Project

General

Profile

Activity

From 02/24/2020 to 03/24/2020

03/24/2020

10:05 PM Bug #44642: cephadm: mgr dump might be too huge
> Do you know why the line "j = json.loads(out)" is choking on the integer value sent by "ceph mgr dump"?
Now I se...
Nathan Cutler
07:38 PM Bug #44669 (Fix Under Review): cephadm: rm-cluster should clean up /etc/ceph
Daniel Pivonka
02:35 PM Bug #44669 (In Progress): cephadm: rm-cluster should clean up /etc/ceph
Daniel Pivonka
02:57 PM Bug #44729: cephadm enter using docker is broken
ls works though... Jan Fajerski
02:56 PM Bug #44729 (Can't reproduce): cephadm enter using docker is broken
... Sebastian Wagner

03/23/2020

04:16 PM Bug #44720 (Need More Info): rook: rgw: allow realm != zone
Sage Weil
04:16 PM Bug #44719 (New): rook: align rgw client names with orch and cephadm
client.rgw.$realm.$zone[.$id] Sage Weil
04:11 PM Feature #44718 (Fix Under Review): NFS ganesha (mgr/cephadm)
Sebastian Wagner
04:10 PM Feature #44718 (Resolved): NFS ganesha (mgr/cephadm)
mgr/cephadm Sebastian Wagner
04:10 PM Feature #43688 (Resolved): NFS ganesha
Sebastian Wagner
02:36 PM Bug #44701 (Resolved): ganesha selinux denial
Sebastian Wagner
01:56 PM Documentation #44716 (Resolved): orchestrator/cephadm: document ceph orch apply -i -
... Sebastian Wagner
12:16 PM Backport #44710 (Resolved): octopus: doc/cephadm: replace `osd create` with `apply osd`
https://github.com/ceph/ceph/pull/34355 Nathan Cutler
08:33 AM Bug #44642 (Fix Under Review): cephadm: mgr dump might be too huge
Sebastian Wagner
08:31 AM Bug #44642: cephadm: mgr dump might be too huge
I don't know what caused this. Might actually be an artifact of our podman hang. prio=low for now.
edit: oh, you c...
Sebastian Wagner

03/20/2020

10:45 PM Bug #44642: cephadm: mgr dump might be too huge
Now, with both cephadm and container at 15.1.1-168-g06ecd31e39 I am seeing "cephadm bootstrap" fail on "ceph mgr dump... Nathan Cutler
06:48 PM Bug #44701 (Resolved): ganesha selinux denial
... Sage Weil
06:28 PM Feature #44628: cephadm: Add initial firewall management to cephadm
yeah, I also don't like to create a new dependency from the dashboard to cephadm Sebastian Wagner
05:08 PM Feature #44628: cephadm: Add initial firewall management to cephadm
I'm inclined to just open both, because the dashboard might move between ssl and not ssl. otherwise we need to make t... Sage Weil
05:10 PM Feature #44576 (Resolved): cephadm: Restart Prometheus, if a new node_exporter or alertmanager is...
This already works. Sage Weil
05:05 PM Bug #44669: cephadm: rm-cluster should clean up /etc/ceph
What should the behavior here be? Check if the /etc/ceph config has the same fsid, and if so, remove it + the keyrin... Sage Weil
05:04 PM Bug #44699 (Closed): cephadm: removing services leaves configs behind
Some of the configs are created by cephadm itself. The user might have created some too, but the config history will... Sage Weil
05:02 PM Bug #44698 (Duplicate): cephadm: removing daemons leaves auth keys behind
Sage Weil
02:19 PM Feature #43839 (Fix Under Review): enhance `host ls`
Sebastian Wagner
01:39 PM Feature #43839 (In Progress): enhance `host ls`
Sebastian Wagner
12:13 PM Bug #44692 (Pending Backport): doc/cephadm: replace `osd create` with `apply osd`
Sebastian Wagner
11:38 AM Bug #44692 (Fix Under Review): doc/cephadm: replace `osd create` with `apply osd`
Sebastian Wagner
11:33 AM Bug #44692 (Resolved): doc/cephadm: replace `osd create` with `apply osd`
Sebastian Wagner
12:06 PM Feature #43689 (Fix Under Review): cephadm: iscsi
Sebastian Wagner
11:43 AM Bug #43890 (Fix Under Review): cephadm: default hardcoded to non-ceph dockerhub
Sebastian Wagner

03/19/2020

07:19 PM Bug #44615 (Resolved): cephadm: reconfig of removed daemon
Sage Weil
02:47 PM Feature #44599 (Fix Under Review): cephadm: check-host: Returns only a single problem
Daniel Pivonka
09:28 AM Cleanup #44676 (Resolved): cephadm: Replace execnet (and remoto)
[[https://github.com/pytest-dev/execnet]] is in maintenance mode. ... Sebastian Wagner

03/18/2020

11:55 PM Bug #44673 (Rejected): cephadm: `orch apply` and `orch daemon add` use completely different code ...
... which is not obvious to users and they will use this interchangeably. Which is not really a good idea.
We sho...
Sebastian Wagner
10:43 PM Feature #44622 (Resolved): orch daemon add -i spec.yaml
Sage Weil
03:16 PM Bug #44642 (In Progress): cephadm: mgr dump might be too huge
Sebastian Wagner
02:05 PM Bug #44642 (New): cephadm: mgr dump might be too huge
Sebastian Wagner
01:48 PM Bug #44669 (Resolved): cephadm: rm-cluster should clean up /etc/ceph
... Sebastian Wagner
01:14 PM Bug #44401 (Resolved): cephadm: check host performed every time through serve loop
Sage Weil
01:14 PM Bug #44607 (Resolved): cephadm: apply(): Traceback, if host doesn't exist
Sage Weil

03/17/2020

04:12 PM Bug #44642 (Rejected): cephadm: mgr dump might be too huge
seems to be a downstream issue. Sebastian Wagner
02:51 PM Bug #44642 (Resolved): cephadm: mgr dump might be too huge
... Sebastian Wagner
04:11 PM Feature #44599 (In Progress): cephadm: check-host: Returns only a single problem
Sebastian Wagner
04:08 PM Feature #44599 (Rejected): cephadm: check-host: Returns only a single problem
Sebastian Wagner
03:30 PM Bug #44644 (Closed): cephadm: RGW: updating the spec doesn't update the mon store
when creating RGW running... Sebastian Wagner
03:20 PM Backport #43993: mimic: ceph orchestrator rgw rm: no valid command found
As `ceph orchestrator rgw rm` doesn't exist for mimic, what about just close this? Sebastian Wagner
02:25 PM Bug #44607 (Fix Under Review): cephadm: apply(): Traceback, if host doesn't exist
Sebastian Wagner
12:46 PM Feature #44622 (Fix Under Review): orch daemon add -i spec.yaml
Sebastian Wagner
10:21 AM Feature #44622 (In Progress): orch daemon add -i spec.yaml
Sebastian Wagner

03/16/2020

05:43 PM Bug #44629 (Can't reproduce): cephadm: prometheus: graph queries are not working correctly
graph queries are not working correctly. The use of instance and
exported_instance needs some investigation. On the ...
Sebastian Wagner
05:41 PM Feature #44628 (Resolved): cephadm: Add initial firewall management to cephadm
we open both 8080 and 8443 for dashboard even when the default is
https. We should probably do one or the other, not...
Sebastian Wagner
05:30 PM Documentation #44600 (Resolved): cephadm: use ssh-copy-id
Sebastian Wagner
04:08 PM Bug #44597 (Resolved): cephadm: Traceback, if ssh key is not on the remote host
Sage Weil
02:26 PM Feature #44625 (Resolved): cephadm: test dmcrypt
we need to verify it. Sebastian Wagner
01:18 PM Feature #44622 (Resolved): orch daemon add -i spec.yaml
Sage Weil

03/14/2020

03:14 PM Bug #44615 (Resolved): cephadm: reconfig of removed daemon
... Sage Weil

03/13/2020

05:00 PM Bug #44609 (Resolved): cephadm: grafana: cert problem prevents dashbaord integration
SSL cert problem prevents embedding out of the box.
Is the problem that ssl_verify is true by default? or that we...
Sebastian Wagner
04:59 PM Bug #44608 (Resolved): cephadm: grafana: bound to 127.0.0.1
after deploying I noticed that it was bound to 127.0.0.1, which blocks
client access from other machines. Should thi...
Sebastian Wagner
04:56 PM Bug #44607 (Resolved): cephadm: apply(): Traceback, if host doesn't exist
when deploying a daemon, with a host for placement - if the host doesn't
exist you get a trackback. This scenario sh...
Sebastian Wagner
04:55 PM Feature #44606 (Resolved): cephadm: RGW firewall + static port
how is the firewall being handled? AFAIK, the port is a parameter on
the rgw_frontend setting, so it could be un...
Sebastian Wagner
04:29 PM Bug #44604 (Can't reproduce): cephadm: RGW: missing spec / mon store validation
should the deployment of rgw first check the presence of a minimum set
of parms defined in the config store - if no...
Sebastian Wagner
04:25 PM Bug #44603 (Rejected): cephadm: `ls --refresh` shows Tracebacks in the log
With a host down that had daemons deployed, a --refresh shows trackbacks in the mgr log from the failed connect attem... Sebastian Wagner
04:23 PM Bug #44602 (Resolved): cephadm: `orch ls` shows daemons as online, despite host is down
With a host down that had daemons deployed:
ceph orch ls didn't show services as affected even after a --refresh i...
Sebastian Wagner
04:20 PM Feature #44601 (New): cephadm: Mix of hosts: with and without firewall
We allow a mix of hosts that either have firewall or not. I think this
should be part of the checks - either all hos...
Sebastian Wagner
04:14 PM Documentation #44600 (Resolved): cephadm: use ssh-copy-id
adding a new host:
Passing the ceph.pub key to new hosts could use the...
Sebastian Wagner
04:13 PM Feature #44599 (Resolved): cephadm: check-host: Returns only a single problem
Adding a host:
If checks fail, they show one at a time, forcing the admin to repeat
the command to get passed eac...
Sebastian Wagner
04:11 PM Bug #44598 (Resolved): cephadm: Traceback, if Python 3 is not installed on remote host
Adding a host:
if python3 isn't on the target, you get a traceback with OSError:
cannot send(already closed?) err...
Sebastian Wagner
04:10 PM Bug #44597 (Resolved): cephadm: Traceback, if ssh key is not on the remote host
Adding a host:
if the ssh key isn't on the new target you hit a trackback - which doesn't inspire confidence.
Sebastian Wagner
03:58 PM Feature #44402 (Resolved): cephadm: more complete smoke test that can be run with vstart
Sage Weil
03:56 PM Feature #44581 (Resolved): cephadm pause and cephadm resume
Sebastian Wagner
03:55 PM Bug #44569 (Resolved): NotImplementedError not caught
Sebastian Wagner
03:23 PM Cleanup #44379 (Won't Fix): orchestrator: {to,from}_json inconsistent
not worth the effort right now. Sebastian Wagner
02:53 PM Documentation #44284: cephadm: provide a way to modify the initial crushmap
Per our discussion today, using `cephadm bootstrap -c /root/ceph.conf` is the correct way to set initial crushmap or ... Ricardo Marques
02:55 AM Bug #44587 (New): failed to write <pid> to cgroup.procs:
... Sage Weil

03/12/2020

05:28 PM Bug #44440 (Resolved): cephadm should be able to infer running container
Sage Weil
02:44 PM Feature #44581 (Resolved): cephadm pause and cephadm resume
if the serve() thread is in a loop breaking all your daemons, people will want to pause it. Sebastian Wagner
12:52 PM Feature #44578 (Rejected): cephadm: verify Grafana works with Prometheus HA
Is Grafana correctly configured when a Prometheus instance is added, for example:
* Is HA working in the Grafana d...
Sebastian Wagner
12:51 PM Bug #44577 (Closed): cephadm: reconfigure Prometheus on MGR failover
we have to make sure, Prometheus knows the new prometheus exporter endpoint:
* Generate a new prometheus config po...
Sebastian Wagner
12:44 PM Feature #44576 (Resolved): cephadm: Restart Prometheus, if a new node_exporter or alertmanager is...
P needs to know the new targets / configuration Sebastian Wagner
12:37 PM Bug #37514 (Can't reproduce): mgr CLI commands block one another (indefinitely if the orchestrato...
CLI commands should now respond swiftly. (cephadm and rook) Sebastian Wagner
12:36 PM Feature #39093 (Rejected): mgr/orchestrator: add `ceph orchestrator wait`
out of scope for now. Sebastian Wagner
12:33 PM Feature #43705: cephadm: on config change, restart appropriate daemons
partially: https://github.com/ceph/ceph/pull/33855 Sebastian Wagner
12:28 PM Feature #43839 (New): enhance `host ls`
Sebastian Wagner
12:19 PM Bug #44270: Under certain circumstances, "ceph orch apply" returns success even when no OSDs are ...
Which means, we have to track which nodes are scanned and bail out, if we don't have the inventory yet? Sebastian Wagner
12:15 PM Bug #44270: Under certain circumstances, "ceph orch apply" returns success even when no OSDs are ...
new workaround: https://github.com/ceph/ceph-salt/pull/109 Sebastian Wagner
12:10 PM Bug #44559: cephadm logs an invalid stat command
just to clarify, ... Sebastian Wagner
12:07 PM Bug #44569 (Fix Under Review): NotImplementedError not caught
Sebastian Wagner

03/11/2020

08:21 PM Bug #44569 (Resolved): NotImplementedError not caught
with cephadm for example,... Sage Weil
08:21 PM Feature #43694 (Resolved): cephadm: flag dashboard user to change password
Daniel Pivonka
02:57 PM Bug #44559 (New): cephadm logs an invalid stat command
Nathan Cutler
02:30 PM Bug #44559: cephadm logs an invalid stat command
Thanks Kris - updated the bug description. Nathan Cutler
12:06 PM Bug #44559: cephadm logs an invalid stat command
Shouldn't that be... Kristoffer Gronlund
11:50 AM Bug #44559 (Fix Under Review): cephadm logs an invalid stat command
Nathan Cutler
11:46 AM Bug #44559 (Can't reproduce): cephadm logs an invalid stat command
When I run "cephadm bootstrap", I see the following in the log:... Nathan Cutler
02:52 PM Bug #44272 (Resolved): on SUSE, crash daemon starts but then always stops a couple minutes later
Sage Weil
11:17 AM Bug #44557 (Resolved): cephadm: error on run-tox-cephadm test
Kefu Chai
09:14 AM Bug #44557 (Fix Under Review): cephadm: error on run-tox-cephadm test
Kiefer Chang
08:19 AM Bug #44557 (Resolved): cephadm: error on run-tox-cephadm test
run-tox-cephadm test fails with:... Kiefer Chang
09:56 AM Backport #43994 (Need More Info): luminous: ceph orchestrator rgw rm: no valid command found
mimic backport attempt was closed. presuming non-trivial Nathan Cutler
08:05 AM Feature #44556 (Resolved): cephadm: preview drivegroups
The osd deployment in cephadm happens async in the background.
When using drivegroups, it may be not always clear...
Joshua Schmid

