All we need is gdb with correct version of binaries and their debuginfo rpms. If the coredump happens to be on RHEL7.4 and you have a RHEL7.3, good luck installing correct binaries. Yum may bail out with broken dependencies, and worse, if you provide --skip-broken, it may happily install a version other than what you need. Of course, downgrading packages is even worse!
Technically, you can install all the binary rpms (if you can download them!) in a different location or just use rpm2cpio to get the shared libraries and let gdb know where your shared libraries are. It is possible this way, but docker seems a much better option! Here are the steps I used. Bonus, if you mess up your docker container, spin a new one from scratch!
Make sure your base system can run docker. Some older versions of RHEL7 have issues. I would start with RHEL7.5 or the latest and install docker.
Download the coredump and other details on the docker host, e.g. at /data
Create a docker container with correct version as the system that took the coredump (Using RHEL7.4 here, power Linux docker image names might be different). Also use --privileged to avoid issues with SELinux.:
docker run -dt --privileged --name rh74 --hostname rh74 -v /data:/data:rw registry.access.redhat.com/rhel7.4 Or for CentOS docker run -dt --privileged --name cos77 --hostname cos77 -v /data:/data:rw docker.io/library/centos:centos7.7.1908 Or for Ubuntu docker run -dt --privileged --name ub20 --hostname ub20 -v /data:/data:rw ubuntu:20.04
Now execute bash in the running container as below (previous step would have put you in the container too. You need to run this only if you happen to exit the shell from the previous step):
docker exec -it rh74 /bin/bash
You are in a bash shell! Use yum to install what ever rpms you need. The following yum install may fail, but you have a better chance of working as the docker container is close to the system that took coredump!:
awk '{print $2}' dso_list > /tmp/rpms # edit /tmp/rpms to remove rpms that don't exist in yum repos. Also ganesha doesn't depend on glibc-common, so it won't be there in the dso_list file but will present issues with broken dependency. Adding the glibc-common with exact version as glibc needed to the /tmp/rpms will work! yum install $(cat /tmp/rpms)Now run gdb and install any debuginfo rpms that you need.
And if your docker container is stopped for some reason, you can always restart it and then execute bash as below (nothing is lost):
docker start rh74 docker exec -it rh74 /bin/bash