Switch to DuckDuckGo Search
   August 3, 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 | 31 | >

Toggle Join/Part | bottom
[00:02:28] <KungFuJesus> pkg's terminology for updates makes no sense
[00:02:42] <KungFuJesus> "Updating modified actions"
[00:02:45] <KungFuJesus> wtf does that mean?
[00:03:15] <KungFuJesus> reticulating splines
[00:08:43] <jollyd> sobolev embedding
[00:20:13] *** jollyd <jollyd!~alarcher@lfbn-nic-1-115-172.w2-15.abo.wanadoo.fr> has quit IRC (Quit: Leaving.)
[00:21:36] *** alanc <alanc!~alanc@129.157.69.53> has quit IRC (Remote host closed the connection)
[00:22:03] *** alanc <alanc!~alanc@129.157.69.53> has joined #illumos
[00:22:17] *** neirac <neirac!~cneira@190.162.109.53> has joined #illumos
[00:25:52] *** kkantor <kkantor!~kkantor@c-24-118-59-107.hsd1.mn.comcast.net> has quit IRC (Ping timeout: 245 seconds)
[00:27:46] *** ptribble <ptribble!~ptribble@cpc92716-cmbg20-2-0-cust138.5-4.cable.virginm.net> has quit IRC (Quit: Leaving)
[00:28:30] <LeftWing> KungFuJesus: I believe a package is a collection of "actions", each of which has a type, like "file" or "user"
[00:29:16] <KungFuJesus> I guess there really isn't a better word. I don't think "transaction" would have been any less vague or fitting
[00:29:31] <LeftWing> If you look at "man -s 5 pkg" you'll see a description of the terminology
[00:29:57] <KungFuJesus> update didn't resolve the NFS issue, but given then illumos-gate as of lately, I'm think it's going to solve a whole buttload of smbd issues
[00:30:09] <KungFuJesus> why did it take so long for that stuff to get merged from Nexenta?
[00:30:15] <LeftWing> They have been busy.
[00:30:35] <KungFuJesus> I mean those things go back way before the acquisition
[00:30:37] <KungFuJesus> like, 2014
[00:30:45] <LeftWing> We can only integrate things that somebody actually tries to integrate
[00:31:02] <MilkmanDan> pkg-do needful --please
[00:31:19] <KungFuJesus> I think I was bit by the exhausted auth sockets bug at least once on one server
[00:31:38] <neirac> jbk hi, have you started working on tickless ? I think I could resume work on that one hopefully next week
[00:31:42] <KungFuJesus> I'm hoping the oplock stuff is fixed, because I had to specifically disable that to make MS office documents not take 2 minutes to open
[00:32:23] <KungFuJesus> I've definitely seen the user can't delete a directory bug
[00:32:39] <KungFuJesus> I've seen the can't view ACLs from windows if they don't have a mapping to POSIX bug
[00:32:55] <jbk> neirac: not yet
[00:33:13] <jbk> i've got plenty to keep me busy :)
[00:34:11] *** andy_js <andy_js!~andy@94.12.192.123> has quit IRC (Quit: andy_js)
[00:35:26] <neirac> jbk, ok I'll try to take latest smartos and apply the patches again I won't delete the repo again :)
[00:44:58] <gitomat> [illumos-gate] 8903 IPPF paths have bad logic -- Michael McConville <mmcco at mykolab dot com>
[00:46:41] *** aszeszo <aszeszo!~aszeszo@unaffiliated/aszeszo> has joined #illumos
[01:05:57] <jbk> ok.. if you have any questions, just let us know
[01:12:10] <richlowe> the only words pkg could use other than actions would be "stuff" or "things" or such.
[01:12:16] <richlowe> which I mean, I'm not saying that'd be worse
[01:30:35] *** neirac <neirac!~cneira@190.162.109.53> has quit IRC (Remote host closed the connection)
[02:08:31] *** amrfrsh <amrfrsh!~Thunderbi@190.2.145.106> has quit IRC (Remote host closed the connection)
[02:21:10] <LeftWing> tsoome: Is there a plan for pulling loader updates from FreeBSD? An OmniOS ISO does not work under illumos bhyve UEFI, but a FreeBSD 12-RELEASE ISO does.
[02:22:05] *** amrfrsh <amrfrsh!~Thunderbi@190.2.145.106> has joined #illumos
[02:22:19] <LeftWing> It fails super early, before it gets to loader.conf even
[02:25:01] <LeftWing> https://pasteboard.co/IqU2pcZ.png
[02:35:33] *** aszeszo <aszeszo!~aszeszo@unaffiliated/aszeszo> has quit IRC (Remote host closed the connection)
[02:35:59] *** aszeszo <aszeszo!~aszeszo@unaffiliated/aszeszo> has joined #illumos
[02:39:31] <richlowe> I assume he slowed down during his null purge
[02:54:20] *** aszeszo <aszeszo!~aszeszo@unaffiliated/aszeszo> has quit IRC (Quit: aszeszo)
[03:05:36] *** freakazoid0223 <freakazoid0223!~matt@pool-96-227-98-169.phlapa.fios.verizon.net> has left #illumos ("Leaving")
[03:12:54] *** jelmd <jelmd!~purple@p5DC0433D.dip0.t-ipconnect.de> has quit IRC (Ping timeout: 244 seconds)
[03:14:59] *** kkantor <kkantor!~kkantor@c-24-118-59-107.hsd1.mn.comcast.net> has joined #illumos
[03:16:15] *** jelmd <jelmd!~purple@p5DC0433D.dip0.t-ipconnect.de> has joined #illumos
[03:19:31] *** kkantor <kkantor!~kkantor@c-24-118-59-107.hsd1.mn.comcast.net> has quit IRC (Ping timeout: 248 seconds)
[03:20:00] *** ed209 <ed209!~ed209@165.225.128.67> has quit IRC (Remote host closed the connection)
[03:20:08] *** ed209 <ed209!~ed209@165.225.128.67> has joined #illumos
[03:26:42] *** abf_ <abf_!~abf@2601:404:c101:234::cdb> has joined #illumos
[03:30:29] <LeftWing> Yeah. I was just trying to reason about the upstream source and it seems like there has been... refactored...
[03:30:39] <LeftWing> *like it
[03:51:28] *** jcea <jcea!~Thunderbi@2001:41d0:1:8a82:7670:6e00:7670:6e00> has quit IRC (Quit: jcea)
[03:53:29] *** dlg <dlg!~dlg@toy.eait.uq.edu.au> has quit IRC (Read error: Connection reset by peer)
[03:53:36] *** dlg <dlg!~dlg@toy.eait.uq.edu.au> has joined #illumos
[03:53:53] *** dlg <dlg!~dlg@toy.eait.uq.edu.au> has quit IRC (Read error: Connection reset by peer)
[03:58:38] *** dlg <dlg!~dlg@toy.eait.uq.edu.au> has joined #illumos
[04:16:09] <jbk> the dramatic pause suggests a degree of understatement :P
[04:24:23] *** dlg <dlg!~dlg@toy.eait.uq.edu.au> has quit IRC (Read error: Connection reset by peer)
[04:24:31] *** dlg <dlg!~dlg@toy.eait.uq.edu.au> has joined #illumos
[04:24:50] *** dlg <dlg!~dlg@toy.eait.uq.edu.au> has quit IRC (Read error: Connection reset by peer)
[04:29:32] *** dlg <dlg!~dlg@toy.eait.uq.edu.au> has joined #illumos
[04:45:09] <richlowe> # Use $@ instead of the "unreliable" $*
[04:53:54] <LeftWing> Eh?!
[04:54:10] <LeftWing> It reliably does something, just not the same something, surely
[04:56:19] <richlowe> also, why did none of our EOF happy brethren get to printmgr?
[04:58:12] <richlowe> sometimes, the hardest part of the job is staying professional in code comments.
[05:01:23] <LeftWing> Seems like it could probably go. :P
[05:01:47] <richlowe> LeftWing: volunteering?
