00:44 <watzon> Quicktype should add support for IDL

00:44 <watzon> As an input format

10:59 <FromGitter> <kingsleyh> hi - does anyone know what this error is when I compile my code? Program received and didn't handle signal ILL (4)

11:03 <FromGitter> <asterite> Looks similar to a segfault

11:06 <FromGitter> <kingsleyh> the binary compile and starts up - and then loads 1 record from the db and then stops with that error

11:06 <FromGitter> <kingsleyh> dmesg says: axen-master[49975] trap invalid opcode ip:6f90a3 sp:7f36c06aaae0 error:0 in axen-master[400000+819000]

11:10 <FromGitter> <kingsleyh> I ran the same thing on another EC2 instance and it did not have the problem

11:20 <FromGitter> <naqvis> sounds like a hidden bug which get triggered under certain conditions. if same executable is working fine on other box, then try to investigate the environment issues like dependent libraries versions or some environment specific settings

11:20 <FromGitter> <Daniel-Worrall> and the different dbs

11:26 <FromGitter> <kingsleyh> I'll copy the db from the other server and try it to rule that out

11:26 <FromGitter> <kingsleyh> I have a core dump file now

11:26 <FromGitter> <kingsleyh> how can I inspect it?

11:26 <FromGitter> <kingsleyh> I can install gdb - do I just run: gdb core

11:27 <FromGitter> <kingsleyh> or is there more to it than that

11:28 <FromGitter> <naqvis> is your binary built with debug info?

12:07 <FromGitter> <kingsleyh> yeah

12:10 <FromGitter> <kingsleyh> ok I created a new m4.large instance on AWS and checked out my code and build the binary

12:12 <FromGitter> <kingsleyh> it works with the supplied big db

12:19 <FromGitter> <kingsleyh> the same exact situation on the old m4.large does not work and give the illegal exception :(

12:20 <FromGitter> <kingsleyh> can I static compile via github actions? or static compile not working yet?

12:21 <FromGitter> <Daniel-Worrall> yes, tho I don't have the action file at hand

12:21 <FromGitter> <kingsleyh> can I use ubuntu-latest?

12:22 <FromGitter> <kingsleyh> or have to use something else?

12:22 <FromGitter> <Daniel-Worrall> for static, you'll need alpine

12:22 <FromGitter> <kingsleyh> ok does gh actions have one or you have to do something special?

12:22 <FromGitter> <Daniel-Worrall> There's an official docker image

12:22 <FromGitter> <kingsleyh> ah ok so you build it via docker in gh actions

12:27 <FromGitter> <kingsleyh> ok got it

13:16 <FromGitter> <kingsleyh> the static binary produced by docker using alpine linux works on the t3 instance on AWS but does not work on the m4.large instance

13:17 <FromGitter> <kingsleyh> any ideas about this?

13:36 <FromGitter> <naqvis> are you running same OS on both instances? t3 vs m4.large instances varies in hw specs only. So please be more specific on underlying OS, error you are receiving

13:42 <FromGitter> <kingsleyh> yeah

13:42 <FromGitter> <kingsleyh> works on m5.large

13:43 <FromGitter> <kingsleyh> but the static one doesn't work on m4.large

13:43 <FromGitter> <kingsleyh> I tried a couple new instances of m4

13:45 <FromGitter> <kingsleyh> none of the m4 sizes work with the static

13:45 <FromGitter> <kingsleyh> but all the t3 and m5 ones I did do work with the static

13:45 <FromGitter> <kingsleyh> very strange

13:45 <FromGitter> <kingsleyh> I think it's a some kind of crystal bug tho

13:46 <FromGitter> <kingsleyh> as the binary starts up but as soon as it touches the db it core dumps on the m4

13:46 <FromGitter> <kingsleyh> if I build it from source on the m4 then it does work on the fresh instance but it doesn't even work from compiled source on the old m4 I have

13:49 <FromGitter> <naqvis> strange enough observations and findings

13:49 <FromGitter> <naqvis> m4 and m5 only differs on cpu specs

13:49 <FromGitter> <naqvis> shouldn't be something that stops code from running

13:49 <FromGitter> <naqvis> which db your code is accessing when it crashes on m4?

13:50 <FromGitter> <naqvis> using any FFI? or pure crystal code?

13:51 <FromGitter> <naqvis> would be better if you can write a simple code to interact with db in question and do some simple query/operations and just try to run that code to see if its really db related integration which is causing the issue, or there is something else

13:52 <FromGitter> <kingsleyh> sqlite3

13:53 <FromGitter> <kingsleyh> it has a lot of crypto and C stuff

13:53 <FromGitter> <kingsleyh> as well

13:53 <FromGitter> <kingsleyh> but only seems to have issues with the db

13:53 <FromGitter> <naqvis> might be good to just a write a simple script with sqlite3 interaction and run that on m4 (where code is crashing)

13:54 <FromGitter> <naqvis> if that simple code crahses as well, then must be something related to sqlite3 libraries (as crystal sqlite3 functionality is wrapper against its c library)

13:57 <FromGitter> <naqvis> also would be good to invoke `ldd executable` to ensure there are no runtime dependencies listed on your statically compiled binary

13:59 <FromGitter> <naqvis> because passing `--static` doesn't always grantees that generated executable is fully static (below excerpt from crystal wiki) ⏎ ⏎ | When --static is given, linking static libraries is enabled, but it's not exclusive. The produced binary won't be fully static linked if the dynamic version of a library is higher in the compiler's library lookup chain than the static variant (or if the static library is

13:59 <FromGitter> ... entirely missing). In order to build a static binary you need to make sure that static versions of the linked libraries are available and the compiler can find them.

14:09 <FromGitter> <oprypin:matrix.org> very good advice

14:11 <FromGitter> <naqvis> 🌹

14:36 <FromGitter> <kingsleyh> I'm sorry I'm an idiot - I've been up since yesterday - and super tired - the static version does work on m4

14:37 <FromGitter> <kingsleyh> I was using a static that was built before I upgraded the gtihub action to generate a static on alpine linux

14:37 <FromGitter> <kingsleyh> the one I generated on ubuntu-latest - works on all t3 and all m5 - but not on m4

14:37 <FromGitter> <kingsleyh> the one generated on alpine linux - runs on all of them

14:37 <FromGitter> <kingsleyh> so it was user error on my part

14:42 <FromGitter> <kingsleyh> thanks for the help though :)

14:45 <FromGitter> <naqvis> take some rest

14:50 <FromGitter> <kingsleyh> we listed Axentro blockchain on an exchange on Friday so had a lot of issues come up that had to fix super fast

14:51 <FromGitter> <kingsleyh> but now need a rest hahaha

14:51 <FromGitter> <naqvis> congratulations, that's definitely a major milestone you have achieved

14:51 <FromGitter> <naqvis> 👍

14:53 <FromGitter> <kingsleyh> yeah just gotta get the price up now

14:53 <FromGitter> <kingsleyh> it's probably going to cost another ton of cash on marketing and market making

14:53 <FromGitter> <kingsleyh> and add another huge time drain to my day hhahha

15:17 <FromGitter> <jrei:matrix.org> Why using m4?

15:18 <FromGitter> <jrei:matrix.org> At work we use the AMD variants for new hosts, t3a and m5a

15:20 <FromGitter> <jrei:matrix.org> Good work, indeed. So much work you must have done... 😳

15:46 <FromGitter> <kingsleyh> just to save cash

15:46 <FromGitter> <kingsleyh> I switched to t3 and m5 now

16:08 <FromGitter> <jrei:matrix.org> us to yes, the a variants are even cheaper

16:13 <FromGitter> <jrei:matrix.org> there are also m6g ARM variants, but you may have tools preventing to use it and Crystal support is "only" Tier 2