Building master on Slackware x64 14.2 failed

Slackware x64 15.0 is fine tho.

$ monerod --version

Monero 'Oxygen Orion' (v0.17.0.0-b54d5ae38)

TrasherDK[m]: which ubuntu version?

sorry i meant which openssl version

on 14.2

MeowingCat: that's why i recommended to run make depends directly

# openssl version

OpenSSL 1.0.2u 20 Dec 2019

seems ancient?

is it possible to update?

Well, 14.2 is ancient, but rock solid.

Updates are only possible if I want to maintain the packages myself, unless I can point cmake to "local" version.

we added openssl 3.0 support and seems as a consequence old openssl broke

openssl 1.1.x should be recent enough

Even the dev branch "current" is at OpenSSL 1.1.1q 5 Jul 2022

you can tell `cmake -D OPENSSL_ROOT_DIR=`

yes, you need at least 1.1.x or 3.0

Okay. So I need to make a static build of openssl and do a install DESTDIR=$somewhere_local

should work

Cool. I'll try. Thanks.

MeowingCat: yeah, what selsta said. just do "make depends target=aarch64-linux-android"

ofrnxmr[m]: was it you that had the difficulty drift bug? can you compile my latest master-beta branch and see if you can still reproduce it?

selsta:

This one?

I'm not sure why it happened, and have only seen it once. But yeah, will update to the latest and keep logs

yes this one

i think it doesn't show up anymore with the latest version, or maybe i just had bad luck and have to wait longer until i trigger it

.merge+ 8394

Added

.merges

gonna leave 8414 until branch is pending

Soo, I guess I'll also need a local static build of boost.

Boost 105900 (older than 1.62) is too old to link with OpenSSL 1.1.1q

Seth For Privacy: Check DMs.

based on recent comments regards tor setup i made a pull request. monero-project/monero #8423

Thanks, always welcome.

Did someone update the docs, saying ssl below 1.1.1 and a boost that ssl 1.1.1 is happy with, is required for running this thing, leaving behind a shitload of peeps, not running the latest ubuntu/debian ? Who the fuck decided that ssl 3.x was the way to go?

Did you check the commit message ?

Eh, git pull origin? No, witch one?

I mean the message in the commit that added openssl 3.x support and/or removed older ones.

It might explain the rationale.

I'm getting issues/pr's, but have missed something. Clues, please. # would help.

You can grep as fast as I can :)

I wasn't saying "the commit message explains", but really "it might".

I have no clue how to grep a commit message. Using `git log -p --name-only --grep=openssl | less`, searching for '3\.` is not helpful. "ssl: server-side: allow multiple version of TLS" and iOS something. Sorry Moo, I'm a git illiterate. Basic stuff I know.

I usualy just "git log", then /[Oo]pen[sS]

or, if I know where in a file part of the change is, git blame $thatfile

Then git blame $thatfile irrelevantcommit^ till I find a good one.

The last commit, after my last successful build, is :

commit 9750e1fa103539b3e533455500610aae76e253a5... (full message at libera.ems.host/_matrix/media/r0/do…9478e3abe5512e42f52e38ccf492753c4d6)

TrasherDK[m]: Ubuntu 18.04 has a new enough OpenSSL

what distro are you running?

and is it a good idea to use outdated openssl?

Yeah, I guess. My servers are running Slackware 14.2/15.0 The Monero ones are still on 14.2

Vulnerable versions of openssl are patched and released within 1 day.

I seems that openssl 1.1.0 and 1.1.1 need more patching than 1.0.x versions.

Version 1.1.1 will be supported until 2023-09-11 (LTS).

Version 1.0.2 is no longer supported. Extended support for 1.0.2 to gain access to security fixes for that version is available.

I don't think the Slackware guys care too much. They are still pushing updates to 1.0x when required. Latest was `Jun 29 10:32` version `openssl-1.0.2u-x86_64-4_slack14.2`

yes, that's from 2019

If that was the case, the date would have a year. Most of the vulnerabilities discovered are in the recent releases.

Those versions released by the Slackware team, are not just flat copies of the upstream. They are often patched beyond the source.

that's kind of worrisome. who's writing their patches? could be like debian and introduce new vulns

IMO, while OpenSSL 3.0's API kinda sucks, we should be migrating to it since 1.1.1 will be dead soon

Well.. My mainnet, testnet and stagenet are running with openssl `1.0.2u`. All the certificates, connection my network, same. All are TLS 1.2 (I think) I upgraded 5+ hosts to v15.0, and that was a bitch. Doing it to 10+ more kinda turns me off. Anyway. If I can build static dependencies, and build the Monero binaries as static, I will keep the nodes running.

Anyone got a bash script to build boost as static, I'm interested 🙂

"make depends" builds all dependencies as static

make depends target=x86_64-linux-gnu

don't know if it works on your slackware version, we tested it on ubuntu 18.04 and 20.04

that's how we build boost github.com/monero-project/monero/bl…r/contrib/depends/packages/boost.mk

So far it doesn't look too promising: `Failed to connect to samba.org port 443 after 15221 ms`

Ah, now it's going.

So I should probably get rid of `-DBOOST_IGNORE_SYSTEM_PATHS=OFF` when building, right?

Will `cmake bla bla ${path_to_source}` do the trick?

where did you add this?

I have `./build` and `./source`. In `./build` I do `cmake -DCMAKE_BUILD_TYPE=static -D OPENSSL_ROOT_DIR=${openssl} ../source`. This is after the `make depends` in `./source`

`make depends` are still grinding.

is this a separate question in case make depends doesn't work?

otherwise i'm not sure what you are trying to do

`make depends` didn't work. boost failed:

`CMake Warning at /usr/share/cmake-3.5/Modules/FindBoost.cmake:725 (message):... (full message at libera.ems.host/_matrix/media/r0/do…e4ac68d6d8229e6bf58644a37d64b9bfccc)

I'm at zulu+7, I have to drop off. I'll look at doing a static boost, when I wake up 😴

you can always use docker to build static binaries

soo I updated my OS and now everything seems broken in monero, here is an example backtrace for `monerod`: paste.debian.net/hidden/f210ba9f

thread 9 doesn't even display anything so no idea what this is

this was with a release build, with a debug build i can't reproduce the crash but monerod doesn't sync any blocks

the only other thing i can try is if my db is corrupted for some reason

the crashed thread doesn't show any callstack, and all other threads are either blocke waiting or also with no callstack

i deleted my db and it starts to sync again, but no idea how it got corrupted in the first place

seems like both of my issues were broken wallet cache and broken db

hopefully my SSD isn't broken lol

we don't use lmdb for the wallet cache, right?

this wallet use after free issue is remaining: paste.debian.net/hidden/e6ee8b39

any tips on how to debug this?

right, wallet cache is just a plain file

sounds like your ssd is flaky

hyc: which macos version do you have on your m1 macbook?

the ssd is like half a year old so i would be really surprised if that's the issue, i can consistently reproduce the use after free during wallet scanning