Switch to DuckDuckGo Search
   February 14, 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 | >

Toggle Join/Part | bottom
[00:49:35] *** andy_js <andy_js!~andy@> has quit IRC (Quit: andy_js)
[02:06:51] *** sarvet <sarvet!~sarvet@dslb-094-220-192-134.094.220.pools.vodafone-ip.de> has joined #oi-dev
[02:21:48] *** sarvet <sarvet!~sarvet@dslb-094-220-192-134.094.220.pools.vodafone-ip.de> has quit IRC (Quit: sarvet)
[03:00:03] *** v_a_b <v_a_b!~volker@p57A27B7D.dip0.t-ipconnect.de> has quit IRC (Ping timeout: 245 seconds)
[05:30:44] *** tsundoku <tsundoku!~tsundoku@irc.tsundoku.ne.jp> has quit IRC (Quit: 行ってきまーす)
[05:33:10] *** tsundoku <tsundoku!~tsundoku@irc.tsundoku.ne.jp> has joined #oi-dev
[05:45:12] *** sfisque <sfisque!~Adium@74-95-42-38-Oregon.hfc.comcastbusiness.net> has quit IRC (Quit: Leaving.)
[06:17:22] *** x5eb <x5eb!~seb@seb-hpws2.w1.tele.crt1.net> has joined #oi-dev
[06:18:07] *** _0x5eb_ <_0x5eb_!~seb@seb-hpws2.elen.ucl.ac.be> has quit IRC (Ping timeout: 240 seconds)
[06:19:08] *** x5eb is now known as _0x5eb_
[06:35:09] *** tsundoku <tsundoku!~tsundoku@irc.tsundoku.ne.jp> has quit IRC (Ping timeout: 246 seconds)
[06:41:03] *** tsundoku <tsundoku!~tsundoku@irc.tsundoku.ne.jp> has joined #oi-dev
[08:18:34] <leoric> yes, time is precious... lecturing, deploying new services on one job and a lot of routine on another....
[08:19:19] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has quit IRC (Quit: tsoome)
[08:54:52] *** jimklimov <jimklimov!~jimklimov@> has joined #oi-dev
[08:56:00] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has joined #oi-dev
[09:01:25] *** v_a_b <v_a_b!~volker@p57A27B78.dip0.t-ipconnect.de> has joined #oi-dev
[09:26:08] *** jimklimov <jimklimov!~jimklimov@> has quit IRC (Read error: Connection reset by peer)
[09:26:34] *** jimklimov <jimklimov!~jimklimov@> has joined #oi-dev
[09:53:48] *** ngchk1 <ngchk1!~ngchk1@b2b-92-50-91-166.unitymedia.biz> has quit IRC (Ping timeout: 245 seconds)
[09:53:58] *** ngchk1 <ngchk1!~ngchk1@b2b-92-50-91-166.unitymedia.biz> has joined #oi-dev
[10:05:47] *** ptribble <ptribble!~ptribble@cpc92716-cmbg20-2-0-cust138.5-4.cable.virginm.net> has joined #oi-dev
[10:44:16] *** tsundoku <tsundoku!~tsundoku@irc.tsundoku.ne.jp> has quit IRC (Quit: 行ってきまーす)
[10:49:36] *** tsundoku <tsundoku!~tsundoku@irc.tsundoku.ne.jp> has joined #oi-dev
[11:28:02] *** andy_js <andy_js!~andy@> has joined #oi-dev
[11:33:25] <xenol> hehe, leoric how are you doing?
[11:42:57] <jimklimov> xenol: leoric: I am having a fun adventure in the past months, upgrading a very venerable setup (started as SXCE) and finally reached it into current hipster
[11:43:33] <jimklimov> the new prepackaged VirtualBox worked great, turns out my vboxsvc SMF wrapper supports it seamlessly :)
[11:43:53] <xenol> wow, why not just reinstall
[11:44:05] <jimklimov> lots of stuff set up, and it is fun
[11:44:33] <jimklimov> and not sure how the Enterprise Sun software would behave if I ask their wizards to install on something very not Solaris 10 like :)
[11:44:56] <jimklimov> existing bins and configs work "as is" in the everchaning environment, is good enough
[11:46:21] <jimklimov> upgrades from SXCE to oi_151a4 (initially I believe) were indeed a re-installation, had a new tree of rootfs and zoneroots side by side, mounted from another OI over NFS and handpicked the changes and setups and files so it would work
[11:46:30] *** mnowak_ <mnowak_!mnowak_@nat/suse/x-bxflkkqmewftqaeb> has quit IRC (Quit: Leaving)
[11:46:45] <jimklimov> then it was pkg update to different oi-dev versions, stalled at oi_151a8 some 6 years ago now :)
[11:48:03] <jimklimov> and a jump to hipster-2015 this winter was almost like reinstalling for many of the zones, a lot of stuff changed (crypto defaults, apache configs, etc etc etc) so it took several weeks to sort out invisibly for users - the box is still in production
[11:49:49] <jimklimov> one thing I had hit now in this final step from hipster-2015 to hipster was that linked-image zones introduced between oi_151a8 and hipster-2015 caused running of a lot of "pkg" copies when doing an upgrade with local zones present (and their "ipkg" becoming a linked one)
[11:50:02] <jimklimov> this crashed the Thumper several times due to memory usage
[11:50:29] *** mnowak_ <mnowak_!mnowak_@nat/suse/x-vdnvpfctuoejovzi> has joined #oi-dev
[11:50:54] <jimklimov> had to substitute a dummy /etc/zones/index in the new BE so it would think to upgrade just the GZ image, and then added the local zones back and updated their zoneroots
[11:52:12] <jimklimov> seems that in current hipster this is resolved; at least though adding packages does look into the zones and does some linked action evaluations and executions (no idea what it does though - not installing newly added software there, for sure... maybe bookkeeping?) -- but this does not pressurize the memory or cause crashes
[11:52:55] <jimklimov> also had to do a number of ugly hacks via envvars due to X11 progs not seeing libGL.so
[11:53:39] <jimklimov> in the end it seems "service/opengl/ogl-select" was not installed nor pulled by anybody's dependency
[11:57:31] <jimklimov> so I do have a web of symlinks going from /usr/lib/libGL.so through many hops to /var/run/opengl/lib/libGL.so.1 - but there's nothing there :)
[11:58:17] <jimklimov> but after adding the package and enabling ogl-select service, it all got well
[11:59:09] <andyf> jimklimov - some packages are marked that they need to be in-sync between the GZ image and other linked images, that is probably what it is doing
[11:59:19] <jimklimov> probably
[11:59:27] <jimklimov> and thanks for the vbox work :)
[11:59:35] <jimklimov> much appreciated, can confirm it flies well :)
[12:03:11] <andyf> great.. it will need some more changes later on when bhyve gets upstreamed, but those changes are already in OmniOS :)
[12:03:47] <jimklimov> :)
[12:04:55] <andyf> hopefully DHCP won't :D
[12:06:10] <jimklimov> one other issue I hit was... not sure if a packaging or operator error - from legacy times, there was a big LD_LIBRARY_PATH preset to help all the differently packaged (SUNW, SFW, SMC, ...) softwarez work. In particular, this precluded some OI binaries to run, because the libstdc++.so from gcc-6-runtime was not among those paths, but another (older) binary was => unresolved symbols and generall messy hell :)
[12:07:16] <jimklimov> I wonder whether OS-bundled binaries from the same distro revision that are supposed to play together should refer to full pathnames of such libs?
[12:09:44] <jimklimov> and finally one more caveat was that "virtualbox-additions" differ from the Sun/Oracle EULA additions
[12:10:30] <jimklimov> so my VMs that were configured to use usb2/3, an RDP console as fallback, etc. did not start until I added the downloaded 5.2.24 EULA additions
[12:13:12] <jimklimov> ah... old me
[12:13:24] <jimklimov> The EULA'd pack is "Extensions"
[12:13:31] <jimklimov> and "additions" are for a gues
[12:13:32] <jimklimov> t
[12:13:47] <jimklimov> might say that a bit better in the package descritpion though :-D
[12:14:59] <jimklimov> just in case, the build in OI does not package the VNC-based VRDE remote console?
[12:15:36] <jimklimov> I believe there is one in sources, but Oracle does not build it in their packages - in favor of having the RDP one from EULA'd extension pack
[12:18:24] *** jimklimov <jimklimov!~jimklimov@> has quit IRC (Read error: Connection reset by peer)
[12:19:00] *** jimklimov <jimklimov!~jimklimov@> has joined #oi-dev
[12:26:19] *** jimklimov <jimklimov!~jimklimov@> has quit IRC (Read error: Connection reset by peer)
[12:26:54] *** jimklimov <jimklimov!~jimklimov@> has joined #oi-dev
[12:28:17] *** jimklimov <jimklimov!~jimklimov@> has quit IRC (Read error: Connection reset by peer)
[12:28:24] *** jimklimov <jimklimov!~jimklimov@> has joined #oi-dev
[12:29:55] *** jimklimov <jimklimov!~jimklimov@> has quit IRC (Read error: Connection reset by peer)
[12:30:24] *** jimklimov <jimklimov!~jimklimov@> has joined #oi-dev
[12:37:12] *** jimklimov <jimklimov!~jimklimov@> has quit IRC (Read error: Connection reset by peer)
[12:37:39] *** jimklimov <jimklimov!~jimklimov@> has joined #oi-dev
[12:39:21] *** jimklimov <jimklimov!~jimklimov@> has quit IRC (Read error: Connection reset by peer)
[12:39:39] *** jimklimov <jimklimov!~jimklimov@> has joined #oi-dev
[12:40:25] *** jimklimov <jimklimov!~jimklimov@> has quit IRC (Read error: Connection reset by peer)
[12:40:40] *** jimklimov <jimklimov!~jimklimov@> has joined #oi-dev
[12:41:50] <jollyd> BE promotion failed.
[12:41:52] <jollyd> gahhh
[12:47:00] *** jimklimov <jimklimov!~jimklimov@> has quit IRC (Read error: Connection reset by peer)
[12:47:12] *** jimklimov <jimklimov!~jimklimov@> has joined #oi-dev
[13:12:05] <jollyd> even after zfs promote the dataset is considered as a clone
[13:12:08] <jollyd> ...
[13:15:21] <tsoome> hm?
[13:17:35] <jollyd> tsoome: I tried to follow the documentation to clone a zone
[13:18:00] <jollyd> I ended up with a zone that would not attach or freeze and a rather nice mess
[13:18:23] <tsoome> manual cloning?
[13:18:44] <andyf> A linked image zone? (ipkg)
[13:19:38] <jollyd> a nlipkg zone
[13:20:03] <jollyd> I followed the tutorial from the solaris docs; it does not work
[13:20:19] <jollyd> ended up with a half cloned zone
[13:20:30] <andyf> It won't work properly for any zone brand which has its BEs managed by libbe
[13:20:32] <jollyd> cloned manually the dataset according to another doc
[13:20:52] <tsoome> was it with zoneadm detach?
[13:21:03] <tsoome> it done*
[13:21:20] <andyf> and OI has this patch... upstream_merge/2019021401
[13:21:22] <jollyd> zone was detached yes
[13:21:25] <andyf> er, https://github.com/OpenIndiana/oi-userland/blob/oi/hipster/components/openindiana/illumos-gate/patches/0001-XXX-Add-nlipkg-zone-brand.patch
[13:22:39] <andyf> The problem is that it's only detached in the current BE.. if you were to roll back to an older one it would be there
[13:54:16] <jollyd> andyf: thanks for the explanation, I guess I should reinstall the zone from scratch
[13:54:51] <jollyd> unfortunately this is a pain and I do not have time for that so it will wait until next month
[13:54:53] <andyf> you can probably recover if necessary (by manually cleaning up datasets etc.)
[13:55:40] <jollyd> andy: I am not sure how the parentbe hash is set in the zfs metadata
[13:56:44] <andyf> uuid=`zfs get -H -o value org.opensolaris.libbe:uuid /`
[13:56:52] <andyf> zfs set org.opensolaris.libbe:parentbe=$uuid $zbe
[13:58:01] *** andy_js <andy_js!~andy@> has quit IRC (Ping timeout: 246 seconds)
[13:59:49] <jollyd> andyf: so just to be sure parentbe is really the uuid of the GZ?
