Is this a BUG REPORT or FEATURE REQUEST?: Bug report
NGINX Ingress controller version: 0.17.1
Kubernetes version: 1.10
Environment:
uname -a): 3.10What happened:
This instruction in nginx.conf:
load_module /etc/nginx/modules/ngx_http_opentracing_module.so;
opentracing_load_tracer /usr/local/lib/libjaegertracing_plugin.so /etc/nginx/opentracing.json;
opentracing on;
Causes a segmentation fault and an Nginx crash at startup:
35 Segmentation fault
The instruction causing the crash is opentracing_load_tracer.
I verified the contents of /etc/nginx/opentracing.json and it looks fine.
How to reproduce it
I can reproduce this with a simple backed nginx.conf and the config test command: nginx -t.
Anything else we need to know:
This is similar to #2222 which was also a segfault caused by the tracing plugin, except it was for Zipkin instead of Jeager.
+1
NGINX Ingress controller version: 0.17.1
Kubernetes version: 1.10.3
Environment:
Cloud provider: Baremetal
OS: CentOS 7
Kernel (e.g. uname -a): 4.16.2-1.el7
What happened:
kernel: traps: nginx[101819] general protection ip:7f68e20688ce sp:7ffd3568a0f0 error:0 in libopentracing.so.1.5.0[7f68e2064000+6000]
How to reproduce it:
config:
enable-opentracing: "true"
jaeger-collector-host: jaeger-collector.jaeger.svc.dev.cluster
@bgagnon @shurshun this was fixed by #2826
You can use the image quay.io/kubernetes-ingress-controller/nginx-ingress-controller:dev
Please remove the securityContext section from the deployment.
Closing. Please reopen if the issue persists after testing the mentioned image
I can confirm that the segfault is fixed in the latest image tag (:dev).
What's the timeline for getting this out of dev?
@jaredallard 0.18.0 contains the fix
Most helpful comment
+1
NGINX Ingress controller version: 0.17.1
Kubernetes version: 1.10.3
Environment:
Cloud provider: Baremetal
OS: CentOS 7
Kernel (e.g. uname -a): 4.16.2-1.el7
What happened:
How to reproduce it: