cephadm: error trying to get ceph auth entry for crash daemon#35274
Merged
sebastian-philipp merged 1 commit intoceph:masterfrom Jun 25, 2020
Merged
cephadm: error trying to get ceph auth entry for crash daemon#35274sebastian-philipp merged 1 commit intoceph:masterfrom
sebastian-philipp merged 1 commit intoceph:masterfrom
Conversation
Contributor
|
This would generate a big chaos of existing daemons using the bare name and new daemons using the fqdn. I think changing this for octopus is very late now. Do you think we can also fix this by improving ceph/src/pybind/mgr/cephadm/utils.py Lines 18 to 30 in 9e1e025 ? |
Member
Author
I thing that not using the real name of the host is going to cause more problems in the future. Besides that, this adds the need to use a "consensus" about when to use the real name or the abbreviated name... another source of problems. |
Contributor
|
don't know. Ceph already prefers bare host names. Imagine if you simply want to change the domain of a cluster. That would be extremely complicated, if we add this to the daemon names. |
Contributor
|
I have the feeling that doing this radical change of creating new daemons with a different naming scheme requires some more thought. Might be something to do for pacific? |
Member
Author
Avoiding the error with a not wonderful trick. (In spanish we call this kind of things "ñapa") |
3 tasks
src/pybind/mgr/cephadm/module.py
Outdated
Comment on lines
+1649
to
+1651
| partial_auth_key = daemon_id | ||
| if daemon_type == 'crash': # | ||
| partial_auth_key = host |
Contributor
There was a problem hiding this comment.
can you move this code into utils.name_to_auth_entity?
Member
Author
There was a problem hiding this comment.
Done.
Note: It seems that mgr also changed his auth key
Member
Author
|
Solved a couple of tricky collateral "mypy" issues. |
Contributor
|
Contributor
|
ping @jmolmo |
sebastian-philipp
approved these changes
Jun 18, 2020
src/pybind/mgr/cephadm/module.py
Outdated
Comment on lines
+890
to
+899
| host, # type str | ||
| entity, # type str | ||
| command, # type str | ||
| args, # type List[str] | ||
| addr = "", # type Optional[str] | ||
| stdin = "", # type Optional[str] | ||
| no_fsid = False, # type Optional[bool] | ||
| error_ok = False, # type Optional[bool] | ||
| image = False, # type Optional[str] | ||
| env_vars= None # type Optional[List[str]] |
Contributor
There was a problem hiding this comment.
why change this to the old style type annotation?
Member
Author
There was a problem hiding this comment.
Because I didn't know what is the preferred annotation type... by your comment I assume it is the "new" annotation style. I will change.
Comment on lines
+19
to
39
| def name_to_auth_entity(daemon_type, # type: str | ||
| daemon_id, # type: str | ||
| host = "" # type Optional[str] = "" | ||
| ): | ||
| """ | ||
| Map from daemon names to ceph entity names (as seen in config) | ||
| Map from daemon names/host to ceph entity names (as seen in config) | ||
| """ | ||
| daemon_type = name.split('.', 1)[0] | ||
| if daemon_type in ['rgw', 'rbd-mirror', 'nfs', 'crash', 'iscsi']: | ||
| return 'client.' + name | ||
| if daemon_type in ['rgw', 'rbd-mirror', 'nfs', "iscsi"]: | ||
| return 'client.' + daemon_type + "." + daemon_id | ||
| elif daemon_type == 'crash': | ||
| return 'client.' + daemon_type + "." + host | ||
| elif daemon_type == 'mon': | ||
| return 'mon.' | ||
| elif daemon_type in ['osd', 'mds', 'mgr', 'client']: | ||
| return name | ||
| elif daemon_type == 'mgr': | ||
| return daemon_type + "." + daemon_id | ||
| elif daemon_type in ['osd', 'mds', 'client']: | ||
| return daemon_type + "." + daemon_id | ||
| else: | ||
| raise OrchestratorError("unknown auth entity name") |
Contributor
There was a problem hiding this comment.
you want to add a pytest for this?
If your cluster has nodes with a . in the name. This will happen. Signed-off-by: Juan Miguel Olmo Martínez <jolmomar@redhat.com>
Contributor
|
Failures:
|
| raise OrchestratorError('no hosts defined') | ||
| out, err, code = self._run_cephadm( | ||
| host, None, 'pull', [], | ||
| host, '', 'pull', [], |
Contributor
There was a problem hiding this comment.
unfortunately, the upgrade test totally failed: http://pulpito.ceph.com/swagner-2020-06-23_11:55:14-rados:cephadm-wip-swagner-testing-2020-06-23-1057-distro-basic-smithi/5172323/
I'd like to verify this PR in a new run.
Contributor
|
jenkins test make check |
3 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
If your cluster has nodes with a <.> in the name. This will happen.
Is the case if your hosts use a FQDN.
It seems that we changed the name of the host when using the <get_unique_name> function... Maybe I miss something .. probably ... but why we needed to change the name of the host?
I have removed the "problematic lines"
Fixes: https://tracker.ceph.com/issues/45726
Details:
Extracted the line where we tried to get auth details for the crash daemon, and executed alone:
Checking what entries we have with "crash":
so.. it seems that we are searching using a wrong key (client.crash.ceph-node-00) instead the valid key (client.crash.ceph-node-00.cephlab.com)
Digging in the code results that we need "ename" in the daemonid...
and the daemon_id is wrong ...