[14:00:01] <andyf> of the active BE in the GZ
[14:00:21] <andyf> that's how zone BEs are rolled back and forward along with the GZ BE
[14:00:23] <jollyd> ok that's consistent
[14:01:38] <andyf> A zone's BE is linked to a GZ BE.. it's not something I really like but I understand why it is that way
[14:03:48] <jollyd> andyf: in the end turned out that my current BE was set as a clone of a zfs dataset from the original BE, and moreover the be data was missing
[14:04:07] <jollyd> andyf: it seems that your steps helped, thanks
[14:10:28] <andyf> Great. Cloning is one of the things on my list to look at one day..
[14:30:45] <jollyd> erf now at least a dozen packages failing in the gcc8 build zone...
[15:09:25] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has quit IRC (Ping timeout: 246 seconds)
[15:41:30] <jimklimov> jollyd: from my practice, clones (or actually snapshots before them) do not inherit current properties of the snapshotted dataset
[15:42:21] <jimklimov> so I have to reapply properties after cloning, there's lots of generations of such few-liners in my different ZFS/zone-related scripts :)
[15:44:54] <jimklimov> e.g. https://github.com/jimklimov/illumos-splitroot-scripts/blob/master/bin/beadm-clone.sh#L46
[15:58:13] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has joined #oi-dev
[16:01:04] <leoric> mno-hime: I've rechecked virtualbox. Builds fine here
[16:01:37] <mnowak_> leoric, reproducibly fails for me. must be some missing dep
[16:02:04] <mnowak_> 64-bit .so are built, though
[16:05:44] <leoric> mnowak_: do you build inside virtualbox ?
[16:06:04] <mnowak_> leoric, nope, illumos zone
[16:06:25] <leoric> is 32-bit libGL usable?
[16:09:39] <mnowak_> leoric, I am not at that system right now. may check if the pkg is not broken
[16:17:33] <jimklimov> SPRB selftest:
[16:17:33] <jimklimov> \"detailedMessage\":\"It is likely that your Blood Alcohol Content (BAC) exceeds the threshold for making sensible decisions regarding pull requests. Please try again on Monday.\"
[16:17:50] <jimklimov> (a jenkins plugin)
[16:17:55] <jimklimov> lol :)
[16:21:14] <jollyd> why updating the zone would panic the whole system remains a mystery...
[16:21:33] <jimklimov> to my recent practice, PKG eats alot :)
[16:21:51] <jimklimov> so when parallelized ovr 30 zones, it was a big lot :)
[16:22:55] <jollyd> 1 zone on an idle machine with 20 cores and 48GB RAM :)
[16:28:33] *** jimklimov1 <jimklimov1!~jimklimov@> has joined #oi-dev
[16:30:16] *** jimklimov <jimklimov!~jimklimov@> has quit IRC (Ping timeout: 246 seconds)
[16:32:25] <jimklimov1> ETOOMANYRESOURCES ;D
[17:29:00] *** jimklimov1 <jimklimov1!~jimklimov@> has quit IRC (Ping timeout: 250 seconds)
[19:25:54] *** jimklimov <jimklimov!~jimklimov@ip-86-49-250-123.net.upcbroadband.cz> has joined #oi-dev
[20:33:23] <tomww> :)
[23:15:37] *** ptribble <ptribble!~ptribble@cpc92716-cmbg20-2-0-cust138.5-4.cable.virginm.net> has quit IRC (Quit: Leaving)
[23:35:24] *** alanc <alanc!~alanc@> has quit IRC (Remote host closed the connection)
[23:35:51] *** alanc <alanc!~alanc@> has joined #oi-dev
[23:35:51] *** ChanServ sets mode: +o alanc
[23:41:15] *** js <js!~js@heap.zone> has quit IRC (Quit: ZNC 1.7.1 - https://znc.in)
[23:43:21] *** js <js!~js@heap.zone> has joined #oi-dev

   February 14, 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 | >