Correctly obtain name Ceph service names #9
Conversation
| @@ -173,17 +173,22 @@ func genContainerCgroupPath(ctype cgroups.ControlType, id string, cgroup2 bool, | |||
| node_hostname = strings.Split(node_hostname, ".")[0] | |||
|
|
|||
| // from `container_name` get the name of the ceph service for which cgroup path is to be found | |||
| ceph_service_name := strings.Split(container_name, node_hostname)[0] | |||
| ceph_service_name = strings.Trim(ceph_service_name, "-") | |||
| ceph_service_name = strings.SplitAfter(ceph_service_name, fmt.Sprintf("ceph-%s-", ceph_fsid))[1] | |||
There was a problem hiding this comment.
ceph_service_name is not used afterwards anymore, in that case it can completely go away?
There was a problem hiding this comment.
I realized using ceph_service_name improves code readability, so I've reintroduced the variable. In the latest changes, we obtain ceph service's name from the container name.
This change is needed because of https://bugzilla.redhat.com/show_bug.cgi?id=2279805
There was a problem hiding this comment.
This might have changed after the comment, but right now ceph_service_name is used at L191.
Ceph containers like "ceph-nfs-pacemaker" don't have either ceph-fsid nor node's hostname in their names. This change fixes parsing names of such containers.
|
I'm not familiar enough with go strings functions. What's the change here? The code looks really similar. Was it failing on the strings.* calls when something was missing in the string and the strings.Contains conditions fix that? |
|
Sorry about the previous comment. Since that I learnt to read and read the PR description. I think I see how the code works for |
|
@vyzigold Yes, for ceph-nfs-pacemaker it would be |
This change is related to parsing the names of Ceph containers used to create the path which helps us reach the socket files of ceph services inside
/var/lib/ceph/<ceph_fsid>/<ceph_service_name>.All Ceph containers are named in a pattern
ceph-<fsid>-mon-controller-0. However, ifceph-nfs-pacemakercontainer is deployed the current logic fails to parse container names and the code breaks because it doesn't follow the pattern as other containers.This change fixes this logic by splitting only when fsid exists in the container name.