jomitsch

Activity Graph

Page 1 of 1

2018-03-26

[16:38:06] jomitsch: has left #RubyOnRails: ("WeeChat 1.9.1")

2018-03-23

[12:41:27] jomitsch: hey all, I asked this yesterday but didn't get much of a response, we are facing a pretty frustating issue where our development rails server hangs on a code change, I think I have narrowed it down to ActionDispatch::Reloader with this logic here https://paste.fedoraproject.org/paste/EBixCIbRwRimCgbESLkxnA - does anyone have any ideas around this or a way to debug further? We are pretty stuck
[12:41:44] jomitsch: this is on rails 5.1.4
[13:23:50] jomitsch: dminuoso: yes, I can reproduce it regularly, thanks for the suggestions! I'll try those
[14:06:57] jomitsch: Inside: ya, thanks, but I'm able to reproduce the issue with spring stopped
[14:32:14] jomitsch: dminuoso: so I sent a "kill -7" to the ruby process pid when the server hangs and got this https://paste.fedoraproject.org/paste/Z8PSEWhNz0MKSyluaGSl1w (can provide full output if needed), unfortunately not much to go on there
[14:32:56] jomitsch: or at least, not much I can decipher from it :)
[14:38:32] jomitsch: dminuoso: I tried that middleware, but I don't see anything showing up in /rails/locks?
[14:38:54] jomitsch: dminuoso: I don't even see a /rails directory, am I missing something obviou?
[14:47:10] jomitsch: dminuoso: https://paste.fedoraproject.org/paste/i0s2fZZpUICu-ruoej8L1A
[14:47:55] jomitsch: I put it before Rack::Sendfile like in the example, which we do appear to be using
[14:48:33] jomitsch: dminuoso: I actually turned spring off
[14:48:48] jomitsch: dminuoso: oh I took that line out
[14:49:00] jomitsch: dminuoso: let me add it back and 'rails middleware' again
[14:50:11] jomitsch: ok now I see it:
[14:50:13] jomitsch: use ActionDispatch::DebugLocks
[14:50:47] jomitsch: dminuoso: yes
[14:50:55] jomitsch: dminuoso: I'm trying it again to double check
[14:53:49] jomitsch: dminuoso: ya can confirm, no /rails/lock on the system
[14:54:10] jomitsch: dminuoso: Is that folder created by the middleware? Do you have to run the server as root?
[14:56:55] jomitsch: dminuoso: oh ha! ok found it
[14:58:44] jomitsch: dminuoso: its showing this https://paste.fedoraproject.org/paste/QyEOY4mhXb2jtgS-KrY8jA
[15:01:14] jomitsch: removing my debugging middleware since I see that in the traceback
[15:03:15] jomitsch: dminuoso: I'm guessing this thread 2 is where to poke around? https://paste.fedoraproject.org/paste/57i~vPCgMbRZzGbsJugM5g
[15:03:46] jomitsch: dminuoso: seems like that is using the ActionDispatch::Reloader code
[16:23:49] jomitsch: dminuoso: ok cool
[16:24:12] jomitsch: dminuoso: for when you get back, if I comment out this line, it looks like the issue goes away https://git.io/vx4gN
[16:24:24] jomitsch: dminuoso: I have no idea the implications of this change though :)
[16:31:48] jomitsch: dminuoso: thanks, let me check on that
[16:39:51] jomitsch: dminuoso: I see a couple suspect places https://git.io/vx4ah and https://git.io/vx4Vf
[16:40:03] jomitsch: I'll try debugging to see who is holding the lock

2018-03-22

[18:44:01] jomitsch: has joined #RubyOnRails
[19:01:21] jomitsch: hey all! My team and I are having a pretty frustrating server hang issue in our development environment. We are discussing it on our community forum here https://community.theforeman.org/t/dev-server-hangs-on-code-change/8184/11?u=john_mitsch - I think I narrowed it down to an issue with ActionDispatch::Reloader, but would like to confirm I am making a correct assumption here? Details on how I came up
[19:01:23] jomitsch: with this are here https://paste.fedoraproject.org/paste/EBixCIbRwRimCgbESLkxnA and we are using 5.1.4
[19:53:17] jomitsch: Inside: yes, but apparently another dev tried disabling it and still saw the server hang issue
[19:58:14] jomitsch: just confirmed that myself as well
[19:58:22] jomitsch: still seeing the issue with spring stopped
[20:51:20] jomitsch: any ideas on how to debug this further?
[20:51:29] jomitsch: or if I'm looking in the right direction?