[05:02:06] <LeftWing> +++NO CARRIER
[05:06:20] *** jubal <jubal!~jubal@226-5-237-24.gci.net> has joined #illumos
[05:09:29] <jbk> ATDT *69
[05:12:35] *** kkantor <kkantor!~kkantor@c-24-118-59-107.hsd1.mn.comcast.net> has joined #illumos
[05:33:46] <richlowe> I'm so confused
[05:36:35] *** dlg <dlg!~dlg@toy.eait.uq.edu.au> has quit IRC (Read error: Connection reset by peer)
[05:36:44] *** dlg <dlg!~dlg@toy.eait.uq.edu.au> has joined #illumos
[05:50:02] *** dlg <dlg!~dlg@toy.eait.uq.edu.au> has quit IRC (Ping timeout: 245 seconds)
[05:51:32] *** jubal <jubal!~jubal@226-5-237-24.gci.net> has quit IRC (Quit: jubal)
[05:54:37] *** kkantor <kkantor!~kkantor@c-24-118-59-107.hsd1.mn.comcast.net> has quit IRC (Ping timeout: 245 seconds)
[05:55:23] *** dlg <dlg!~dlg@toy.eait.uq.edu.au> has joined #illumos
[06:04:48] *** mgerdts <mgerdts!~textual@2600-6c44-0c7f-ec89-5191-2305-ce54-92b3.dhcp6.chtrptr.net> has quit IRC (Quit: My MacBook has gone to sleep. ZZZzzz…)
[06:09:09] *** dlg_ <dlg_!~dlg@toy.eait.uq.edu.au> has joined #illumos
[06:12:07] *** dlg <dlg!~dlg@toy.eait.uq.edu.au> has quit IRC (Ping timeout: 245 seconds)
[06:16:52] *** dlg <dlg!~dlg@toy.eait.uq.edu.au> has joined #illumos
[06:20:03] *** dlg_ <dlg_!~dlg@toy.eait.uq.edu.au> has quit IRC (Ping timeout: 245 seconds)
[06:44:00] *** jubal <jubal!~jubal@226-5-237-24.gci.net> has joined #illumos
[06:52:20] *** jelmd <jelmd!~purple@p5DC0433D.dip0.t-ipconnect.de> has quit IRC (Ping timeout: 258 seconds)
[07:03:22] *** baojg_ <baojg_!~baojg@162.243.44.213> has quit IRC (Ping timeout: 245 seconds)
[07:10:52] *** baojg <baojg!~baojg@162.243.44.213> has joined #illumos
[07:52:13] <tsoome> LeftWing: is that omnios iso up to date with gate?
[07:53:35] <LeftWing> tsoome: I've tried both a stock one and one that I've built from recent bits (I'm trying to test the virtio stuff)
[07:53:58] <LeftWing> I just adjusted the build process to disable creating the "hybrid" ISO (with the fdisk/installboot business)
[07:54:05] <LeftWing> That allowed it to work once
[07:54:20] <LeftWing> But then, when rebooting, it failed in the way it failed with the hybrid
[07:54:33] <LeftWing> I suspect it's the existence of more than one ESP that's causing the issue...
[07:55:32] <tsoome> hm, we do not have 2 esp:)
[07:56:26] <tsoome> which bhyve is that?
[07:56:30] <LeftWing> The SmartOS bhyve
[07:56:31] <tsoome> smartos?
[07:56:33] <tsoome> ok
[07:56:51] <tsoome> could you make firmware file available for me?
[07:56:54] <LeftWing> Sure
[07:56:57] <LeftWing> One sec
[07:57:30] <LeftWing> https://us-east.manta.joyent.com/jmc/public/tmp/uefi-rom.bin
[07:57:57] <tsoome> I only have fbsd bhyve atm
[07:58:06] <tsoome> but lets see:)
[07:59:29] <tsoome> there was a bug in loader where it did fail to find handle for boot device, but I thought we have that fix already…
[08:00:00] <tsoome> might be some other stupid corner case…
[08:25:34] <tsoome> seems so, at least I can replicate the issue often enough.
[08:26:23] <LeftWing> tsoome: So it looks like in efipart_updatecd() we actually find _five_ handles of type MEDIA_DEVICE_PATH subtype MEDIA_CDROM_DP
[08:26:43] <LeftWing> One of them has an alias that matches the device we eventually try to find
[08:27:12] <LeftWing> But efipart_cdinfo_add() keeps overwriting the same cdinfo entry
[08:27:42] <tsoome> yea, but the issue is not limited to cd, so it has to be separate issue:)
[08:27:50] <tsoome> Load Device: PciRoot(0x0)/Pci(0x5,0x0)/HD(1,GPT,199AB29A-F956-E544-BF86-DE472FE8F1E4,0x100,0x80000)
[08:28:24] <LeftWing> It looks like efipart_hdinfo_add() has the same kind of logic in it
[08:30:05] <LeftWing> So, if I comment out this: https://github.com/illumos/illumos-gate/blob/master/usr/src/boot/sys/boot/efi/libefi/efipart.c#L384-L388
[08:30:27] <LeftWing> It starts working, using "cd1" (instead of what I imagine is the expected "cd0") for currdev
[08:32:16] <tsoome> do you get the device names in both cases?
[08:32:51] <LeftWing> How do you mean
[08:33:14] <tsoome> lsdev -v
[08:33:25] <tsoome> device path, i meant
[08:34:22] <tsoome> ah, i see the error for my case
[08:34:44] <tsoome> I have: Load Device: PciRoot(0x0)/Pci(0x5,0x0)/HD(1,GPT,199AB29A-F956-E544-BF86-DE472FE8F1E4,0x100,0x80000)
[08:34:54] <LeftWing> https://sysmgr.org/~leftwing/tmp/cds.png
[08:34:57] <tsoome> but the device is listed: PciRoot(0x0)/Pci(0x5,0x0)
[08:41:53] <LeftWing> It seems like efi_devpath_match() has changed a bit
[08:46:52] <LeftWing> Though perhaps not in a way that matters
[08:47:48] <tsoome> there are bugs for sure. first of all, i do not want to see PciRoot(0x0)/Pci(0x5,0x0) for disk path
[08:48:08] <LeftWing> Bugs in the UEFI ROM?
[08:48:19] <tsoome> no, loader code
[08:48:22] <LeftWing> Ahh OK
[08:49:10] <tsoome> secondly, if we have multiple handles for same device, the rest should be collected in aliases..
[08:49:23] <tsoome> or like that:D
[08:57:07] <tsoome> https://paste.ec/paste/fKtGyFW3#QL4XO0krBayeuTnFwPDUaniZDqnNEPh4DKX80oY9z2U
[08:57:38] <tsoome> even on vmware fusion there are some dubious ones — the ata cdrom and disk for example.
[08:57:50] <tsoome> somehow I have missed those…
[09:01:15] *** phyre__ <phyre__!~phyre___@165.54.223.87.dynamic.jazztel.es> has joined #illumos
[09:12:49] <tsoome> btw, can we get bhyve on gate?:P
[09:13:47] <LeftWing> Not tonight, but yes, I think that's a good idea.
[09:16:07] <tsoome> btw, this issue does depend on firmware. the freebsd bhyve-uefi.fd does boot
[09:16:38] <tsoome> that is, we can find the handle there
[09:21:10] <LeftWing> Right, for the CDROM case it seems like it depends on the order of the handles
[09:34:07] <LeftWing> Of the five cdrom handles, two of them appear to represent el torito boot things
[09:34:30] <LeftWing> (If I remove the "cdboot" bit from mkisofs, there is then only four of them)
[09:42:06] <LeftWing> https://sysmgr.org/~leftwing/tmp/cds2.png -- it looks like we really do need to store more than one pd_alias value
[09:42:46] <tsoome> yes, we need to store all aliases
[09:43:54] <LeftWing> Alright, well, I need to get some sleep!
[09:46:15] <tsoome> and the problem with lsdev -v output is that it is possible we do have partition node listed directly on pci node, without having “container” for hd. so when we chop off last node, we do get pci node and not the disk...
[09:52:14] *** aszeszo <aszeszo!~aszeszo@unaffiliated/aszeszo> has joined #illumos
[10:00:36] <tsoome> ok, my boot from hdd failed when second disk did not had partition table.
[10:02:02] *** aszeszo <aszeszo!~aszeszo@unaffiliated/aszeszo> has quit IRC (Quit: aszeszo)
[10:12:11] *** Diagon <Diagon!DiagonalAr@gateway/vpn/protonvpn/diagonalarg> has joined #illumos
[10:33:25] *** andy_js <andy_js!~andy@94.12.192.123> has joined #illumos
[10:43:41] *** Diagon <Diagon!DiagonalAr@gateway/vpn/protonvpn/diagonalarg> has quit IRC (Quit: Leaving)
[10:52:57] *** phyre__ <phyre__!~phyre___@165.54.223.87.dynamic.jazztel.es> has quit IRC (Quit: Saliendo)
[10:53:12] *** phyre__ <phyre__!~phyre___@165.54.223.87.dynamic.jazztel.es> has joined #illumos
[10:53:35] *** aszeszo <aszeszo!~aszeszo@unaffiliated/aszeszo> has joined #illumos
[11:09:20] *** sjorge <sjorge!~sjorge@unaffiliated/sjorge> has quit IRC (Quit: 410 Gone)
[11:23:32] *** sjorge <sjorge!~sjorge@unaffiliated/sjorge> has joined #illumos
[11:43:08] *** pvs <pvs!sid19318@lopsa/member/pvs> has quit IRC (Ping timeout: 252 seconds)
[11:43:20] *** pvs <pvs!sid19318@lopsa/member/pvs> has joined #illumos
[11:43:30] *** Smithx10 <Smithx10!sid243404@gateway/web/irccloud.com/x-choflpyqtlvzlcdu> has quit IRC (Ping timeout: 252 seconds)
[11:43:52] *** Tsesarevich <Tsesarevich!Tsesarevic@fluxbuntu/founder/joejaxx> has quit IRC (Ping timeout: 252 seconds)
[11:44:14] *** bdha <bdha!~bdha@2001:19f0:5:5568:5400:2ff:fe25:993e> has quit IRC (Ping timeout: 252 seconds)
[11:44:30] *** bdha <bdha!~bdha@45.77.108.166> has joined #illumos
[11:44:58] *** arekinath <arekinath!~arekinath@pdpc/supporter/professional/arekinath> has quit IRC (Ping timeout: 252 seconds)
[11:44:58] *** trisk <trisk!~trisk@2601:196:4700:3f0b:8:20ff:fe3e:890e> has quit IRC (Ping timeout: 252 seconds)
[11:47:00] *** Tsesarevich <Tsesarevich!Tsesarevic@fluxbuntu/founder/joejaxx> has joined #illumos
[11:47:07] *** arekinath <arekinath!~arekinath@pdpc/supporter/professional/arekinath> has joined #illumos
[11:47:51] *** trisk <trisk!~trisk@2601:196:4700:3f0b:8:20ff:fe3e:890e> has joined #illumos
[11:58:40] *** Smithx10 <Smithx10!sid243404@gateway/web/irccloud.com/x-lhnzqybpwghnbpmu> has joined #illumos
[12:13:15] *** jollyd <jollyd!~alarcher@lfbn-nic-1-115-172.w2-15.abo.wanadoo.fr> has joined #illumos
[12:17:52] *** jollyd <jollyd!~alarcher@lfbn-nic-1-115-172.w2-15.abo.wanadoo.fr> has quit IRC (Ping timeout: 268 seconds)
[12:20:00] *** 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
[12:38:09] *** BH23 <BH23!~BH23@santoroj.plus.com> has joined #illumos
[12:46:47] *** aszeszo_ <aszeszo_!~aszeszo@unaffiliated/aszeszo> has joined #illumos
[12:48:27] *** aszeszo <aszeszo!~aszeszo@unaffiliated/aszeszo> has quit IRC (Ping timeout: 258 seconds)
[13:25:04] *** BH23 <BH23!~BH23@santoroj.plus.com> has quit IRC (Ping timeout: 272 seconds)
[13:37:16] *** sjorge <sjorge!~sjorge@unaffiliated/sjorge> has quit IRC (Remote host closed the connection)
[13:38:21] *** nexgen <nexgen!~nexgen@144.172.68.147> has joined #illumos
[13:41:30] *** sjorge <sjorge!~sjorge@unaffiliated/sjorge> has joined #illumos
[14:09:29] *** abf_ <abf_!~abf@2601:404:c101:234::cdb> has quit IRC (Quit: Leaving)
[14:22:59] *** jelmd <jelmd!~purple@p4FDDE0A2.dip0.t-ipconnect.de> has joined #illumos
[15:25:23] *** jollyd <jollyd!~alarcher@37.168.171.7> has joined #illumos
[15:44:24] *** phyre__ <phyre__!~phyre___@165.54.223.87.dynamic.jazztel.es> has quit IRC (Ping timeout: 272 seconds)
[15:45:17] <jollyd> Any version of openmpi since 3.0 has failed on illumos. It turns out that the reason is that pmix (https://github.com/pmix) is broken on illumos.
[15:47:34] <jollyd> Lock file management seems to be the root cause...
[15:48:00] <jollyd> The execution fails at line 143 of https://github.com/pmix/pmix/blob/master/src/mca/gds/ds12/gds_ds12_lock_fcntl.c
[15:48:39] <tsoome> that is, on open() ?
[15:48:45] <jollyd> Yes
[15:48:48] <tsoome> why?
[15:49:35] <jollyd> That's what I am trying to figure out given that the directory for the lock file is created and that no error is thrown after the file creation
[15:49:48] <tsoome> truss does not tell?
[15:51:11] <tsoome> you should see the error from truss or dtrace:)
[15:52:36] <jollyd> tsoome: I am obviously way out of my comfort/knowledge zone which is numerics and parallel computing :) so I'll try to truss first
[15:53:47] <tsoome> normally, open O_RDONLY should fail if either file is not there or it is about permissions
[15:55:03] <jollyd> erf tests are run with perl... I hate that
[15:55:04] <jollyd> ...
[15:55:23] <jollyd> why is there always some perl or python in the way...
[15:55:28] <tsoome> :)
[16:03:19] <jollyd> tsoome: any particular option that would be good to pass to truss?
[16:04:10] <tsoome> it depends, -f if it is child process you want to truss
[16:04:20] <tsoome> so it will follow forks
[16:04:46] <tsoome> and -o truss.txt :D
[16:04:58] <tsoome> so you can grep it later:D
[16:06:22] <jollyd> erf... with truss -f the test passes...
[16:06:48] <tsoome> but without?
[16:08:46] <jollyd> it passes as well
[16:08:52] <jollyd> weird...
[16:09:00] *** kayront <kayront!~kayront@unaffiliated/kayront> has joined #illumos
[16:10:34] <jollyd> same for another test
[16:12:35] <jollyd> ok got it
[16:27:37] <jollyd> tsoome: indeed the lock does not exist when open is called
[16:27:39] <jollyd> ...
[16:28:10] <tsoome> so the actual error did happen before
[16:28:31] <tsoome> as we attempt to open it readonly:)
[16:28:53] <tsoome> unless we should be allowed to fail there.
[16:29:02] <jollyd> yup but I do not see any attempt to create it before...
[16:29:32] <tsoome> no other open for that name?
[16:29:36] <jollyd> I do not think this code is used on Linux since they have another strategy using non-portable pthread extensions
[16:29:38] <jollyd> nop
[16:29:45] <tsoome> ah.
[16:30:04] <tomww> !!
[16:30:18] <tsoome> so there is a chance that this code is just bad:D
[16:32:31] <jollyd> unfortunately most scientific computing code becomes more and more linux specific
[16:32:57] <jollyd> and even with code that used to be portable, support for other systems is removed
[16:38:45] <jollyd> they use this http://man7.org/linux/man-pages/man3/pthread_rwlockattr_getkind_np.3.html
[16:41:40] <jollyd> the truss output is at https://gist.github.com/alarcher/d9ad260c6fa12ccb5e73af406ecc33e2
[16:42:59] <jollyd> that actually means that PMIX_PROC_IS_SERVER(pmix_globals.mypeer) fails for all the MPI process...
[16:43:10] <jollyd> returns false I meant
[17:04:38] *** lystra <lystra!~lystra@d53-64-11-169.nap.wideopenwest.com> has quit IRC (Ping timeout: 245 seconds)
[17:08:10] *** lystra <lystra!~lystra@d53-64-11-169.nap.wideopenwest.com> has joined #illumos
[17:23:31] *** hotbox <hotbox!~hotbox@2001:41d0:fe8f:b70a:20d:b9ff:fe47:7c05> has quit IRC (Ping timeout: 276 seconds)
[17:23:44] *** heroux <heroux!sandroco@gateway/shell/insomnia247/x-pfpxzzjpqwyehavd> has quit IRC (Ping timeout: 268 seconds)
[17:23:56] *** cantstanya <cantstanya!~chatting@gateway/tor-sasl/cantstanya> has quit IRC (Ping timeout: 260 seconds)
[17:23:57] *** hotbox <hotbox!~hotbox@2001:41d0:fe8f:b70a:20d:b9ff:fe47:7c05> has joined #illumos
[17:25:11] *** heroux <heroux!sandroco@gateway/shell/insomnia247/x-wxwsqbbqbxndbnvp> has joined #illumos
[17:28:23] *** cantstanya <cantstanya!~chatting@gateway/tor-sasl/cantstanya> has joined #illumos
[17:32:21] <jollyd> tsoome: I followed the breadcrumbs
[17:33:35] <jollyd> tsoome: turns out we end up at https://github.com/pmix/pmix/blob/master/src/mca/pshmem/mmap/pshmem_mmap.c#L90
[17:37:21] *** m4rley <m4rley!~m4rley@207.148.96.120> has quit IRC (Quit: leaving)
[17:37:39] *** m4rley <m4rley!~m4rley@207.148.96.120> has joined #illumos
[17:38:49] *** cartwright <cartwright!~chatting@gateway/tor-sasl/cantstanya> has joined #illumos
[17:39:09] <jollyd> posix_fallocate returns 22
[17:39:58] <tsoome> 22 is EINVAL
[17:40:33] *** jellydonut <jellydonut!~quassel@s91904421.blix.com> has joined #illumos
[17:41:58] <jollyd> ok... so offset is 0 and size is 4096
[17:42:10] <jollyd> nothing crazy
[17:42:29] <tsoome> or the underlying file system does not support this operation
[17:42:38] <tsoome> pid is on tempfs maybe?
[17:43:01] <tsoome> or something like…
[17:43:51] <jollyd> it tries to create on /tmp
[17:45:27] <jollyd> I did see a fcntl(14, F_ALLOCSP, ..) failure in truss
[17:45:29] <tsoome> try with TMPDIR=${HOME} maybe?
[17:45:30] <jollyd> makes sense
[17:48:24] <tsoome> yes, it seems tmpfs does not support F_ALLOCSP
[17:48:34] <tsoome> time to write one maybe:)
[17:52:24] *** cantstanya <cantstanya!~chatting@gateway/tor-sasl/cantstanya> has quit IRC (Ping timeout: 260 seconds)
[17:56:10] <jollyd> tsoome: it does not help if I set the base path to ~/tmp
[17:56:27] <tsoome> ~ is shell specific
[17:56:51] <tsoome> you need to use $HOME or direct path
[17:57:14] <jollyd> I did write the whole path
[17:57:19] <tsoome> ah.
[17:57:20] <tsoome> hm.
[17:59:17] <jollyd> erf
[17:59:32] <jollyd> the reason was that other processes did not inherit it
[18:00:29] <tsoome> :D
[18:01:06] <jollyd> passing tmp in home one test succeeds
[18:01:46] <tsoome> ok, so you can file an FRE - to have tempfs to support F_ALLOCSP
[18:01:47] <jollyd> but for the test at hand fallocate still fails
[18:01:51] <tsoome> :D
[18:03:38] <jollyd> base_path /export/home/alarcher/tmp/pmix_dstor_ds21_12116
[18:03:38] <jollyd> mmap segment create
[18:03:38] <jollyd> posix_fallocate 14, 0, 4096
[18:03:38] <jollyd> posix_fallocate return 22
[18:03:38] <jollyd> default boo boo
[18:03:43] <jollyd> erf..
[18:03:59] <jollyd> wonder if there is another peculiarity with automounted dirs
[18:04:56] <tsoome> with lofs?
[18:05:11] <tsoome> it may be.
[18:07:07] *** aszeszo <aszeszo!~aszeszo@unaffiliated/aszeszo> has joined #illumos
[18:09:43] <jollyd> same failure on normal fs
[18:10:04] *** aszeszo_ <aszeszo_!~aszeszo@unaffiliated/aszeszo> has quit IRC (Ping timeout: 272 seconds)
[18:17:00] *** Asgaroth <Asgaroth!~Asgaroth@212.2.172.228> has joined #illumos
[18:17:14] <jollyd> 14098/2: open("/scratch/alarcher/tmp/pmix_dstor_ds12_14098/initial-pmix_shared-segment-0", O_RDWR|O_CREAT, 0600) = 14
[18:17:15] <jollyd> 14098/2: fcntl(14, F_ALLOCSP, 0xFE19695C) Err#22 EINVAL
[18:18:25] <jollyd> tsoome: sorry to pester you again... how do I dig further to check what fails in fcntl?
[18:19:14] <tsoome> what file system is used there, zfs?
[18:19:45] <jollyd> yes
[18:19:48] <tsoome> for that file, that is
[18:20:23] <jollyd> zfs everywhere
[18:20:25] <jollyd> :)
[18:21:09] <tsoome> can you try with ufs?
[18:21:36] <tsoome> just to test an idea…
[18:22:16] <tsoome> you can create it on zvol or lofi mapped file.
[18:23:39] <tsoome> as it seems we might not support F_ALLOCSP there..
[18:24:49] <tsoome> zfs_space() is only supporting F_FREESP.
[18:26:14] *** merzo <merzo!~merzo@198-11-133-95.pool.ukrtel.net> has joined #illumos
[18:28:01] <jollyd> https://illumos.topicbox.com/groups/zfs/Tea969450ed27ed13-M2927eedf540196f1a1c876db
[18:28:13] <jollyd> tsoome: is it what you are referring to?
[18:34:32] <tsoome> yes. I was looking into the code, however:D
[18:35:08] <tsoome> anyhow, that code you have, should have fallback for EINVAL:(
[18:38:33] <jollyd> tsoome: should I just ignore if EINVAL?
[18:39:09] <tsoome> am, no it means that logic did fail, so something alternative should be used instead
[18:40:47] <tsoome> perhaps they just wanted to make sure they have that 4k space
[18:40:57] <jollyd> tsoome: it seems FreeBSD ignores EINVAL when posix_fallocate is used on ZFS
[18:41:19] <tsoome> freebsd as?
[18:41:38] <tsoome> you mean, freebsd specific port of that code?
[18:42:57] <jollyd> sorry I misread https://groups.google.com/forum/#!topic/bsdmailinglist/thoBnXPBSRU
[18:43:40] <jollyd> if I naively ignore EINVAL all the pmix tests pass... :S
[18:51:38] <jollyd> probably I should just disable HAVE_POSIX_FALLOCATE at all
[18:51:50] <tsoome> that is likely better option
[18:57:29] <jollyd> if this make openmpi work again on illumos this will be very frustrating
[18:57:30] <jollyd> ...
[19:06:17] *** lystra <lystra!~lystra@d53-64-11-169.nap.wideopenwest.com> has quit IRC (Ping timeout: 245 seconds)
[19:11:30] *** jcea <jcea!~Thunderbi@2001:41d0:1:8a82:7670:6e00:7670:6e00> has joined #illumos
[19:16:44] *** lystra <lystra!~lystra@d53-64-11-169.nap.wideopenwest.com> has joined #illumos
[19:22:08] *** lystra <lystra!~lystra@d53-64-11-169.nap.wideopenwest.com> has quit IRC (Ping timeout: 245 seconds)
[19:23:15] *** jollyd <jollyd!~alarcher@37.168.171.7> has quit IRC (Quit: Leaving.)
[19:23:52] *** lystra <lystra!~lystra@d53-64-11-169.nap.wideopenwest.com> has joined #illumos
[19:30:27] *** MilkmanDan <MilkmanDan!~dan@wilug/expat/MilkmanDan> has quit IRC (Ping timeout: 245 seconds)
[19:31:27] *** MilkmanDan <MilkmanDan!~dan@wilug/expat/MilkmanDan> has joined #illumos
[20:00:51] *** andy_js <andy_js!~andy@94.12.192.123> has quit IRC (Read error: Connection reset by peer)
[20:01:29] *** andy_js <andy_js!~andy@94.12.192.123> has joined #illumos
[20:30:53] *** lystra <lystra!~lystra@d53-64-11-169.nap.wideopenwest.com> has quit IRC (Ping timeout: 245 seconds)
[20:34:28] *** lystra <lystra!~lystra@d53-64-11-169.nap.wideopenwest.com> has joined #illumos
[20:39:50] *** awordnot <awordnot!~awordnot@c-73-210-60-203.hsd1.il.comcast.net> has quit IRC (Ping timeout: 268 seconds)
[20:40:37] *** BH23 <BH23!~BH23@santoroj.plus.com> has joined #illumos
[20:46:36] *** awordnot <awordnot!~awordnot@c-73-210-60-203.hsd1.il.comcast.net> has joined #illumos
[21:28:35] *** BH23 <BH23!~BH23@santoroj.plus.com> has quit IRC (Ping timeout: 248 seconds)
[21:51:09] *** aszeszo <aszeszo!~aszeszo@unaffiliated/aszeszo> has quit IRC (Remote host closed the connection)
[21:51:16] *** aszeszo_ <aszeszo_!~aszeszo@unaffiliated/aszeszo> has joined #illumos
[22:12:38] *** merzo <merzo!~merzo@198-11-133-95.pool.ukrtel.net> has quit IRC (Ping timeout: 272 seconds)
[22:20:00] *** ed209 <ed209!~ed209@165.225.128.67> has quit IRC (Remote host closed the connection)
[22:20:06] *** ed209 <ed209!~ed209@165.225.128.67> has joined #illumos
[23:01:49] *** wilbury <wilbury!otis@m68k.sinclairql.eu> has quit IRC (Ping timeout: 252 seconds)
[23:07:01] *** wilbury <wilbury!otis@m68k.sinclairql.eu> has joined #illumos
[23:11:43] *** wilbury <wilbury!otis@m68k.sinclairql.eu> has quit IRC (Ping timeout: 252 seconds)
[23:25:42] *** jellydonut <jellydonut!~quassel@s91904421.blix.com> has quit IRC (Quit: jellydonut)
[23:27:29] *** jellydonut <jellydonut!~quassel@s91904421.blix.com> has joined #illumos
[23:32:54] *** wilbury <wilbury!otis@m68k.sinclairql.eu> has joined #illumos
[23:57:56] *** andy_js <andy_js!~andy@94.12.192.123> has quit IRC (Quit: andy_js)
top

   August 3, 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 | 31 | >