appear to be a regression
[16:18:56] <jbk> not sure what's going on with that.. (basically all writes to the terminal stop happening.. though I can't tell if they're being discarded, being queued, or what)
[16:19:23] <jbk> input from the terminal still works -- you can run commands, etc. just never see any output in the shell that invoked zlogin -I
[16:21:39] <andyf> but only with -(capital i)?
[16:21:46] <jbk> yeah
[16:21:57] <andyf> ok, so probably not necessary for gate at least
[16:21:57] <jbk> which makes me wonder if maybe it's a zfd bug
[16:36:41] *** mnrmnaugh <mnrmnaugh!~mnrmnaugh@unaffiliated/mnrmnaugh> has joined #illumos
[17:14:00] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has joined #illumos
[18:12:29] *** Kurlon <Kurlon!~Kurlon@bidd-pub-03.gwi.net> has quit IRC (Ping timeout: 265 seconds)
[18:22:18] <sjorge> v_a_b, I got it to work... I had a type in pam.conf which was causing issues
[18:22:37] <sjorge> v_a_b are you also storing rbac and sudo info in LDAP?
[18:22:55] <sjorge> I have found the schema for sudo so I want to get that working next, but I still have not found info on Solaris RBAC
[18:24:54] <sjorge> Will also need to look into pam and homedir creation
[18:29:24] *** Kurlon <Kurlon!~Kurlon@bidd-pub-03.gwi.net> has joined #illumos
[18:37:53] *** Kurlon <Kurlon!~Kurlon@bidd-pub-03.gwi.net> has quit IRC (Ping timeout: 260 seconds)
[18:43:49] <v_a_b> sjorge Not yet... I am planning to put ssh keys into LDAP. RBAC will be more of a test, in case customers might want it. But I am planning so many things :-(
[18:46:23] <v_a_b> Solaris 11.4 comes with a nice OpenLDAP server setup script that creates a running slapd configuration, complete with all schema extensions needed for Solaris.
[18:52:44] *** Kurlon <Kurlon!~Kurlon@bidd-pub-03.gwi.net> has joined #illumos
[18:58:08] <jbk> you can get them out of the idsconfig script
[19:04:59] <v_a_b> Yep. The nice thing about the S11.4 ldap server config script is that it not only replaces idsconfig, but works for either DSEE or OpenLDAP (which needs slightly different tweaks IIRC).
[19:07:10] <jbk> unfortunately, the way the RBAC data is mapped to ldap seems to repeat the mistakes of the past (even at the time it was done)
[19:07:15] <jbk> i mean it works
[19:07:42] <jbk> but basically treats ldap as a 'dumb store'
[19:29:16] <sjorge> jbk you mean ldap, the global text file ?
[19:29:41] <sjorge> Looks like pam_mkhomedir can't be compiled against our pam
[19:29:50] <sjorge> linux seems to have diverged :/
[19:29:50] <jbk> it used to
[19:30:06] <sjorge> Going to look into autofs, as a solution
[19:30:08] <jbk> well the linux pam modules had a number of problems
[19:30:19] <jbk> like for years, they were never thread safe
[19:31:00] <sjorge> Ooh nice, someone did it before https://ops.faithlife.com/?p=179
[19:31:04] <sjorge> Going to try that after dinner
[19:33:39] <LeftWing> You could certainly make it work
[19:34:00] <LeftWing> Our PAM is fundamentally capable
[19:35:01] <jbk> IIRC, one of the biggest things is linux made some functions parameters const, we still don't
[19:35:10] <LeftWing> We could! :D
[19:36:44] <jbk> yes, but ISTR I think that ends up altering the pam_sm backend too, whic technically would be breaking the interface
[19:37:20] <jbk> though probably only to the point that anything that wouldn't be fixed by just updating the function signatures is probably doing something they shouldn't
[19:53:39] *** wacki <wacki!~wacki@i577B845D.versanet.de> has joined #illumos
[20:03:14] *** Kurlon <Kurlon!~Kurlon@bidd-pub-03.gwi.net> has quit IRC (Ping timeout: 240 seconds)
[20:10:06] *** aaro <aaro!aaro@gateway/shell/xshellz/x-sxaqfqkqbhnfefbp> has quit IRC (K-Lined)
[20:33:37] *** aaro <aaro!aaro@gateway/shell/xshellz/x-xquhvuzvjhvdzyhv> has joined #illumos
[21:21:37] *** man_u <man_u!~manu@89-92-19-81.hfc.dyn.abo.bbox.fr> has quit IRC (Quit: man_u)
[21:22:47] *** Kurlon <Kurlon!~Kurlon@bidd-pub-03.gwi.net> has joined #illumos
[21:35:27] *** BOKALDO <BOKALDO!~BOKALDO@> has quit IRC (Quit: Leaving)
[21:38:26] *** kev009 <kev009!~kev009@ip72-222-200-117.ph.ph.cox.net> has quit IRC (Remote host closed the connection)
[21:57:55] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has quit IRC (Quit: tsoome)
[22:00:25] *** tsoome <tsoome!~tsoome@> has joined #illumos
[22:09:12] *** wacki <wacki!~wacki@i577B845D.versanet.de> has quit IRC (Quit: Lingo: www.lingoirc.com)
[22:22:40] *** kev009 <kev009!~kev009@ip72-222-200-117.ph.ph.cox.net> has joined #illumos
[22:52:34] *** jimklimov <jimklimov!~jimklimov@> has joined #illumos
[23:01:41] *** edef <edef!edef@NixOS/user/edef> has quit IRC (Ping timeout: 268 seconds)
[23:01:57] *** edef <edef!edef@NixOS/user/edef> has joined #illumos
[23:03:38] *** Kurlon_ <Kurlon_!~Kurlon@bidd-pub-04.gwi.net> has joined #illumos
[23:04:46] <jbk> LeftWing: how usable is your vioscsi driver? :P
[23:04:53] <LeftWing> lol
[23:04:58] <LeftWing> Why do you ask
[23:05:53] <jbk> it would be incredibly useful to be able to have a disk device in a VM I can offline (as in cause ndi_devi_offline() to be called)
[23:06:40] <jbk> trying to recreate illumos#4454 -- mgerdts was trying to use iscsi w/ it, but the number of iscsi bugs he kept hitting kept growing
[23:07:14] *** Kurlon <Kurlon!~Kurlon@bidd-pub-03.gwi.net> has quit IRC (Ping timeout: 258 seconds)
[23:07:25] <jbk> so i'm wondering if there's something that while perhaps not entirely polished, might be good enough to recreate + tet fix
[23:07:55] <LeftWing> Ahh
[23:08:00] *** Kurlon_ <Kurlon_!~Kurlon@bidd-pub-04.gwi.net> has quit IRC (Ping timeout: 258 seconds)
[23:08:03] <LeftWing> I mean, I have made disks come and go
[23:08:05] <LeftWing> on GCE
[23:08:09] <LeftWing> You could give it a whirl!
[23:08:56] <LeftWing> jbk: I think my latest work is in https://github.com/jclulow/illumos-gate/commits/gce
[23:09:10] <jbk> virtualbox appears to support it (as a hypervisor), so my hope was I could spin up an illumos guest under it, chill w/ dtrace, and get my panic
[23:10:21] <jbk> ahh cool.. thanks i'll give it a whirl
[23:10:29] <rmustacc> May be worth checking of nvemadm detach with a virtual nvme disk can trigger it.
[23:10:53] <rmustacc> I'm not sure if that path will technically cause an nvme_devi_offline() though.
[23:11:23] <jbk> do any hypervisors emulate nvme? or just pass-thru?
[23:12:57] <dsockwell> I've thought it was progress when they started emulating SATA instead of just SCSI and IDE
[23:13:41] <dsockwell> Not that I know anything specific but I wouldn't hold my breath
[23:16:02] <rmustacc> jbk: qemu, vmware, etc.
[23:18:15] <andyf> jbk even bhyve...
[23:20:41] <jbk> oh wow.. so it does.. and it looks like it uses blockif for the backend..
[23:21:26] <andyf> and can be memory backed too.
[23:48:22] *** gh34 <gh34!~textual@cpe-184-58-181-106.wi.res.rr.com> has quit IRC (Quit: Textual IRC Client: www.textualapp.com)
[23:51:08] <jbk> hrm..
[23:52:02] <jbk> nvmeadm detach fails w/ device busy once i add the device to a zpool.. though i wonder if I can work around that..
[23:56:37] *** Kurlon <Kurlon!~Kurlon@cpe-67-253-141-249.rochester.res.rr.com> has joined #illumos

   January 3, 2020
< | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31