-
Notifications
You must be signed in to change notification settings - Fork 547
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Passenger standalone 5.0.6 crashes with SIGSEGV #1469
Comments
Another crash log, this time including the detailed
|
Thank you for reporting this. We will investigate this issue. |
Hi @karlsvec. Due to lack of debugging symbols, it isn't entirely clear what the cause of this problem is. The few readable backtraces in the report seem to make little sense. However, we fixed a crash a few days ago which was related to handling of sub-URI deployments. Are you using sub-URI deployments? We've also modified the Debian packages to include a -dbg version which includes debugging symbols, so that in the future users can send us better bug reports. |
@FooBarWidget I'm not using a sub-URI deployment, but I'll try the |
@FooBarWidget I'm seeing the SIGSEGV crash in the latest UPDATE: Instead of continuing to spam this issue with long crash logs embedded in comments, here's a link to a Gist: https://gist.github.com/karlsvec/db8d0354ab540b987737. This Gist contains the most recent crash log, including the
I'll post a crash log with the detailed |
Thanks, that report is a lot clearer. It has a lot of resemblance with #1459. Might even be the same issue. |
Using passenger docker on aws (m3.xlarge - 15GiB men, 4 cores, node.js app) occasionally causes our app to crash, we wouldn't have noticed anything but users started complaining that their sessions would drop occasionally. Unfortuneately I am unable to reproduce this issue and I can't make sense out of the error output. We had a hunch that maybe IE 11 was causing the crash, but again we can't reproduce the error Here is the log output: (log level 7) (there are 3 errors in the log, it looks like the subsequent two are semi related to the first SegFault) It happens quite often so if i can be of help by twiddling flags or anything let me know. https://gist.github.com/blakmatrix/c7907607272fcd016ccf I will note that I reduced min process to 1 vs. 4 and I get the same issue. |
We have similar issue. Passenger 5.0.6 as nginx module (from standard passenger packages for Debian), like in #1459 Standard Rails app, nothing unusual. There are not much requests, because it's a staging server, but from time to time we see 502 error on frontend and similar crash logs on the server. PS: Sorry for offtopic, but I was not able to find a way to downgrade a package via apt-get to 4.0.59 version. Any chance that someone advises me how to do it? |
@dhampik Passenger 4 packages are now in a separate APT repository: https://blog.phusion.nl/2015/03/08/passenger-4-apt-repository-now-available/ |
I will add my own stack trace, I don't know if this is related or not (perhaps a wholly different bug). I am on Mac OS X 10.10.3, with passenger built as an Apache 2.4.10 module |
@karlsvec @blakmatrix @stormsilver your traces look similar indeed, are you using Out-Of-Band Garbage Collection? |
unsure, I use whatever is the default in passenger-docker |
@blakmatrix we have a range of images for passenger-docker, which one are you using? Ruby? Version? |
@OnixGH Yes I am - here's the relevant snippet. This is a Rails 4.2.1 app under Ruby 2.1.2p95 running in development mode. What else can I tell you? if defined?(PhusionPassenger)
# enable OOB GC, there is more to do here to take advantage of the Ruby 2.1 GC, but it doesn't work with Passenger's smart spawn, which is a huge bummer: http://blog.phusion.nl/2014/01/31/phusion-passenger-now-supports-the-new-ruby-2-1-out-of-band-gc/
PhusionPassenger.require_passenger_lib 'rack/out_of_band_gc'
use PhusionPassenger::Rack::OutOfBandGc, strategy: :counting, frequency: 5
# use PhusionPassenger::Rack::OutOfBandGc, :strategy => :gctools_oobgc
end |
@OnixGH: Yes, I am using out-of-band GC. My configuration is the same as @stormsilver's. I'm on Ruby 2.1.6. |
@OnixGH docker-full |
We've found a way to reproduce the crash. It seems to trigger only when OOBGC is enabled, which @karlsvec @stormsilver @shirosaki also confirmed to have enabled. |
@blakmatrix do you have OOBGC enabled (like @stormsilver posted, for example in your config.ru)? It's something that your app configuration controls, regardless of which passenger image or version you use. |
Bug found & fixed in stable (coming in 5.0.8). |
@OnixGH our docker is configured for a node.js application. As far as I know I'm not using OOBGC. |
Huzzah! Nice work, thank you!
|
process selection method was only suitable for enabled processes, and not disabling/disabled processes. Closes GH-1469.
@blakmatrix thanks for following up. OOBGC process pausing is only relevant for ruby, so you might be experiencing a different issue. In any case it would help a lot if you had a trace with debugging symbols like the others did (you can install passenger-debuginfo to get this). |
I am having this same issue with Phusion Passenger version 5.0.9 Standard Rails App and high number of requests. But it manifests regardless of the traffic. I am not sure of my Garbage Collector, but I am more than willing to help if someone can provide some direction. I can post my dump here: https://gist.github.com/germs12/f52b511d411eaae96688 |
Because I was seeing this error in production (shame on me) I didn't have a lot of time to debug. I upgraded to the latest I'd do that if I found myself experiencing this issue in the future. |
Hello everyone. We encounter exactly the same problem. I checked at our stack trace and it's quite the same than the others. I was looking for some OOBGC info into our configuration files but no clue. We have a bitnami redmine installation (Ruby 2.0.0 and Passenger 5.0.4) and i can't figured out where i could find something. First of all, I was looking for information on the bitnami community forum but here it's the only place where i finally find someone with the same issue. I checked our apps config.ru files but there isn't much information in it. If anyone has a clue, except than upgrade the passenger version, (we can't) it would be appreciated, Thx (As you probably saw i'am not a native english speaker so please excuse me if i burned your eyes ;-) ) |
@MisterBarbudo Upgrade passenger and it should fix the issue. |
@MisterBarbudo I'm assuming that by saying you can't upgrade the passenger version yourself you mean that bitnami is controlling it? In that case it seems to me that you should contact their support, we can't do much but fix the bug and that was more than 3 months ago. |
First of all, thanks for yours answers @germs12 @OnixGH yes totally. It's an all inclusive package and i don't really know much about passenger so i don't want to risk to break something I was looking for a way to upgrade passenger but our server can't download packages from the web (the only way i found so far to upgrade ). I wiil look at it tomorow. Thx. |
Hello guys, Eventually, I got the time to work on this issue. I managed to upgrade Passenger from 5.0.4 to 5.0.20 and you were right, it fixes the app. Thx. |
👍 |
I'm seeing occasional crashes with 5.0.6 standalone using the official packages from the APT repository on Ubuntu 14.04 LTS.
I'm using Nginx as a proxy for Passenger standalone. Passenger is configured to use the built-in engine and listen on a Unix socket. The crash is generally accompanied by something like this in Nginx's error log:
The text was updated successfully, but these errors were encountered: