Switch to DuckDuckGo Search
   September 20, 2019  
< | 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 | >

Toggle Join/Part | bottom
[00:00:22] *** dlg <dlg!~dlg@toy.eait.uq.edu.au> has quit IRC (Ping timeout: 244 seconds)
[00:00:34] *** dlg <dlg!~dlg@toy.eait.uq.edu.au> has joined #illumos
[00:12:58] *** gwr <gwr!~gwr@151.203.114.175> has quit IRC (Quit: This computer has gone to sleep)
[00:23:26] *** merzo <merzo!~merzo@243-8-132-95.pool.ukrtel.net> has quit IRC (Ping timeout: 276 seconds)
[00:39:39] *** merzo <merzo!~merzo@48-50-133-95.pool.ukrtel.net> has joined #illumos
[00:48:17] <AmyMalik> danmcd: So Net used to not be part of the Solaris OS
[00:48:30] <AmyMalik> That's...
[00:49:21] <LeftWing> There are multiple decades of history here; operating systems have evolved a lot over the life of the code we now have in illumos
[00:50:11] <AmyMalik> LeftWing: Decades of history my brain is not ready to comprehend.
[00:52:45] <AmyMalik> danmcd: so that'd be like me calling my breakfast today the omelette/bacon consolidation
[00:52:53] <AmyMalik> (with a lot less history)
[01:05:17] <LeftWing> I do enjoy a good omelette/bacon consolidation
[01:05:30] <AmyMalik> LeftWing: lol what
[01:06:11] <AmyMalik> LeftWing: If I ever open up a UNIX themed diner, should the pork belly and eggs breakfast be called the "omelet(te)/bacon consolidation"?
[01:06:24] <AmyMalik> Or does that not work
[01:07:35] <LeftWing> I feel like a UNIX-themed diner would have to have a lot of menu items that people are not proud of, but which perhaps seemed like a good idea at the time, that you'd need to skip past to get to the edible food
[01:08:42] *** jcea <jcea!~Thunderbi@2001:41d0:1:8a82:7670:6e00:7670:6e00> has quit IRC (Remote host closed the connection)
[01:08:48] <LeftWing> And there'd be a small crowd of peristent Yelp reviewers that feel diners went off the rails after about 1980
[01:11:57] *** jcea <jcea!~Thunderbi@2001:41d0:1:8a82:7670:6e00:7670:6e00> has joined #illumos
[01:12:04] <arekinath> oh no you don't want to order that dish, that gets prepared in the STREAMS part of the kitchen
[01:12:09] <AmyMalik> LeftWing: No, that'd be a Linux themed diner
[01:12:34] <AmyMalik> LeftWing: So, a UNIX diner has vegan version of everything?
[01:12:46] <AmyMalik> Shit you're not proud of but which seemed like a good idea back in 2019
[01:13:27] <AmyMalik> arekinath: fun times, the streams part of the kitchen is actually the cleanest. It's also the slowest, though. Average wait times for streams dishes are between 3 and 5 hours because many of them are prepared sous vide.
[01:13:54] <AmyMalik> And often the cooks run out of memory repeatedly, so they forget about what's on the cooker, so instead of getting tender bacon, you always get crispy.
[01:14:20] <arekinath> exactly, you never get what you want and it's slow and more expensive :P
[01:14:31] <arekinath> next door to the UNIX diner is a small flying saucer with the door hanging off the top hinge and a sign saying "plan 9, beware of the leopard"
[01:14:53] <AmyMalik> Nobody complains, though, because the frybread is always perfectly crispy.
[01:15:52] <AmyMalik> arekinath: yeah, we bought them out ages ago, and we still have the money set aside to fix that, but management doesn't care until the door falls off
[01:16:31] <LeftWing> I feel this metaphor has gotten away from us
[01:17:05] <arekinath> also plan9 ostensibly serves most of the same menu but nobody goes there for that, it only gets business because everyone knows they also sell LSD
[01:17:06] <AmyMalik> Oh and if you come in wearing a Sun, Joyent, Nexenta or OmniTI badge, you can't order anything with TUN or TAP in the name
[01:17:39] <AmyMalik> you can only order the PPP version, which comes pre-salted.
[01:17:49] <LeftWing> Hey, now, Joyent ship the TUN/TAP driver in SmartOS!
[01:17:52] <AmyMalik> This metaphor has fallen away from us.
[01:18:15] <AmyMalik> LeftWing: Yes, they do, but that driver is under the GPL (unless they've paid max_mk and k. aizawa for CDDL terms)
[01:18:24] <AmyMalik> Or are we citing the Ubuntu/ZFS precedent for this?L
[01:19:01] <LeftWing> It's actually fine to ship a GPL kernel module. It's not like the OS kernel is a derived work of the TUN/TAP module.
[01:19:06] <LeftWing> It's just using the same interfaces anybody else can use.
[01:19:26] <LeftWing> We haven't put it in our gate, though, so that it's clearly separate from the CDDL bits in source form
[01:19:49] <LeftWing> I suppose I should start saying "They" :P
[01:19:57] <AmyMalik> LeftWing: You got fired didn't you
[01:20:05] <LeftWing> Yes, I'm basically useless!
[01:20:13] <AmyMalik> Is it fine to ship a CDDL kernel module on a GPL kernel (Ubuntu and ZFS)?
[01:20:21] <AmyMalik> LeftWing: are you basically "fired; still on the payroll" case?
[01:20:32] <LeftWing> I personally feel that it is, but I'm also not a lawyer.
[01:20:48] <LeftWing> It seems to hinge on whether you feel the ZFS module is a derived work of the Linux kernel or not.
[01:21:48] <andyf> Aw, I'm late to the party
[01:21:51] <bahamat> I'm also not a lawyer, but CDDL says that the binaries can be relicensed.
[01:22:04] <andyf> the metaphor looks spent..
[01:22:10] <AmyMalik> andyf: build on it anyway!
[01:22:12] <AmyMalik> :D
[01:22:13] <LeftWing> andyf: Yeah I think we killed the horse
[01:22:51] <AmyMalik> Y'all already know this already, but I'll pretend you don't because I need to practice explaining things, the Ubuntu and ZFS case involves a GNU+Linux distribution including a zfs kernel module (but not loading it by default) in its default distribution. No other GNU+Linux distribution does this that I know of.
[01:23:00] <LeftWing> AmyMalik: As for my departure, it had been seven years and I was ready to try doing something else. I'll still be furniture around here, though, even if it's just in my spare time.
[01:23:09] <AmyMalik> yes, I said already twice.
[01:23:28] <AmyMalik> LeftWing: Holy. So you left 7 years ago and Joyent are still 'we' to you. That's... Nuts.
[01:23:53] <LeftWing> Well I only left a couple of weeks ago. The adjustment of pronouns will take a little while.
[01:24:01] <AmyMalik> OH.
[01:24:34] <AmyMalik> LeftWing: How many lines long is your non-compete?
[01:24:45] <LeftWing> This is California! We don't really do those.
[01:24:51] <AmyMalik> andyf: why not keep on flogging the metaphor anyway? it was fun.
[01:25:06] <andyf> I think it peaked with STREAMS and sous-vide
[01:25:10] <LeftWing> ha
[01:25:14] <AmyMalik> what.
[01:25:26] <AmyMalik> There's other obscure kernel interfaces aren't there?
[01:25:31] *** amrfrsh <amrfrsh!~Thunderbi@134.19.189.92> has quit IRC (Quit: amrfrsh)
[01:26:03] <andyf> There is so much strangeness...
