-
bahamat
jclulow: I looked at moving section 8 to be before section 1, but that program is a C file, and it's got some *weird* preferences in there about decisions it makes.
-
bahamat
Or, before 1B, rather.
-
jclulow
oh?
-
bahamat
Yeah.
-
jclulow
I mean I wrote it, so, happy to answer any questions haha
-
bahamat
It was *not* clear to me at all how I could establish that preference.
-
jclulow
Ah
-
bahamat
Dan looked at it with me and he couldn't figure it out either.
-
jclulow
haha
-
jclulow
I have created a puzzle
-
jclulow
one sec I'll make a diff
-
bahamat
I have a jira for it, let me see if I wrote down what stumped us
-
bahamat
OS-8436
-
jinni
-
bahamat
Looks like I didn't...
-
bahamat
-
jclulow
Yeah based on my comment it's still going to put 1B before 8, sigh
-
bahamat
Yeah.
-
jclulow
Ok, a new program then!!
-
jclulow
one moment
-
bahamat
It was one of those things where I didn't know why *this* program in particular, or *this* behavior. I considered just outright replacing it with a static list, but I didn't feel like I understood why it works the way it does before I started mucking around in there.
-
jclulow
can you gist a $MANIFEST file
-
jclulow
so that I don't have to do a build haha
-
bahamat
I think so.
-
bahamat
It might be a bit stale, but not terribly so.
-
jclulow
that's ok
-
jclulow
as long as it's post manual renumbering
-
bahamat
So, man/manifest, yeah?
-
jclulow
err
-
jclulow
probably the final altogether one
-
jclulow
like the thing that has everything that we give to builder
-
jclulow
assuming we still do that
-
jclulow
The reason it's a program rather than a static file FWIW is that it needs to be a complete list, or man(1) will silently ignore the missing (sub)sections and I didn't want people to have to notice new sections showing up later
-
bahamat
-
jclulow
-
jclulow
that one
-
jclulow
-
bahamat
OK, that's the generated one
-
jclulow
yeah
-
bahamat
-
jclulow
tah!
-
bahamat
Ok, chatgrep to the rescue, yet again. Basically, I did the commit, we realized it didn't work and then were like how tf does this even work anyway? and then yeah, discovering that subsections get stacked with the section period, realized that it wasn't straightforward.
-
bahamat
But something else that I didn't get was that after my change the order was unchanged *at all*.
-
bahamat
It didn't do 1,1c,1b,8,2, even though section 8 had a lower priority..
-
bahamat
It was still 7,8,9.
-
bahamat
But then after re-reading it, I would expect it to be 1m,1,1b, and 3socket,3c,3,4, but that's not the case on those either.
-
bahamat
The main section is *always* first, and ordering seems to only affect subsections.
-
bahamat
I thought about trying to make a pseudo 1.8 section that was all symlinks to section 8 to fool the search order, but then when displayed it would still look like section 8, but I figured that was too much of a hack.
-
jclulow
MANSECTS=1,1c,8,1b,2,3,3c,3avl,3bsm,3c_db,3cfgadm,3commputil,3contract,3cpc,3curses,3dat,3devid,3devinfo,3dlpi,3dns_sd,3elf,3exacct,3ext,3fcoe,3fstyp,3gen,3gss,3jedec,3kstat,3kvm,3ldap,3lgrp,3lib,3m,3mail,3malloc,3mp,3mpapi,3mvec,3nsl,3nvpair,3ofmt,3pam,3papi,3picl,3picltree,3pool,3proc,3project,3resolv,3rpc,3rsm,3sasl,3scf,3sec,3secdb,3sip,3socket,3head,3stmf,3sysevent,3tecla,3utempter,3uuid,3vnd
-
jclulow
,3volmgt,3x11,3xau,3xaw,3xcurses,3xext,3xmu,3xnet,4,4d,4fs,4i,4ipp,4m,4p,5,5b,7,9,9e,9f,9p,9s,
-
jclulow
that looks right I think
-
jclulow
-
bahamat
Awesome
-
bahamat
I'll work on getting that merged tomorrow.
-
bahamat
huy: You've probably got something either in an image or a vm that the web ui isn't able to parse properly. If you have any images that you obtained from a source other than images.smartos.org, that's a likely culprit. If not, then it's probably an instance.
-
tozhu
bahamat: I would like to ask a question regarding how to re-compile uuid.node, there is my log, not sure if could help to take a look?
pastebin.com/7eyAU9mJ how to fix the uuid.node ?
-
bahamat
tozhu: That doesn't look like it's something we built.
-
bahamat
tozhu: But it seems pretty straightforward that you've got a 64-bit thing trying to dynamically link with a 32-bit thing.
-
bahamat
Make both of them the same bitness and that should resolve your issue.
-
bahamat
node-libuuid (which is what it looks like you're using? But I'm not entirely sure), seems to support both 32 and 64-bit. So it should just be a matter of passing the right flags to it.
-
bahamat
But neither node-libuuid nor rdsagent comes from us, so I'm not the best person to ask.
-
tozhu
yes, it was we built
-
tozhu
bahamat, may I know how to you build it?
-
tozhu
just run ‘npm build’ ?
-
tozhu
bahamat: Thank you for your answer
-
tozhu
may I know how to you compile the uuid.node?
-
bahamat
I don't.
-
bahamat
There are some places where we use it (
hound.tritondatacenter.com/?q=libuu…id&i=nope&files=package.json&repos=), but there's literally no special handling for it. Just `npm install`
-
tozhu
okay, thank you very much
-
huy
bahamat: does it matter if it’s running or not ?
-
bahamat
huy: No, it reads the JSON so that it can report the vm properties.
-
bahamat
An easier way to narrow it down without having to destroy them is to comment them out of /etc/zones/index and see if the ui has a better time of it.
-
bahamat
If you can figure out which one(s) cause the error then we can almost certainly figure out what it is getting tripped up on.
-
huy
It’s probably a docker image I’ve been trying in the past
-
bahamat
Well, like I said, if you can identify it, we can fix it.
-
huy
bahamat: it was a bhyve instance but I just deleted it, can't tell what was causing the problem sorry
-
tozhu
hello all, where could get the correct signature? I hit error like this:
pastebin.com/YxcZNz0q
-
jperkin
tozhu: what set are you upgrading from and to?
-
tozhu
I have upgrade from 2021Q4 to trunk
-
jperkin
ok, if you follow the "Standard (upgrade)" instructions at
pkgsrc.smartos.org/install-on-illumos it should work
-
tozhu
-
tozhu
last week, it works; but today, my colleagur, set the URL to 2023Q4, did pkgin up
-
tozhu
then all things don’t work
-
tozhu
so, I don’t know hot fix it,
-
jperkin
you can't just update the URL, you need the upgrade kit as that contains the new pgp key
-
jperkin
you can either use the trunk upgrade kit and then fixup the URLs, or use
pkgsrc.smartos.org/packages/SmartOS…tstrap-2023Q4-x86_64-upgrade.tar.gz
-
tozhu
okay, thank you very much
-
tozhu
I’ll try
-
jperkin
I don't publish the LTS upgrade kits on the website as it would complicate things and it's rare that someone would want to use them, but they're all there if you want to just browse around
-
tozhu
if I build everything from source, the bootstrap and its tool is been built automatically?
-
jperkin
if you want to bootstrap everything yourself then you wouldn't use any of the kits, though you would need at least a compiler from somewhere
-
tozhu
is there some docs to guide?
-
tozhu
I wan’t mirror a pkgsrc, seems the software can by synced, but the bootstrap is a link, can’t be sync-ed
-
jperkin
there's the pkgsrc guide at
netbsd.org/docs/pkgsrc
-
tozhu
jperkin, Thank you very much for the answer, Thank you
-
jperkin
shouldn't be any links, they're just regular files?
-
jperkin
also 99% of the stuff I've written is up at
github.com/TritonDataCenter/pkgbuild so you can reuse any of that
-
tozhu
no, I just curl on the machine to get the bootstrapxxx.gz file that I use rsync from pkgsrc.smartos.org the file is only 2k, but I use broswer to get the file, then it’s a big file
-
tozhu
yes, I have read that, it’s real usful
-
tozhu
have you complete the docs that guide how to build all package in parall? :-)
-
jperkin
the sizes via rsync look ok to me, e.g.
-
jperkin
-
jperkin
-rw-r--r-- 18202492 2024/01/16 10:33:23 bootstrap-trunk-x86_64-20240116.tar.gz
-
tozhu
okay, I’ll change my command option, currently I use ‘rsync -avP’
-
tozhu
jperkin, Thank you very much for the great help
-
jperkin
for bulk building I use pbulk, there are some docs and all the scripts I use are in the pkgbuild repo, but I would caveat that with it being quite sensitive to changes so if you want to set up your own environment it's probably worth learning from my stuff but implementing your own so you will know where to look if things go wrong
-
jperkin
I can't publish 100% because a lot of my builds use custom NFS mounts etc, so there will always be a certain amount of local configuration required
-
tozhu
okay, got it, thank you very much
-
jperkin
in my spare time I've been working on a better tool to handle all of this stuff, but it's quite a big project and I don't have a huge amount of spare time ;)
-
tozhu
got it
-
tozhu
so, can I help you to do something? jperkin :-)
-
jperkin
I don't have anything public yet, other than the backend lib I wrote a few years back (
github.com/jperkin/pkgsrc-rs), but will be happy to share once I have a baseline I'm reasonably happy with
-
tozhu
jperkin: what’s the issue, is there any problem with CDN system?
pastebin.com/bmEKYPQ2
-
tozhu
the file is only 3260 byte
-
jperkin
yes, that's correct
-
jperkin
the upgrade kits are just new config files and a pgp keyring
-
tozhu
okay, thank you