An unkown unknown issue happened on 3rd March that caused the shared health record, terminology service, client registry and the HIM to be restarted by the NDC. From that day, the Health Information Mediator couldnt couldn't receive message messages from remote Health centers. The Jembi Technical team made a deep diagnosis and identified some issue that might have caused the crash and try to fix them.
This page is for documenting the the fixes applied to the registries after the crash.
Hostname Issue
After migration, The the hostname were set in a non persistent way using the command
hostname registryname
which doesnt persist doesn't persist the hostname into the file located in /etc/hostname
...
Note that the hostname will be full set after restarting the registry.
/boot full issue
Warning |
---|
Note that the following fix involves upgrading the kernel as well as removing older kernels. Ensure that the system is properly backed up before proceeding. |
The ubuntu instances that the various registries are running have automatically been updating the linux kernel. Without the old kernels being removed, the /boot partitions on some of the registries have become full as the old kernels remain there. Due to this Ubuntu then fails to install the latest kernel, breaking the package manager. The following is a possible fix for this issue (and was successfully applied to the HIM server):
Establish the status of the /boot drive by running
Code Block language bash df -ah
If /boot is not indeed full then this fix will not apply.
- Manually remove old kernels from the /boot directory to free up space. Make sure to leave at the latest kernel.
With space free, the installation of the latest kernel can proceed:
Code Block language bash sudo aptitude build-dep sudo apt-get -f install
If all goes well, the latest kernel should be installed. The old manually deleted kernel still just need to be cleanup up from the package manager:
Code Block sudo apt-get remove linux-image- #{for each kernel manually deleted}