[01:26:11] <jbk> well there's DLPI, though that'd be a subset of STREAMS i suppose
[01:26:19] <andyf> did you know that `netstat` basically gets its information via SNMP over /dev/arp?
[01:26:21] <AmyMalik> Like how sometimes customers ask to run DTrace on their dishes and they end up watching every ingredient go in.
[01:26:23] *** amrfrsh <amrfrsh!~Thunderbi@134.19.189.92> has joined #illumos
[01:26:41] <jbk> andyf: yes, and once I discovered that, I immediately stopped looking :P
[01:26:46] <LeftWing> The Management Information Base!
[01:26:56] <AmyMalik> Okay, this is going from funny to sad isn't it
[01:27:07] <andyf> Or that `pfiles` injects an "agent" thread into the target process so that it can run getsockopt() et al. from inside the process?
[01:27:20] <LeftWing> So, the Agent LWP is both Too Clever and kind of neat
[01:27:27] <jbk> heh
[01:27:35] <AmyMalik> andyf: hoohh what
[01:27:41] <AmyMalik> LeftWing: If joyent is ever hiring for a sous chef for a Unix themed diner, do you think you'll volunteer yourself?
[01:27:43] <andyf> or that there are libc functions that shell out to ksh
[01:27:49] <andyf> I think that's it for discoveries this week..
[01:27:51] <jbk> wordexp
[01:27:51] <AmyMalik> andyf: list please.
[01:28:04] <LeftWing> I think there are a lot of places where we might reasonably add direct interfaces to supplant things done with the Agent LWP today
[01:29:05] <LeftWing> The pfiles grab on a process with lots of sockets is somewhat expensive
[01:29:08] <andyf> jbk DLPI is an example of a nice interface though
[01:30:00] <LeftWing> Valid reasons to ditch the Agent LWP for something could include: doing it that way has too much of a performance impact on the victim process
[01:30:21] <LeftWing> and also: we'd like to allow a different permissions/privileges model than "have proc access to the victim"
[01:30:26] <andyf> Yes, the performance of things like 'lsof -i' on illumos is quite bad
[01:31:03] <andyf> With my netstat change I only grab the process if it has any BSD sockets open, so it's as light touch as I can make it
[01:31:17] <andyf> As it happens, there is no need to grab if it just has TLI/XTI
[01:31:17] <LeftWing> Adding the /proc/$pid/cmdline stuff is one example of a new direct interface that can have better performance and permissions characteristics
[01:31:44] <LeftWing> That's interesting! Perhaps we can fix that on the BSD sockets side in the end
[01:31:49] <arekinath> I'm currently making syscalls with the agent lwp and Psyscall to forcibly change the audit context of a process from outside :D
[01:32:01] <LeftWing> arekinath: Is that working out? :D
[01:32:27] <andyf> To be fair, the agent stuff does seem robust
[01:32:28] <LeftWing> It is certainly a neat way to prototype something like that, anyway
[01:32:28] <arekinath> it is!
[01:32:36] <AmyMalik> Who knows. Maybe a diner analogy is what will end up being the genesis of a gateable TUN/TAP driver. Unlikely though
[01:32:44] *** wiedi_ <wiedi_!~wiedi@ip5b4096a6.dynamic.kabel-deutschland.de> has joined #illumos
[01:32:48] <LeftWing> andyf: It has had a lot of bug fixes (I did one myself!) so I think it's pretty good now yeah
[01:32:57] *** wiedi <wiedi!~wiedi@ip5b4096a6.dynamic.kabel-deutschland.de> has quit IRC (Read error: Connection reset by peer)
[01:33:17] <andyf> Certainly good enough that it isn't considered at all dangerous to run pfiles against a process..
[01:33:42] <andyf> (or maybe those who know how it works feel differently)
[01:34:00] <LeftWing> The only danger I'm aware of is that if you have lots of sockets we can hang up the process for a while
[01:34:06] <LeftWing> While we're gathering all the information
[01:34:14] <AmyMalik> How could that be optimised?
[01:34:29] <AmyMalik> Is there information that can be gathered from outside of the process?
[01:34:40] <LeftWing> By providing an alternative interface (through procfs maybe?) to get the information without taking locks that the victim process needs to operate
[01:34:44] <andyf> Linux (and Solaris) has /proc/<pid>/fdinfo - that's one way that extra information could be retrieved
[01:35:06] <LeftWing> Right, we could provide a side hatch like that
[01:35:17] <LeftWing> That's basically the same model as /proc/$pid/cmdline
[01:35:34] <andyf> IIRC it's /proc/<pid>/fdinfo/<fd>
[01:36:06] <LeftWing> So it is
[01:36:53] <LeftWing> Yeah, this is the right kind of thing
[01:37:01] <LeftWing> If you look at fdinfo in https://docs.oracle.com/cd/E88353_01/html/E37852/proc-5.html
[01:38:05] <andyf> That's almost what the Pfdinfo_iter function returns
[01:38:07] <jbk> new project! :P
[01:38:21] <LeftWing> andyf: I imagine it was an evolution of that
[01:38:31] <andyf> I'd also need address family and socket type though (if S_IFSOCK)
[01:39:05] <LeftWing> andyf: I think those would be entries in pr_misc[]?
[01:39:40] <andyf> oh, so they are
[01:39:46] <LeftWing> PR_SOCKETNAME is a prsockaddr_t which has prsock_family
[01:39:54] <LeftWing> and PR_SOCKOPT_TYPE has the type I guess?
[01:40:15] <andyf> yes, well I get the type today with getsockopt(SO_TYPE)
[01:40:32] <LeftWing> Yup
[01:40:33] <andyf> and PR_SOCKETNAME should be struct sockaddr, so I have ->sa_family
[01:41:07] <LeftWing> I feel like we could just implement this as-is
[01:41:15] <LeftWing> I don't immediately see anything wrong with it
[01:43:39] <andyf> What's the legality of implementing this as Solaris compatible, based on that man page?
[01:43:50] <LeftWing> Pretty sure that's fine
[01:44:16] <LeftWing> We'd be writing our own code
[01:44:26] <LeftWing> We have a prfdinfo_t already
[01:44:34] <andyf> we already have... yes, that
[01:44:57] <AmyMalik> if we were perusing sol11, we'd have gotten it on paper from oracle that they wouldn't sue us in connection with this
[01:45:22] <AmyMalik> because of how litigious they can be
[01:45:38] <LeftWing> andyf: I think we'd have to keep our prfdinfo_t as-is though. I think it's in core files.
[01:45:48] <andyf> yes, the comment says so..
[01:46:12] <LeftWing> But like, we could make the fdinfo file compatible in layout and just give it a new type name or something
[01:46:33] <andyf> cool project
[01:46:42] <despair86> > can't order UTUN if Oracle
[01:46:56] * despair86 reads scrollback
[01:46:57] <AmyMalik> eau: ?
[01:47:01] <AmyMalik> despair86: *
[01:47:04] <AmyMalik> eau: apologies
[01:47:23] <andyf> I suppose changing the core file format has associated difficulties
[01:47:59] <eau> hoi
[01:48:02] <eau> AmyMalik:?
[01:48:16] <AmyMalik> eau: i tab-failed
[01:48:16] <LeftWing> andyf: I mean, we could jump through some hoops to keep the old definition with a different name, etc
[01:48:19] <eau> ahh ok
[01:48:21] <eau> :)
[01:48:31] <eau> hey LeftWing here!
[01:48:43] <LeftWing> Hello!
