Reset configmap log_file option #208
Conversation
The log_file config option in DEFAULT section is used for both nsx-ncp-config and nsx-node-agent-config. So, the log_file will be used for bootstrap, ncp and node-agent pods once it is set. All pods will write log in the same log_file. However, ncp and node-agent write log_file with ncp user, bootstrap pod write log_file with root user. The log_file is set in operator configmap, boostrap, ncp and node-agent pod will be restarted. If bootstrap pod create log_file before than ncp/node-agent pods is trying to access. Bootstrap pod creates and write log_file with root users. In such case, the ncp/node-agent will hit Permission denied issue when writing log_file and the ncp/node-agent pod will be crashed. This patch is to make a workaround solution to fix this issue: 1. Reset log_file to the default None value so that ncp, bootstrap and node-agent will use their own default logfile, respectively. 2. Print a log message when detecting log_file option is set by user.
| @@ -158,6 +168,16 @@ func (adaptor *ConfigMapOc) validateConfigMap(configmap *corev1.ConfigMap) []err | |||
| if err != nil { | |||
| appendErrorIfNotNil(&errs, errors.Wrapf(err, "mtu is invalid")) | |||
| } | |||
| // Check DEFAULT section log_file option | |||
There was a problem hiding this comment.
this part is the same as 132-141, maybe add a method instead of same code block.
There was a problem hiding this comment.
We are trying to merge the two func (adaptor *ConfigMapOc) validateConfigMap(configmap *corev1.ConfigMap) and func (adaptor *ConfigMapK8s) validateConfigMap(configmap *corev1.ConfigMap) since the logic of these two functions are basically the same. but it's a little bit rush to do it in the operator 4.1.0. We can consider this in the next release.
| // Revert DEFAULT log_file to default value | ||
| fillDefault(cfg, "DEFAULT", "log_file", "", true) | ||
| // Write config back to ConfigMap data | ||
| (*data)[operatortypes.ConfigMapDataKey], err = iniWriteToString(cfg) |
There was a problem hiding this comment.
Do we need to reset the value call iniWriteToString if in any case we are failing validation and therefore setting operator in degraded state?
There was a problem hiding this comment.
if there is failing or error when calling iniWriteToString, we write error in NCP install CRD status, so, it means NCP install CRD is a degraded state. I feel this approach should be ok.
The log_file config option in DEFAULT section is used for both nsx-ncp-config and nsx-node-agent-config.
So, the log_file will be used for bootstrap, ncp and node-agent pods once it is set. All pods will write log in the same log_file. However, ncp and node-agent write log_file with ncp user, bootstrap pod write log_file with root user.
The log_file is set in operator configmap, boostrap, ncp and node-agent pod will be restarted. If bootstrap pod create log_file before than ncp/node-agent pods is trying to access. Bootstrap pod creates and write log_file with root users. In such case, the ncp/node-agent will hit Permission denied issue when writing log_file and the ncp/node-agent pod will be crashed.
This patch is to make a workaround solution to fix this issue: