Bug 2210873
| Summary: | assimilate_{{ tripleo_cephadm_cluster }}.conf required if `--crush-hierarchy` is used | ||
|---|---|---|---|
| Product: | Red Hat OpenStack | Reporter: | Emilien Macchi <emacchi> |
| Component: | tripleo-ansible | Assignee: | Manoj Katari <mkatari> |
| Status: | MODIFIED --- | QA Contact: | Alfredo <alfrgarc> |
| Severity: | low | Docs Contact: | |
| Priority: | low | ||
| Version: | 17.1 (Wallaby) | CC: | fpantano, fpiccion, jelynch, johfulto, mkatari, rheslop |
| Target Milestone: | ga | Keywords: | Triaged |
| Target Release: | 17.1 | ||
| Hardware: | Unspecified | ||
| OS: | Unspecified | ||
| Whiteboard: | |||
| Fixed In Version: | tripleo-ansible-3.3.1-1.20230620010933.03ab6c5.el9ost | Doc Type: | Bug Fix |
| Doc Text: |
Before this update, deploying Red Hat Ceph Storage (RHCS) with the `--crush-hierachy` parameter failed with an 'assimilate.conf not found' error when setting crush rules.
+
With this update, the RHCS CLI does not bind mount `assimilate.conf` and RHCS sets crush rules successfully.
|
Story Points: | --- |
| Clone Of: | Environment: | ||
| Last Closed: | Type: | Bug | |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
Also weird: ``` [root@controller-0 ~]# file /home/ceph-admin/assimilate_ceph.conf /home/ceph-admin/assimilate_ceph.conf: ASCII text ``` The file exists... And the file doesn't exist on the 2 other controllers. 1. assimilate_ceph.conf is an argument passed to the cephadm bootstrap command. 2. cephadm boostrap is only run on one node (the first controller node by default) 3. Thus, it's expected that it not be present on the other two controllers Sure Francesco, working on it. Another way to encounter this bug is by using CephCrushRules in your Heat environment files. Same fix (884597) should address it though. *** Bug 2228783 has been marked as a duplicate of this bug. *** Updated doc text looks good to me. |
Description of problem: When using deployed-ceph with `--crush-hierarchy` parameter to the ceph deploy command, a file has to exist but should not be needed by default. Version-Release number of selected component (if applicable): RHOS-17.1-RHEL-9-20230525.n.1 How reproducible: Deploy ceph with `--crush-hierarchy` parameter and without `--config` nor pre-created ceph.conf. Actual results: The ceph deployment will fail: 2023-05-29 14:59:42,544 p=165056 u=stack n=ansible | 2023-05-29 14:59:42.544281 | fa163e4e-4a57-698b-b613-0000000000d7 | FATAL | insert new default crush rule into daemon to prevent restart | controller-0 -> 172.16.200.13 | item=controller-2 | error={"ansible_loop_var": "item", "changed": false, "cmd": ["podman", "run", "--rm", "--net=host", "--ipc=host", "--volume", "/var/lib/ceph/305f99dc-9741-55f8-879a-82b88672831a/config/:/etc/ceph:z", "--volume", "/home/ceph-admin/assimilate_ceph.conf:/home/assimilate_ceph.conf:z", "--volume", "/var/run/ceph/305f99dc-9741-55f8-879a-82b88672831a:/var/run/ceph:z", "--entrypoint", "ceph", "registry-proxy.engineering.redhat.com/rh-osbs/rhceph:6-115", "--admin-daemon", "/var/run/ceph/ceph-mon.controller-2.shiftstack.test.asok", "config", "set", "osd_pool_default_crush_rule", "3"], "delta": "0:00:00.048437", "end": "2023-05-29 18:59:42.521240", "item": "controller-2", "msg": "non-zero return code", "rc": 125, "start": "2023-05-29 18:59:42.472803", "stderr": "Error: statfs /home/ceph-admin/assimilate_ceph.conf: no such file or directory", "stderr_lines": ["Error: statfs /home/ceph-admin/assimilate_ceph.conf: no such file or directory"], "stdout": "", "stdout_lines": []} 2023-05-29 14:59:42,545 p=165056 u=stack n=ansible | [WARNING]: ('controller-0 -> 172.16.200.13', 'fa163e4e-4a57-698b-b613-0000000000d7') missing from stats Expected results: ceph.conf is not required if I don't want to provide one, so ceph-cli should not try to bind mount /home/ceph-admin/assimilate_ceph.conf which doesn't exist.