[01:49:58] * despair86 slinks away for a bit
[01:52:02] *** despair86 is now known as RightWing
[01:52:16] * RightWing returns
[01:53:47] <AmyMalik> ...
[01:55:34] <RightWing> i _don't_ bite...unlike _you_ :-p
[01:56:15] *** gwr <gwr!~gwr@151.203.114.175> has joined #illumos
[01:57:55] <andyf> I'd actually like to expand it in the core file too, the `pfiles` on a core can show more information.. but I don't know what that might break
[01:58:32] <andyf> but it has a knock-on effect on things like elfdump
[01:58:54] <RightWing> > Samsung ships the TAP driver
[01:59:05] <RightWing> yeah but it doesn't ship the header, oddly enough
[01:59:06] <AmyMalik> don't Samsung ship Linux though, not illumos?
[01:59:13] <AmyMalik> or am I delusional
[01:59:21] <LeftWing> Samsung is a large multinational family of companies
[01:59:23] <RightWing> joyent hasn't existed as an independent entity for about three years now...
[01:59:28] *** gh34 <gh34!~textual@cpe-184-58-181-106.wi.res.rr.com> has quit IRC (Quit: Textual IRC Client: www.textualapp.com)
[01:59:31] <LeftWing> I'm pretty sure they're shipping a lot of things
[01:59:33] <AmyMalik> who bought Joyent?
[01:59:35] <RightWing> tru
[01:59:43] <LeftWing> Including shipping containers
[02:00:08] <AmyMalik> Did you know that Hyundai (not the car manufacturer) makes some particularly reliable heavy machinery?
[02:00:18] <LeftWing> I have heard that
[02:00:37] <RightWing> yeah JP seems to like their conglomerates
[02:00:52] <AmyMalik> RightWing: korea, you mean
[02:01:00] <AmyMalik> japan does too but this isn't japan, this is korea
[02:01:02] <RightWing> them too
[02:01:43] <RightWing> i _think_ the thing Hyundai spun off was its electronics division, which now belongs to the state phone company there (hence, _SK_ hynix)
[02:02:45] <AmyMalik> I'm pretty sure hyundai may have sold out multiple divisions
[02:03:15] <RightWing> yeah that was the most recent thing i can recall atm
[02:03:27] * RightWing smiles slightly
[02:06:41] <AmyMalik> did Samsung buy Joyent tho?
[02:06:51] <RightWing> indeed
[02:08:16] <RightWing> well i'm an idiot
[02:08:42] <RightWing> i just remembered i'd made a personal copy of the 2016 skeletal onnv tree
[02:08:49] <RightWing> on github
[02:09:05] <AmyMalik> onnv stands for what? os/net nevada?
[02:09:20] <RightWing> ye
[02:10:40] <RightWing> does XView handle multiple monitors at all? I know CDE doesn't
[02:11:26] <RightWing> i might dig up a copy of the open-source release of xview/openwin and give it a spin
[02:12:28] <danmcd> That onnv tree, IIRC, is leaked-out-of-Oracle and is toxic to anyone who doesn't want the wrath of Larry.
[02:12:57] <RightWing> yeah that's snv_175, no one brought it up until now
[02:13:00] * RightWing shrugs
[02:13:20] <RightWing> (as i recall, the xview public release was olwm, a few demo *tools, and some other bits and pieces)
[02:13:21] <LeftWing> RightWing: Are you not referring to Gordon's skeletal illumos build repo?
[02:13:25] <RightWing> yes
[02:13:34] <LeftWing> danmcd: It's not actually nevada
[02:13:48] <LeftWing> https://github.com/gwr/on-skel/
[02:13:48] <danmcd> Oh...
[02:14:02] <RightWing> yeah i'd forgotten i'd forked a copy a while back
[02:14:15] <danmcd> ... <emily-litella>Never mind</emily-litella>
[02:14:22] <LeftWing> Ha
[02:14:53] <danmcd> I'm glad you know who she is.
[02:15:56] *** Teknix <Teknix!~pds@69.41.134.110> has quit IRC (Ping timeout: 258 seconds)
[02:16:04] <AmyMalik> danmcd: why is it toxic?
[02:16:16] <LeftWing> danmcd: I feel like this is an SNL thing that Bryan probably showed me
[02:16:28] <danmcd> @AmyMalik if it's what LeftWing says, it's not toxic.
[02:16:43] <RightWing> Reinhilde: snv_175 was long after the end of opensolaris...
[02:17:02] <jbk> do not taunt snv_175 ?
[02:17:04] <RightWing> so having the code gave you no rights, and a big fat lawsuit
[02:17:08] <danmcd> @LeftWing Yes, very much an SNL thing. Bryan was likely 2-4 years old when it aired.
[02:17:13] <AmyMalik> RightWing: am I still showing up as "Reinhilde" on your screen?
[02:17:18] <RightWing> sometimes
[02:17:20] <danmcd> (it == those eps)
[02:17:22] *** Teknix <Teknix!~pds@69.41.134.110> has joined #illumos
[02:19:13] <AmyMalik> danmcd: and who is Larry?
[02:19:25] <danmcd> Larry Ellison of Oracle.
[02:21:19] <AmyMalik> Oh.
[02:35:41] *** Yogurt_ <Yogurt_!~Yogurt@104-7-67-228.lightspeed.sntcca.sbcglobal.net> has quit IRC (Remote host closed the connection)
[02:36:41] *** lystra <lystra!~lystra@d53-64-11-169.nap.wideopenwest.com> has quit IRC (Ping timeout: 276 seconds)
[02:36:48] *** Yogurt <Yogurt!~Yogurt@104-7-67-228.lightspeed.sntcca.sbcglobal.net> has joined #illumos
[02:38:17] *** tozhu <tozhu!~tozhu@221.12.108.74> has joined #illumos
[02:41:36] *** Yogurt <Yogurt!~Yogurt@104-7-67-228.lightspeed.sntcca.sbcglobal.net> has quit IRC (Ping timeout: 268 seconds)
[02:51:15] <AmyMalik> AR9462 - is it still unsupported, or is it a case of, all the driver code is there, it's just not picked up by the atheros drivers? I'm considering installing some description of illumos on my home desktop when I get my next hard drive.
[02:52:35] *** merzo <merzo!~merzo@48-50-133-95.pool.ukrtel.net> has quit IRC (Read error: Connection reset by peer)
[02:54:51] *** gwr <gwr!~gwr@151.203.114.175> has quit IRC (Quit: Leaving)
[02:58:30] *** heroux <heroux!sandroco@gateway/shell/insomnia247/x-kcmrbxzazslmdauo> has quit IRC (Ping timeout: 258 seconds)
[02:59:04] <gitomat> [illumos-gate] 11706 crypto tests need to run 32-bit and 64-bit -- Robert Mustacchi <rm at joyent dot com>
[02:59:29] *** heroux <heroux!sandroco@gateway/shell/insomnia247/x-bkwicewywfvogshg> has joined #illumos
[03:02:35] <danmcd> ping yuripv ?
[03:02:52] <AmyMalik> danmcd: why
[03:05:05] <danmcd> I need to know if Yuri can push or not.
[03:05:33] <danmcd> Oh, that ping had nothing to do with your AR9462 question. Sorry.
[03:06:05] <AmyMalik> oh
[03:06:58] <gitomat> [illumos-gate] 11615 should handle locale charmap aliases -- Yuri Pankov <yuri.pankov at nexenta dot com>
[03:10:31] *** merzo <merzo!~merzo@135-13-133-95.pool.ukrtel.net> has joined #illumos
[03:14:50] <AmyMalik> would BSD UFS and illumos UFS happen to be compatible enough that one can mount the other read/write without causing data corruption?
[03:15:22] *** lystra <lystra!~lystra@d53-64-11-169.nap.wideopenwest.com> has joined #illumos
[03:16:06] <AmyMalik> i need to find a USB stick I don't mind trashing and put an illumos distribution on it
[03:19:14] <jbk> ok after updating to bloody, and updating w/ the new env files.. it keeps deleting smatch and then fails because it can run smatch while building tools
[03:20:01] *** ed209 <ed209!~ed209@165.225.128.67> has quit IRC (Remote host closed the connection)
[03:20:03] <jbk> err can't
[03:20:08] *** ed209 <ed209!~ed209@165.225.128.67> has joined #illumos
[03:22:34] <AmyMalik> jbk: that sucks something intense
[03:22:55] *** tozhu <tozhu!~tozhu@221.12.108.74> has quit IRC (Quit: tozhu)
[03:24:11] *** saskaloon <saskaloon!~toor@interminate.indelect.net> has quit IRC (Ping timeout: 258 seconds)
[03:24:57] <jbk> well it seems like an bit of a circular problem
[03:25:09] <jbk> if you delete the thing you're trying to use while building...
[03:26:33] <jbk> i just went old school and built tools then installed it to /opt/onbld and pointed it at that
[03:27:26] <danmcd> Yeah I encountered that once today with '030, tools-bootstrap failed. A second build SEEMED to work, though.
[03:27:50] <jbk> just i'll need to keep an eye out for any smatch updates
[03:28:08] <jbk> well i think it's because it's using cw to build tools
[03:29:44] *** Yogurt <Yogurt!~Yogurt@c-73-189-45-147.hsd1.ca.comcast.net> has joined #illumos
[03:32:50] *** lystra <lystra!~lystra@d53-64-11-169.nap.wideopenwest.com> has quit IRC (Ping timeout: 240 seconds)
[03:34:02] <danmcd> Makes sense, I s'pose. Ahh, I gotta run, 9:30pmET.
[03:36:27] *** Yogurt <Yogurt!~Yogurt@c-73-189-45-147.hsd1.ca.comcast.net> has quit IRC (Remote host closed the connection)
[03:37:44] <AmyMalik> also, uh
[03:37:54] *** jcea <jcea!~Thunderbi@2001:41d0:1:8a82:7670:6e00:7670:6e00> has quit IRC (Quit: jcea)
[03:47:49] <AmyMalik> hold up
[03:52:18] *** Yogurt <Yogurt!~Yogurt@c-73-189-45-147.hsd1.ca.comcast.net> has joined #illumos
[03:56:45] *** Yogurt <Yogurt!~Yogurt@c-73-189-45-147.hsd1.ca.comcast.net> has quit IRC (Ping timeout: 258 seconds)
[04:07:31] *** lystra <lystra!~lystra@172.58.139.241> has joined #illumos
[04:14:03] *** tozhu <tozhu!~tozhu@36.19.34.169> has joined #illumos
[04:21:29] *** tozhu <tozhu!~tozhu@36.19.34.169> has quit IRC (Quit: tozhu)
[04:23:42] <AmyMalik> "streams means something different when shouted"
[04:25:34] *** tozhu <tozhu!~tozhu@36.19.34.169> has joined #illumos
[04:28:06] <Smithx10> ./thinks about changing my name to XWing... or maybe DarkWing :P
[04:29:19] <arekinath> ZeroWing
[04:29:24] <arekinath> <galaxy brain>
[04:30:16] <AmyMalik> ln -s /usr/share/man/man1m /usr/share/man/man8 (?)
[04:30:39] <Smithx10> WestWing (tv show)
[04:31:32] <AmyMalik> i also think Tcl should be installing its manual pages in 3TCL, not N
[04:31:44] <Smithx10> Holy Grail of Wing
[04:31:45] <Smithx10> https://www.thefreedictionary.com/words-that-end-in-wing
[04:34:21] <Smithx10> PigeonWing , thats a good one
[04:38:46] <jbk> @#!@ java
[05:05:17] <jbk> wtf isn't dmake honoring -e
[05:09:53] <jbk> though how is anyone building illumos-gate w/ jdk8 and not have it die on the javadoc strings?
[05:11:40] <AmyMalik> idk
[05:33:58] *** tozhu <tozhu!~tozhu@36.19.34.169> has quit IRC (Quit: tozhu)
[05:34:21] *** tozhu <tozhu!~tozhu@36.19.34.169> has joined #illumos
[05:58:11] *** lystra <lystra!~lystra@172.58.139.241> has quit IRC (Ping timeout: 276 seconds)
[06:01:22] *** tozhu <tozhu!~tozhu@36.19.34.169> has quit IRC (Quit: tozhu)
[06:38:57] *** nrhtr <nrhtr!~jparker@125.130.247.43.static.as24516.net> has joined #illumos
[06:46:02] *** tozhu <tozhu!~tozhu@36.19.53.41> has joined #illumos
[06:53:20] <gitomat> [illumos-gate] 11702 loader: zalloc cstyle cleanup -- Toomas Soome <tsoome at me dot com>
[07:00:22] *** lystra <lystra!~lystra@d53-64-11-169.nap.wideopenwest.com> has joined #illumos
[07:06:25] *** mnowak_ <mnowak_!~mnowak_@94.142.238.232> has quit IRC (Remote host closed the connection)
[07:06:56] *** mnowak_ <mnowak_!~mnowak_@94.142.238.232> has joined #illumos
[07:26:19] *** tozhu <tozhu!~tozhu@36.19.53.41> has quit IRC (Quit: tozhu)
[07:27:40] *** tozhu <tozhu!~tozhu@36.19.53.41> has joined #illumos
[07:28:35] <jbk> i had to add this: https://gist.github.com/jasonbking/6912f17100362a9f9ecddbee9d7bf7bb
[07:28:54] <jbk> so i'm curious how others were getting it to build w/ the java8 jdk
[07:29:10] <jbk> because they changed the behavior of generating javadocs
[07:29:18] <jbk> (which apparently a lot of people were unhappy)
[07:29:24] <jbk> where up to then, it was rather lax
[07:29:32] <jbk> and now they made it super-strict by default
[07:29:39] <jbk> and the libdtrace_jni javadocs fail that
[07:30:01] <jbk> and it looks like there was a bit in there to add the flag to silence it
[07:30:12] <jbk> the XDOCLINT_OPTION string
[07:30:28] <jbk> so i'm trying to figure out what i'm missing...
[07:30:45] <jbk> or if i should go ahead and open an issue and throw that out for review
[07:31:05] <tsoome> we had some noise when there was some other java switch
[07:31:18] <tsoome> like switch to openjdk or like
[07:31:41] <jbk> (i also wouldn't be opposed to just tossing out some/all of the in-gate java, but that's probably another discussion)
[07:32:05] *** mnowak_ <mnowak_!~mnowak_@94.142.238.232> has quit IRC (Quit: Leaving)
[07:32:28] *** tozhu <tozhu!~tozhu@36.19.53.41> has quit IRC (Client Quit)
[07:32:37] *** mnowak_ <mnowak_!~mnowak_@94.142.238.232> has joined #illumos
[07:33:12] *** tozhu <tozhu!~tozhu@36.19.53.41> has joined #illumos
[08:02:12] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has quit IRC (Quit: tsoome)
[08:08:15] *** idodeclare <idodeclare!~textual@cpe-76-185-177-63.satx.res.rr.com> has quit IRC (Quit: My MacBook has gone to sleep. ZZZzzz…)
[08:09:53] *** aszeszo <aszeszo!~aszeszo@unaffiliated/aszeszo> has quit IRC (Quit: aszeszo)
[08:30:40] *** tozhu <tozhu!~tozhu@36.19.53.41> has quit IRC (Quit: tozhu)
[08:38:45] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has joined #illumos
[08:49:55] *** alanc <alanc!~alanc@129.157.69.38> has quit IRC (Remote host closed the connection)
[08:50:21] *** alanc <alanc!~alanc@129.157.69.38> has joined #illumos
[08:58:16] *** aszeszo <aszeszo!~aszeszo@unaffiliated/aszeszo> has joined #illumos
[09:01:33] <yuripv> the only somewhat important java piece is poold (or can it go too?)
[09:02:08] *** amrfrsh <amrfrsh!~Thunderbi@134.19.189.92> has quit IRC (Ping timeout: 276 seconds)
[09:02:34] <AmyMalik> why would there be java in base?
[09:07:27] *** nrhtr <nrhtr!~jparker@125.130.247.43.static.as24516.net> has quit IRC (Quit: WeeChat 2.4)
[09:08:45] <andyf> jbk you have a knack for finding new problems.. lots of us build gate regularly on omnios and don't have to do anything special in regard to cw etc.
[09:08:58] <andyf> jbk - https://downloads.omniosce.org/env/build-gate-on-bloody.env
[09:09:20] <andyf> although I think the templates in /opt/onbld/env/ should be good too
[09:14:38] <gitomat> [illumos-gate] 11683 Automatically maintain whatis database -- Andy Fiddaman <omnios at citrus-it dot co.uk>
[09:14:39] <gitomat> [illumos-gate] 11687 isaexec() leaks memory if realloc fails -- Andy Fiddaman <omnios at citrus-it dot co.uk>
[09:16:00] <andyf> jbk I would like to understand what's happening though so we can fix it and make it as easy as possible to build on omnios
[09:27:33] *** andy_js <andy_js!~andy@97e29e78.skybroadband.com> has joined #illumos
[09:28:18] <yuripv> AmyMalik: just to make everyone ask that question! :D
[09:29:29] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[09:41:38] *** lystra <lystra!~lystra@d53-64-11-169.nap.wideopenwest.com> has quit IRC (Ping timeout: 240 seconds)
[09:44:10] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[09:55:00] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[09:55:03] *** lystra <lystra!~lystra@d53-64-11-169.nap.wideopenwest.com> has joined #illumos
[10:14:56] *** cartwright <cartwright!~chatting@gateway/tor-sasl/cantstanya> has quit IRC (Ping timeout: 260 seconds)
[10:23:48] *** man_u <man_u!~manu@manu2.gandi.net> has joined #illumos
[10:26:44] *** mnowak_ <mnowak_!~mnowak_@94.142.238.232> has quit IRC (Quit: Leaving)
[10:45:23] *** cartwright <cartwright!~chatting@gateway/tor-sasl/cantstanya> has joined #illumos
[10:47:02] *** ptribble <ptribble!~ptribble@cpc92716-cmbg20-2-0-cust138.5-4.cable.virginm.net> has joined #illumos
[10:52:45] *** mnowak_ <mnowak_!~mnowak_@94.142.238.232> has joined #illumos
[10:58:03] *** merzo <merzo!~merzo@135-13-133-95.pool.ukrtel.net> has quit IRC (Ping timeout: 240 seconds)
[11:09:44] *** andy_js_ <andy_js_!~andy@97e29e78.skybroadband.com> has joined #illumos
[11:18:22] *** c3bolla <c3bolla!c3bolla@gateway/shell/panicbnc/x-uoikgxlcooehkjga> has quit IRC (Quit: PanicBNC - http://PanicBNC.net)
[11:19:10] *** andy_js_ <andy_js_!~andy@97e29e78.skybroadband.com> has quit IRC (Quit: andy_js_)
[11:19:18] *** c3bolla <c3bolla!c3bolla@gateway/shell/panicbnc/x-pxjhhagrdhlvldmk> has joined #illumos
[11:36:23] *** m4rley <m4rley!~m4rley@207.148.96.120> has quit IRC (Ping timeout: 258 seconds)
[11:37:00] *** m4rley <m4rley!~m4rley@207.148.96.120> has joined #illumos
[11:37:54] *** andy_js_ <andy_js_!~andy@97e29e78.skybroadband.com> has joined #illumos
[11:47:18] <jeffpc> yuripv: in unleashed we ripped out poold
[11:47:39] <jeffpc> IIRC, poold is only a policy/management layer
[11:47:52] <jeffpc> so, it can get ripped out pretty easily
[11:48:18] <jeffpc> and if someone really wants the functionality, they can resurrect poold or (better yet) reimplement it
[11:48:20] <andyf> I'm sure somebody had a half written replacement in C.. or was that intrd?
[11:49:18] <jeffpc> that was intrd
[11:49:47] <jeffpc> or at least intrd had such a replacement
[11:49:55] <jeffpc> it's possible someone did the same for poold
[11:54:21] <gitomat> [illumos-gate] 11591 usbser: cast between incompatible function types -- Toomas Soome <tsoome at me dot com>
[11:58:27] <gitomat> [illumos-gate] 11592 sdp: cast between incompatible function types -- Toomas Soome <tsoome at me dot com>
[12:03:49] <jlevon> I would have zero issue with saying goodbye to poold
[12:04:33] <igork> leoric/alp like/use it
[12:05:46] <jlevon> they do?
[12:06:15] <igork> he is using it on OI
[12:07:18] <igork> maybe used - i think he can comment it
[12:13:55] *** andy_js_ <andy_js_!~andy@97e29e78.skybroadband.com> has quit IRC (Quit: andy_js_)
[12:14:18] *** andy_js_ <andy_js_!~andy@97e29e78.skybroadband.com> has joined #illumos
[12:14:42] *** andy_js_ <andy_js_!~andy@97e29e78.skybroadband.com> has quit IRC (Client Quit)
[12:14:52] <leoric> Well, I haven't said I use it, but I really like its concept
[12:15:06] *** andy_js_ <andy_js_!~andy@97e29e78.skybroadband.com> has joined #illumos
[12:15:21] <andyf> There are OmniOS customers that use it too
[12:15:23] <leoric> I've just plaid with it a bit, and it was interesting
[12:15:31] *** andy_js_ <andy_js_!~andy@97e29e78.skybroadband.com> has quit IRC (Client Quit)
[12:15:38] <andyf> We made it optional and not part of the base install, but it's still there
[12:20:01] *** ed209 <ed209!~ed209@165.225.128.67> has quit IRC (Remote host closed the connection)
[12:20:08] *** ed209 <ed209!~ed209@165.225.128.67> has joined #illumos
[12:20:18] *** merzo <merzo!~merzo@185.39.197.205> has joined #illumos
[12:56:40] <jlevon> andyf: OK,fine
[12:58:34] <kayront> do you know if ipf supports breaking the config file in chunks? some sort of include functionality
[12:58:39] <kayront> that would make life a lot easier with templating etc
[13:00:01] <jperkin> you could pre-process it with m4
[13:09:35] *** merzo <merzo!~merzo@185.39.197.205> has quit IRC (Ping timeout: 265 seconds)
[13:11:53] *** gh34 <gh34!~textual@cpe-184-58-181-106.wi.res.rr.com> has joined #illumos
[13:12:07] <tsoome> depends on config, but our SMF services also add ipf rules.
[13:13:10] <kayront> yes
[13:13:17] <kayront> but say you elect not to use the framework
[13:13:23] <kayront> is it possible to include files from ipf.conf?
[13:14:12] <kayront> I understand it should be possible to achieve more or less the same with the firewall framework + custom rules per service
[13:14:36] <kayront> but I want to lock down things a bit more (for instance with the framework it has a pass out rule that allows all traffic out, in this situation I don't want that)
[13:15:10] <jimklimov> kayront: there is no apparent keyword for that in man pages
[13:15:20] <kayront> so I was thinking about adding a variable to my ansible role that specifies whether this host will use the framework or not, and if not adjust with svccfg everything necessary to load /etc/ipf/ipf.conf instead
[13:15:36] <kayront> but then it will get cumbersome to add everything on the main file rather than having per-ansible-role snippets
[13:15:39] <kayront> as I'm doing with pf
[13:16:12] <jimklimov> I think the concept allows for just importing numerous files (or stdin commands) - if you don't flush between, they are appended (IIRC)
[13:17:22] <jimklimov> make a workaround with e.g. /etc/ipf.conf.d/* files, some might be *.disabled (or have another subdir for those), and cat the enabled ones into /etc/ipf.conf ?
[13:17:37] <kayront> that should work
[13:17:39] <kayront> good point
[13:18:15] *** p7mo <p7mo!~p7mo@ricasa01.informatik.uni-bremen.de> has quit IRC ()
[13:19:32] *** jimklimov1 <jimklimov1!~jimklimov@31.7.243.238> has joined #illumos
[13:21:42] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Ping timeout: 245 seconds)
[13:25:42] <jimklimov1> also note the ippool.conf that lets you name address lists and ranges, to keep the rules more simple (e.g. "allow admins to ipmi" with pooled definitions of who are admins and which hosts are ipmi cards)
[13:39:16] *** jimklimov1 <jimklimov1!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[13:39:42] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[13:49:30] *** amrfrsh <amrfrsh!~Thunderbi@134.19.189.92> has joined #illumos
[14:06:33] *** sjorge <sjorge!~sjorge@unaffiliated/sjorge> has quit IRC (Quit: 410 Gone)
[14:08:54] *** sjorge <sjorge!~sjorge@unaffiliated/sjorge> has joined #illumos
[14:32:06] <gitomat> [illumos-gate] 11693 Document mdb(1) ::ps flags -- Jason King <jason.king at joyent dot com>
[14:32:08] *** aszeszo <aszeszo!~aszeszo@unaffiliated/aszeszo> has quit IRC (Remote host closed the connection)
[14:37:09] *** aszeszo <aszeszo!~aszeszo@unaffiliated/aszeszo> has joined #illumos
[14:44:20] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[14:44:52] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[14:45:46] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[14:46:22] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[14:50:43] *** amrfrsh <amrfrsh!~Thunderbi@134.19.189.92> has quit IRC (Remote host closed the connection)
[14:51:10] *** amrfrsh <amrfrsh!~Thunderbi@134.19.189.92> has joined #illumos
[15:15:39] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has quit IRC (Ping timeout: 258 seconds)
[15:17:51] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[15:18:32] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[15:20:31] *** idodeclare <idodeclare!~textual@cpe-76-185-177-63.satx.res.rr.com> has joined #illumos
[15:21:13] *** idodeclare <idodeclare!~textual@cpe-76-185-177-63.satx.res.rr.com> has quit IRC (Client Quit)
[15:22:23] *** Kurlon <Kurlon!~Kurlon@cpe-67-253-141-249.rochester.res.rr.com> has quit IRC (Ping timeout: 276 seconds)
[15:49:19] *** jcea <jcea!~Thunderbi@2001:41d0:1:8a82:7670:6e00:7670:6e00> has joined #illumos
[15:57:08] *** Kurlon <Kurlon!~Kurlon@bidd-pub-03.gwi.net> has joined #illumos
[15:59:54] *** Kurlon_OI <Kurlon_OI!~jcoombs@bidd-pub-03.gwi.net> has quit IRC (Remote host closed the connection)
[16:05:52] *** MilkmanDan <MilkmanDan!~dan@wilug/expat/MilkmanDan> has quit IRC (Ping timeout: 245 seconds)
[16:06:54] *** MilkmanDan <MilkmanDan!~dan@wilug/expat/MilkmanDan> has joined #illumos
[16:16:32] *** DrLou <DrLou!~DrLou@c-73-33-92-169.hsd1.nj.comcast.net> has joined #illumos
[16:21:19] <jbk> andyf: that's what i don't get.. with the java8 changes to javadoc, i don't see how it was possible anyone was successfully building illumos-gate w/o that change, yet they were
[16:21:53] <jbk> (building using the java8 jdk)
[16:27:45] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Ping timeout: 265 seconds)
[16:45:49] <andyf> jbk is it just as simple as export BLD_JAVA_7='#'; export BLD_JAVA_8= ?
[16:46:07] <jbk> no
[16:46:10] <jbk> that's the thing
[16:46:13] <jbk> I have that in my env file
[16:46:26] <jbk> and it does horribly on the libdtrace_jni javadoc build
[16:46:29] <jbk> without that patch
[16:46:39] <jbk> (and it's done that every time i've tried using jdk8)
[16:46:59] <jbk> which is why i'm absolutely baffed how people are building w/ jdk8 without that
[16:47:42] <jbk> and makes me think i'm missing something
[16:58:19] *** perlgod <perlgod!~cullum@puffy.c0ffee.net> has joined #illumos
[17:01:17] *** MilkmanDan <MilkmanDan!~dan@wilug/expat/MilkmanDan> has quit IRC (Ping timeout: 245 seconds)
[17:02:26] *** MilkmanDan <MilkmanDan!~dan@wilug/expat/MilkmanDan> has joined #illumos
[17:16:13] <jbk> i guess i'll file a bug + put the patch out for review and see if anyone chimes in..
[17:20:26] *** BH23 <BH23!~BH23@santoroj.plus.com> has joined #illumos
[17:26:42] *** MilkmanDan <MilkmanDan!~dan@wilug/expat/MilkmanDan> has quit IRC (Ping timeout: 245 seconds)
[17:28:09] *** MilkmanDan <MilkmanDan!~dan@wilug/expat/MilkmanDan> has joined #illumos
[17:32:27] *** Yogurt <Yogurt!~Yogurt@c-73-189-45-147.hsd1.ca.comcast.net> has joined #illumos
[17:39:19] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has joined #illumos
[17:41:19] *** BH23 <BH23!~BH23@santoroj.plus.com> has quit IRC (Ping timeout: 268 seconds)
[17:41:52] *** baojg_ <baojg_!~baojg@162.243.44.213> has quit IRC (Remote host closed the connection)
[17:42:35] *** baojg <baojg!~baojg@162.243.44.213> has joined #illumos
[17:45:03] <andyf> jbk we should try and work out what's going on though. I can build a fresh bloody box and do a gate build.
[17:45:33] <jbk> looking at usr/src/lib/libdtrace_jni/java/Makefile
[17:45:40] <jbk> there's two different targets that invoke javadoc
[17:46:00] <jbk> the 'doc' target and $(DOCAPIDESTDIR)/index.html
[17:46:14] <jbk> i wonder if there's a bit of a (sans bug) harmless race
[17:46:37] <jbk> in that the first one (which includes the XDOCLINT_OPTION var) is usually getting there first and building the pages
[17:46:51] <andyf> does it work with make instead of dmake?
[17:46:53] <jbk> so that when the 'doc' target gets run, it's not doing anything
[17:47:32] <andyf> https://paste.ec/paste/PyaMghXB#n+QixA39p9-yAagy0TIPxlYH6EEHQbgZ+lc3b0cPtkC
[17:47:34] <jbk> except in my case, the doc target is getting there first and building it w/o that option
[17:48:30] <jbk> it's not the compiling of the .java files, it's the javadoc generation that fails
[17:48:37] <andyf> ah yes, if I do `make doc` I see the errors
[17:49:33] <jbk> though i'm not sure what's invoking that target
[17:50:04] <andyf> maybe that's the difference.. although the patch won't hurt
[17:50:13] <andyf> if I do `make clobber install`, it generates .html files
[17:52:38] <jbk> w/o error?
[17:52:49] <andyf> yes, but that is different from the doc: target
[17:54:16] <andyf> So it does look like the doc target is broken, but I can't see why it would be being called
[17:59:50] <jbk> yeah, i can't find where it's being called
[17:59:51] *** tozhu <tozhu!~tozhu@171.221.45.90> has joined #illumos
[18:00:02] <jbk> but i suspect what happens is if the install target gets called first, it's fine
[18:00:16] <jbk> but if not, then that's when you have a problem
[18:00:49] *** KungFuJesus <KungFuJesus!~adamstyli@207.250.97.74> has joined #illumos
[18:01:10] <KungFuJesus> https://illumos.org/rb/r/1311 <-- This looks intriguing
[18:01:36] <andyf> jbk - are you using the version of nightly from gate?
[18:01:40] <andyf> just trying to rule things out..
[18:02:15] <jbk> yes
[18:03:00] <andyf> me too..
[18:03:28] <gitomat> [illumos-gate] 11714 temporarily disable ZFS TRIM support -- Jerry Jelinek <jerry.jelinek at joyent dot com>
[18:03:32] <andyf> The strange this is, I probably do anything up to a dozen builds of gate each day. If it's a race I'd expect to have seen it before
[18:03:36] <andyf> *thing
[18:03:55] <andyf> so it is more likely to be environment, although I can't see how
[18:11:09] <andyf> Sorry, I cannot find any way for the doc target to end up being called. Does the nightly.log help at all?
[18:12:52] <jbk> i'm rerunning it now to see if i can figure that out
[18:13:02] <jbk> because i can't see how it's getting called either
[18:13:11] <jbk> which means it's probably buried in some makefile magic somewhere
[18:14:45] <andyf> I just read through the output of `make -p` and other than it being under the .PRECIOUS symbol, I couldn't find anything
[18:15:59] <wiedi_> KungFuJesus: it was a fun weekend project. Besides the open issues andy found I later noticed that the surrounding code also has some more potential memory leaks in error paths. I haven't found the time to fix those yet.
[18:17:25] <jbk> and IIUC, pmooney's thread priority fix i think also makes sendfile less problematic
[18:18:09] <jlevon> we should upstream that
[18:18:11] <jbk> (basically using sendfile could cause a process to unfairly be prioritized over other stuff)
[18:18:15] <jbk> i thought we did
[18:18:18] <jbk> guess not
[18:18:20] <jbk> :)
[18:20:01] <wiedi_> oh, I wonder how that would affect the performance tests
[18:23:38] <andyf> It is still good to be interrupt a `cp`...
[18:23:46] <jbk> and that...
[18:24:06] <jbk> wiedi_: probably a bigger concern on a busy (or multi-tenant) system
[18:24:59] <wiedi_> andyf: yeah, I still do chunks and not the whole file for that reason
[18:26:28] <andyf> FRC in a makefile is FORCE, right?
[18:27:56] <jbk> yes
[18:29:04] *** man_u <man_u!~manu@manu2.gandi.net> has quit IRC (Ping timeout: 265 seconds)
[18:29:49] <andyf> It would be interesting to see if .make.state has anything in when your build fails
[18:30:32] <jbk> i'll see in a bit.. it might be a bit longer since i also have a smartos-live build running in another zone while this builds in an omnios vm
[18:30:39] <jbk> so it's a bit 'slow' ;)
[18:30:51] *** aszeszo <aszeszo!~aszeszo@unaffiliated/aszeszo> has quit IRC (Quit: aszeszo)
[18:30:51] <sjorge> Oh was the TRIM stuff broken?
[18:32:24] <jbk> there's a report of it trashing a zpool, so it's temporairly being put behind a tunable while it's determined what the cause was
[18:32:51] <jbk> (bad SSD firmware, a sd/sata layer issue, zfs trim bug, etc)
[18:33:09] <jbk> btw, the thread priority bug is https://smartos.org/bugview/OS-7753 for those interested
[18:40:09] *** saskaloon <saskaloon!~toor@interminate.indelect.net> has joined #illumos
[18:55:16] *** Kurlon <Kurlon!~Kurlon@bidd-pub-03.gwi.net> has quit IRC (Ping timeout: 264 seconds)
[18:58:54] <gitomat> [illumos-gate] 11712 installboot: ignore the case of missing loader64.efi or loader32.efi -- Toomas Soome <tsoome at me dot com>
[19:00:10] <KungFuJesus> wiedi_: how big were the measured performance gains
[19:02:50] <wiedi_> it mostly reduced system time. There are some numbers/pictures in https://illumos.topicbox.com/groups/developer/T2429439b431dec43-Mb5da9a7a4874e751cae3078b/experiment-cp1-could-use-sendfile
[19:03:22] <KungFuJesus> system time is still time, though
[19:04:10] <gitomat> [illumos-gate] 11718 gptzfsboot: exit() should call __exit() -- Toomas Soome <tsoome at me dot com>
[19:04:34] <KungFuJesus> small text files being unaffected I think is a somewhat expected side effect, as you see in libcmdutil's writefile implementation, the code you touched was in the > some threshold loop
[19:04:55] <KungFuJesus> the smaller files use the standard unix read and write syscalls
[19:06:20] <wiedi_> I removed the threshold for the sendfile case because I was interested if there would be a difference. For such small files it seems there isn't
[19:07:07] <KungFuJesus> context switches that fit well within LLC probably wouldn't, I would think
[19:10:13] <KungFuJesus> I feel like by default reflink copies can have some weird consequences. Maybe I'm just paranoid. I hate how OS X's finder does it implicitly, even across file system boundaries
[19:10:28] <wiedi_> I also think the benchmarks were done before many of the cpu workarounds were added, so that might also be interesting to retest
[19:10:46] <KungFuJesus> sorry, side tracked by reading the first comment on the ML entry
[19:12:12] <wiedi_> reflink() is on my wishlist too ;)
[19:12:45] <KungFuJesus> it just makes me paranoid. I can think of how many times I did a cp mostly just to get a defragmented copy of something
[19:12:59] <KungFuJesus> and Mac OS's new behavior takes reflinks to the extremely in APFS
[19:13:32] <KungFuJesus> forcing finder to do a real copy to another file system requires going through a context menu and using your damn password
[19:13:41] <KungFuJesus> (though regular cp doesn't)
[19:14:36] <KungFuJesus> extreme*
[19:17:25] <KungFuJesus> I guess maybe how it's implemented in Oracle Solaris isn't so bad. An explicit rather than implicit option makes me feel better, anyway
[19:40:23] <gitomat> [illumos-gate] 11713 zdb should not output binary data on terminal -- Toomas Soome <tsoome at me dot com>
[19:43:03] *** Kurlon <Kurlon!~Kurlon@bidd-pub-03.gwi.net> has joined #illumos
[19:46:16] *** gwr <gwr!~gwr@151.203.114.175> has joined #illumos
[19:51:05] <jbk> ..and of course that build succeeded
[19:55:52] *** jcea <jcea!~Thunderbi@2001:41d0:1:8a82:7670:6e00:7670:6e00> has quit IRC (Remote host closed the connection)
[19:56:07] *** jcea <jcea!~Thunderbi@2001:41d0:1:8a82:7670:6e00:7670:6e00> has joined #illumos
[20:15:30] *** Yogurt <Yogurt!~Yogurt@c-73-189-45-147.hsd1.ca.comcast.net> has quit IRC (Remote host closed the connection)
[20:25:54] *** gitomat <gitomat!~nodebot@165.225.148.18> has quit IRC (Remote host closed the connection)
[20:27:08] *** gwr <gwr!~gwr@151.203.114.175> has quit IRC (Read error: Connection reset by peer)
[20:27:38] *** gwr <gwr!~gwr@151.203.114.175> has joined #illumos
[20:27:57] *** ed209 <ed209!~ed209@165.225.128.67> has quit IRC (Ping timeout: 245 seconds)
[20:28:21] *** gitomat <gitomat!~nodebot@165.225.148.18> has joined #illumos
[20:31:01] *** Yogurt <Yogurt!~Yogurt@c-73-189-45-147.hsd1.ca.comcast.net> has joined #illumos
[20:35:18] *** Yogurt <Yogurt!~Yogurt@c-73-189-45-147.hsd1.ca.comcast.net> has quit IRC (Ping timeout: 245 seconds)
[20:50:48] *** clapont <clapont!~clapont@unaffiliated/clapont> has joined #illumos
[20:52:58] *** gwr <gwr!~gwr@151.203.114.175> has quit IRC (Quit: Leaving)
[21:00:58] <gitomat> [illumos-gate] 11704 loader.efi: filter out very large blocksize in efipart -- Toomas Soome <tsoome at me dot com>
[21:11:17] *** tomww <tomww!~tom@unaffiliated/tomww> has quit IRC (Ping timeout: 245 seconds)
[21:11:47] *** tomww <tomww!~tom@unaffiliated/tomww> has joined #illumos
[21:18:23] <jbk> grr.. i hate these heisenbugs
[21:20:45] <jbk> now twice it's been fine
[21:21:08] <jbk> i'll try it w/o a shadow complier this time.. see if that changes things up enough
[21:21:19] <_mjg> just run it every other time
[21:21:26] <jbk> heh
[21:21:27] <yuripv> :)
[21:21:54] <jbk> i think the fix is still sound, but it would be nice to fully understand what's going on (i.e. how that 'doc' target is being called)
[21:25:03] <jbk> because it's not at all obvious how that's happening
[21:25:53] <jbk> though maybe...
[21:26:12] <jbk> let me add an @echo in it's rule that I can search for in nightly.log
[21:26:51] <yuripv> "the fix is still sound" being ripping that java stuff?
[21:27:15] <yuripv> I wonder if anyone uses anything besides poold
[21:27:20] <jbk> https://gist.github.com/jasonbking/6912f17100362a9f9ecddbee9d7bf7bb
[21:27:25] <yuripv> OI for desktop stuff may be?
[21:28:09] <jbk> if you directly build the doc target in a clean dir, you'll see the javadoc errors
[21:28:20] <jbk> in usr/lib/libdtrace_jni
[21:28:41] <jbk> but there's another target that seems to also build the javadocs (that has that fix already)
[21:31:30] <gitomat> [illumos-gate] 11588 loader: isapnp.c: this statement may fall through -- Toomas Soome <tsoome at me dot com>
[21:37:26] *** lystra <lystra!~lystra@d53-64-11-169.nap.wideopenwest.com> has quit IRC (Ping timeout: 276 seconds)
[21:51:15] *** Yogurt <Yogurt!~Yogurt@c-73-189-45-147.hsd1.ca.comcast.net> has joined #illumos
[21:55:39] *** Yogurt <Yogurt!~Yogurt@c-73-189-45-147.hsd1.ca.comcast.net> has quit IRC (Ping timeout: 246 seconds)
[22:20:08] *** ed209 <ed209!~ed209@165.225.128.67> has joined #illumos
[22:21:28] *** Yogurt <Yogurt!~Yogurt@c-73-189-45-147.hsd1.ca.comcast.net> has joined #illumos
[22:23:41] *** lystra <lystra!~lystra@d53-64-11-169.nap.wideopenwest.com> has joined #illumos
[22:24:33] *** gh34 <gh34!~textual@cpe-184-58-181-106.wi.res.rr.com> has quit IRC (Quit: Textual IRC Client: www.textualapp.com)
[22:26:11] *** Yogurt <Yogurt!~Yogurt@c-73-189-45-147.hsd1.ca.comcast.net> has quit IRC (Ping timeout: 276 seconds)
[22:28:19] *** lystra <lystra!~lystra@d53-64-11-169.nap.wideopenwest.com> has quit IRC (Ping timeout: 265 seconds)
[22:40:38] *** lystra <lystra!~lystra@172.58.142.164> has joined #illumos
[22:53:06] *** jelmd <jelmd!~purple@port-ip-88-150-100-66.reverse.mdcc-fun.de> has quit IRC (Quit: Leaving.)
[22:53:15] *** Yogurt <Yogurt!~Yogurt@c-73-189-45-147.hsd1.ca.comcast.net> has joined #illumos
[22:54:11] *** jelmd <jelmd!~purple@port-ip-88-150-100-66.reverse.mdcc-fun.de> has joined #illumos
[22:58:02] *** Yogurt <Yogurt!~Yogurt@c-73-189-45-147.hsd1.ca.comcast.net> has quit IRC (Ping timeout: 276 seconds)
[22:58:30] *** ptribble <ptribble!~ptribble@cpc92716-cmbg20-2-0-cust138.5-4.cable.virginm.net> has quit IRC (Quit: Leaving)
[23:01:30] *** andy_js <andy_js!~andy@97e29e78.skybroadband.com> has quit IRC (Quit: andy_js)
[23:07:07] *** lystra <lystra!~lystra@172.58.142.164> has quit IRC (Ping timeout: 245 seconds)
[23:10:10] *** lystra <lystra!~lystra@172.58.142.164> has joined #illumos
[23:16:26] *** clapont <clapont!~clapont@unaffiliated/clapont> has quit IRC (Ping timeout: 240 seconds)
[23:21:43] *** Kurlon <Kurlon!~Kurlon@bidd-pub-03.gwi.net> has quit IRC (Ping timeout: 268 seconds)
[23:23:53] *** aszeszo <aszeszo!~aszeszo@unaffiliated/aszeszo> has joined #illumos
[23:38:27] *** clapont <clapont!~clapont@unaffiliated/clapont> has joined #illumos
[23:38:59] <perlgod> Anyone successfully using shell accounts from AD? Currently hitting this pam_krb5 issue: https://www.illumos.org/issues/11671
[23:49:41] *** Yogurt <Yogurt!~Yogurt@c-73-189-45-147.hsd1.ca.comcast.net> has joined #illumos
[23:54:21] *** Yogurt <Yogurt!~Yogurt@c-73-189-45-147.hsd1.ca.comcast.net> has quit IRC (Ping timeout: 265 seconds)
top

   September 20, 2019  
< | 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 | >