[RM-36583] align with create by using short hostname

[changchengx]

If using fully qualified domain name to check monitor
status after "mon add", it will hit the error that can't
find the node file under /var/run/ceph/ which result
in aboring add monitor

Signed-off-by: Changcheng Liu changcheng.liu@intel.com

[ceph-jenkins]

Can one of the admins verify this patch?

[changchengx]

@dmick
When using "mon add" command, such as "ceph-deploy mon add nstcloudcc1.sh.intel.com",
It will check the monitor status with command " sudo ceph --cluster=ceph --admin-daemon /var/run/ceph/ceph-mon.nstcloudcc1.sh.intel.com.asok mon_status". However, the actual file name is "/var/run/ceph/ceph-mon.nstcloudcc1.asok". The reason is it use "fully qualified domain name" as part of the file name. Actually, it should use short hostname.

I've checked the "mon create" command, such as "ceph-deploy mon create nstcloudcc1.sh.intel.com", it use the short hostname to check the monitor status "sudo ceph --cluster=ceph --admin-daemon /var/run/ceph/ceph-mon.nstcloudcc1.asok mon_status" and succeed.

[dmick]

Yes, but why is the call to next() added, and why is the statement order changed?

[changchengx]

@dmick

1. The function mon_hosts is a yield generator. It need use "next" function to get the value.
2. I'st better to use the same logic as "create" function: check the monitor status, then catch any error.

[tchaikov]

@changchengx could you split this commit into two? one for using name returned by mon_hosts(), another for transposing mon_status() nad catch_mon_errors()?

[changchengx]

@tchaikov
I'll check this PR within this week. It was filed about 2 years ago.

[tchaikov]

@tchaikov
I'll check this PR within this week. It was filed about 2 years ago.

the original maintainer does not work on this project anymore. guess that's why your PR was unattended.