03/10/2020

10:19 PM Bug #44397 (Resolved): cephadm: make rgw daemons avoid the same host
Sage Weil
12:59 PM Bug #44397 (Fix Under Review): cephadm: make rgw daemons avoid the same host
Sebastian Wagner
12:43 PM Bug #44397: cephadm: make rgw daemons avoid the same host
https://github.com/ceph/ceph/commit/8330d2f2bd2bb9325ac48accedfecd6dfaab8697 Sebastian Wagner
09:27 PM Bug #44512 (Resolved): mgr/cephadm: `orch ls` doesn't obey filters
Sage Weil
07:59 AM Bug #44512 (Fix Under Review): mgr/cephadm: `orch ls` doesn't obey filters
Kiefer Chang
08:11 PM Bug #44401 (Fix Under Review): cephadm: check host performed every time through serve loop
Daniel Pivonka
04:14 PM Backport #43993 (Need More Info): mimic: ceph orchestrator rgw rm: no valid command found
first attempted backport - https://github.com/ceph/ceph/pull/33159 - was closed Nathan Cutler
03:29 PM Feature #44548 (Resolved): cephadm: persist osd removal queue
cephadm and the corresponding osd_support module currently don't save state of osds that are queued to be removed, he... Joshua Schmid
12:01 PM Feature #43699 (Resolved): mgr/cephadm: osd rm must validate before deletion
Joshua Schmid
12:00 PM Feature #43693 (Resolved): cephadm: replace OSDs
Joshua Schmid
11:54 AM Bug #44272 (Fix Under Review): on SUSE, crash daemon starts but then always stops a couple minute...
Sebastian Wagner
11:45 AM Bug #44272: on SUSE, crash daemon starts but then always stops a couple minutes later
from dmesg:... Sebastian Wagner
11:41 AM Feature #44402: cephadm: more complete smoke test that can be run with vstart
fixed via https://github.com/ceph/ceph/pull/33730 or is there something else missing? Sebastian Wagner
10:42 AM Cleanup #44379: orchestrator: {to,from}_json inconsistent
to,from}_json should not accept strings and instead always accept/return dicts or lists. Sebastian Wagner
03:42 AM Bug #44526 (Resolved): sporatic cephadm bootstrap failures: 'timed out'
Sage Weil

03/09/2020

09:52 PM Feature #43962 (Resolved): cephadm: Make mgr/cephadm declarative
Sage Weil
06:14 PM Bug #44440 (In Progress): cephadm should be able to infer running container
Ricardo Marques
05:27 PM Bug #44526 (Fix Under Review): sporatic cephadm bootstrap failures: 'timed out'
Sage Weil
05:27 PM Bug #44526: sporatic cephadm bootstrap failures: 'timed out'
I think the fundamental problem here is how ceph.in is using librados. One thread is trying to do some work, which i... Sage Weil
04:12 PM Bug #44526: sporatic cephadm bootstrap failures: 'timed out'
ceph.in sets a short 5s timeout for -h, and that's triggering shutdown, but then ceph isn't cleanly stopping...
<p...
Sage Weil
03:55 PM Bug #44526 (Resolved): sporatic cephadm bootstrap failures: 'timed out'
... Sage Weil
04:12 AM Bug #44513 (Resolved): mgr/cephadm: `orch ps --refresh` returns no results
h3. Steps to reproduce
* Create services and list daemons...
Kiefer Chang
04:05 AM Bug #44512 (Resolved): mgr/cephadm: `orch ls` doesn't obey filters
h3. Steps to reproduce
* Create a service, e.g. mgr
* List the service with service_type filter, say `osd`. The r...
Kiefer Chang

03/08/2020

10:30 PM Bug #44253 (Resolved): _apply_service should move services, not just expand/contract
Sage Weil
10:30 PM Bug #44254 (Resolved): scheduler should prefer existing daemon locations
Sage Weil
10:30 PM Bug #44392 (Resolved): mgr/orchestrator: missing SPEC and PLACEMENT field in JSON output of Servi...
Sage Weil
10:30 PM Bug #44491 (Resolved): mgr/cephadm: fail to load service specs after restarting
Sage Weil
10:29 PM Bug #44167 (Resolved): cephadm/ def _update_service: Remove should make use of spec.placement.hosts
Sage Weil
10:29 PM Bug #44302 (Resolved): cehpadm: apply_mon: NotImplementedError
Sage Weil

03/07/2020

05:48 PM Bug #43713 (Resolved): drive group filters: use `and` instead of `or`
Kefu Chai
12:21 AM Bug #44440: cephadm should be able to infer running container
ceph-container proposal for adding a new LABEL - https://github.com/ceph/ceph-container/pull/1604 Ricardo Marques

03/06/2020

09:25 PM Feature #43937 (Rejected): cephadm: make default image configurable
Closing in favor of https://tracker.ceph.com/issues/44440, see https://github.com/ceph/ceph/pull/33781 for more infor... Ricardo Marques
12:29 PM Feature #43937 (Fix Under Review): cephadm: make default image configurable
Ricardo Marques
09:02 PM Bug #44302 (Fix Under Review): cehpadm: apply_mon: NotImplementedError
Sage Weil
07:21 PM Bug #44302 (In Progress): cehpadm: apply_mon: NotImplementedError
From https://github.com/ceph/ceph/pull/33548#issuecomment-591443581
I removed apply_mon because simply reusing _ap...
Sage Weil
07:25 PM Bug #44440: cephadm should be able to infer running container
As described in https://github.com/ceph/ceph/pull/33781#issuecomment-595760420, if 'image' isn't specified then cepha... Ricardo Marques
07:13 PM Bug #44440 (New): cephadm should be able to infer running container
Ricardo Marques
12:40 PM Bug #44313: ceph-volume prepare is not idempotent and may get called twice
33755 fixed this for c-v prepare on a single device, which is what teuthology does. Sage Weil
11:37 AM Bug #44491 (Resolved): mgr/cephadm: fail to load service specs after restarting
h3. Steps to reproduce:
* Enable cephadm backend and add a host mgr0
* Create a mgr daemon*...
Kiefer Chang
08:30 AM Feature #44461 (Pending Backport): cephadm: watch Grafana certificates
Add a period check for the validity of the provided Grafana certificates and raise an health alert if they aren't hea... Patrick Seidensal

03/05/2020

10:50 PM Feature #43937: cephadm: make default image configurable
I've learned that `cephadm bootstrap` is already storing the pulled image path in a config-key:... Ricardo Marques
07:47 PM Feature #43937 (In Progress): cephadm: make default image configurable
Ricardo Marques
06:18 PM Feature #43937: cephadm: make default image configurable
I would prefer to be able to set a MON config-key with the default value.
If env variable is defined, it should ov...
Ricardo Marques
06:06 PM Feature #43937: cephadm: make default image configurable
Sebastian Wagner wrote:
> prio low, as this can be done by setting the env variable system-wide
How does one set ...
Nathan Cutler
06:57 PM Backport #43993 (New): mimic: ceph orchestrator rgw rm: no valid command found
Shyukri Shyukriev
06:29 PM Bug #44440 (Duplicate): cephadm should be able to infer running container
Nathan Cutler
02:07 PM Bug #44440 (Resolved): cephadm should be able to infer running container
If I use "cephadm" to deploy a Ceph cluster using "Container A" (not the default one) and I have that cluster running... Nathan Cutler
03:44 PM Bug #44272: on SUSE, crash daemon starts but then always stops a couple minutes later
OK, some more information:... Nathan Cutler
11:18 AM Bug #44272: on SUSE, crash daemon starts but then always stops a couple minutes later
OK, I will reproduce, obtain dmesg output, and post here.
One thing I did notice is that, with the upstream contai...
Nathan Cutler
07:00 AM Bug #44392 (In Progress): mgr/orchestrator: missing SPEC and PLACEMENT field in JSON output of Se...
Kiefer Chang

03/04/2020

11:25 PM Feature #44429 (Rejected): cephadm: make upgrade work with 'packaged' mode
In order to upgrade when the cephadm binary is installed via a package, mgr/cephadm needs to update the cephadm packa... Sage Weil
07:34 PM Bug #44273 (Can't reproduce): Getting "stray daemon osd.3 on host admin not managed by cephadm" o...
not getting this anymore Nathan Cutler
11:27 AM Feature #44414 (Resolved): bubble up errors during 'apply' phase to 'cluster warnings'
Since we moved to a fully declarative approach which handles most of the deployment in the background (k8-like) it be... Joshua Schmid
01:54 AM Bug #44390 (Resolved): cephadm: fail to create daemons
Kiefer Chang

03/03/2020

08:47 PM Bug #44401 (In Progress): cephadm: check host performed every time through serve loop
Daniel Pivonka
08:41 PM Bug #44401 (Resolved): cephadm: check host performed every time through serve loop
This should only check every N seconds (say, 10 minutes) Sage Weil
08:42 PM Feature #44402 (Resolved): cephadm: more complete smoke test that can be run with vstart
Frequently we are making (mgr/cephadm and cephadm) code changes and are developing against vstart. It would be nice ... Sage Weil
08:33 PM Feature #44205 (Resolved): cephadm: push/apply config.yml
Sage Weil
06:20 PM Bug #44397 (Resolved): cephadm: make rgw daemons avoid the same host
A test verifying this behavior was removed, check history for test_rgw_update_fail Sage Weil
11:10 AM Bug #44392 (Fix Under Review): mgr/orchestrator: missing SPEC and PLACEMENT field in JSON output ...
Kiefer Chang
10:07 AM Bug #44392 (Resolved): mgr/orchestrator: missing SPEC and PLACEMENT field in JSON output of Servi...
A new column `SPEC` was added in PR https://github.com/ceph/ceph/pull/33553.
And PLACEMENT field was added in PR htt...
Kiefer Chang
03:44 AM Bug #44390 (Fix Under Review): cephadm: fail to create daemons
Kiefer Chang
03:38 AM Bug #44390 (In Progress): cephadm: fail to create daemons
Might be a regression from https://github.com/ceph/ceph/pull/33658/files#diff-8b586ec9c3ad3e8421a8858888f7ddf0R2067. Kiefer Chang
03:36 AM Bug #44390 (Resolved): cephadm: fail to create daemons
I hit this error when creating OSDs:... Kiefer Chang

03/02/2020

05:35 PM Cleanup #44379 (Won't Fix): orchestrator: {to,from}_json inconsistent
sometimes to_json returns a dict (that can be fed to json.dumps), sometimes it returns the JSON string. We should be... Sage Weil

02/28/2020

05:20 PM Documentation #44354: cephadm: Log messages are missing
https://github.com/ceph/ceph/pull/33627 is an attempt to clarify the cephadm log gathering docs, and it has some rela... Nathan Cutler
04:43 PM Documentation #44354 (Duplicate): cephadm: Log messages are missing
There is a bug in master/octopus which causes OSDs to crash in msgr V2 without any load on the cluster. But that's no... Nathan Cutler
04:52 PM Bug #43713 (Fix Under Review): drive group filters: use `and` instead of `or`
Sebastian Wagner
02:33 PM Bug #44028 (Resolved): cephadm: usability: failing to add an osd, useless message
I don't think we can make the error message any more useful. We IMO have to fix bugs in c-v and cephadm if they appear. Sebastian Wagner
02:30 PM Bug #44272 (Triaged): on SUSE, crash daemon starts but then always stops a couple minutes later
Sebastian Wagner
02:30 PM Bug #44312 (Duplicate): ceph-volume prepare is not idempotent and may get called twice
Sebastian Wagner
02:28 PM Bug #44313: ceph-volume prepare is not idempotent and may get called twice
Sage Weil wrote:
> An alternative workaround would be to make cephadm look for 'RuntimeError: skipping vg_nvme/lv_4,...
Sebastian Wagner
02:04 PM Bug #44313: ceph-volume prepare is not idempotent and may get called twice
-Your QA run really had a excessive amount of duplicates-:
Ah, each mon logs the command....
Sebastian Wagner
02:27 PM Bug #44167 (In Progress): cephadm/ def _update_service: Remove should make use of spec.placement....
Sebastian Wagner
02:26 PM Bug #44167: cephadm/ def _update_service: Remove should make use of spec.placement.hosts
should be fixed by https://github.com/ceph/ceph/pull/33523 Sebastian Wagner
02:24 PM Feature #43696: cephadm: check that units start
I'm inclined to close this as won't fix. What shell we do?
h3. Wait, till the process is responsive?
Like http...
Sebastian Wagner
01:49 AM Feature #44308 (Resolved): mgr/cephadm: Enable alertmanager configuration in mgr/cephadm (the orc...
Sage Weil

02/27/2020

07:03 PM Bug #44121 (Resolved): calling cephadm shell again looses bash history
Sage Weil
12:38 PM Bug #44270: Under certain circumstances, "ceph orch apply" returns success even when no OSDs are ...
For ceph-salt, I have a workaround here:
https://github.com/ceph/ceph-salt/pull/99
Nathan Cutler

02/26/2020

08:50 PM Bug #44313: ceph-volume prepare is not idempotent and may get called twice
similar failure, this time a 'daemon mon add' dup:
/a/sage-2020-02-26_08:10:43-rados-wip-sage2-testing-2020-02-25-...
Sage Weil
08:27 PM Bug #44313: ceph-volume prepare is not idempotent and may get called twice
One possible fix would be to make ceph-volume itself idempotent, so that calling prepare on an already-prepared devic... Sage Weil
08:26 PM Bug #44313 (Resolved): ceph-volume prepare is not idempotent and may get called twice
symptom is a failure like so:... Sage Weil
08:24 PM Bug #44312 (Duplicate): ceph-volume prepare is not idempotent and may get called twice
Sage Weil
04:59 PM Bug #43680 (Resolved): parallelize osd provisioning
Sebastian Wagner
01:46 PM Feature #44205 (Fix Under Review): cephadm: push/apply config.yml
Joshua Schmid
10:41 AM Feature #44308 (Resolved): mgr/cephadm: Enable alertmanager configuration in mgr/cephadm (the orc...
Kristoffer Gronlund
09:44 AM Bug #44302 (Fix Under Review): cehpadm: apply_mon: NotImplementedError
Sebastian Wagner
09:08 AM Bug #44302 (Resolved): cehpadm: apply_mon: NotImplementedError
... Sebastian Wagner
09:32 AM Feature #44305 (Resolved): mgr/cephadm: Add support for removing MONs
As a future follow up of https://tracker.ceph.com/issues/44302 Sebastian Wagner

02/25/2020

03:07 PM Feature #44287 (Rejected): cephadm: Graceful Shutdown of the Whole Ceph Cluster
even though the use case is limited, it would make it possible for users to stop their production clusters for schedu... Sebastian Wagner
02:59 PM Bug #44272: on SUSE, crash daemon starts but then always stops a couple minutes later
Rethinking. I think this is an apparmor problem. Adding the output of dmesg would be helpful. Sebastian Wagner
02:41 PM Bug #44272: on SUSE, crash daemon starts but then always stops a couple minutes later
related: https://github.com/opencontainers/runc/issues/2236
After reading the code at container_linux.go:389, podm...
Sebastian Wagner
11:57 AM Documentation #44284: cephadm: provide a way to modify the initial crushmap
Actually, when this workaround is used, the option ends up being set in the MON store, and is reflected in the initia... Nathan Cutler
11:51 AM Documentation #44284 (Resolved): cephadm: provide a way to modify the initial crushmap
If we don't edit the map when bootstrapping, the CRUSH map has to be edited at runtime, which is non-trivial.
Howe...
Sebastian Wagner

02/24/2020

07:29 PM Bug #41746 (Resolved): mgr/rook: `ceph orchestrator device ls` doesn't set `available`
this is working now AFAICS Sage Weil
07:29 PM Bug #43838: cephadm: Forcefully Remove Services (unresponsive hosts)
One option is to have them 'ceph orch host rm $hostname'... Sage Weil
07:27 PM Bug #44121 (Fix Under Review): calling cephadm shell again looses bash history
Sage Weil
07:19 PM Bug #44270 (Triaged): Under certain circumstances, "ceph orch apply" returns success even when no...
i bet the problem is that the drive inventory isn't populated yet immediately after bootstrap. Sage Weil
03:12 PM Bug #44270: Under certain circumstances, "ceph orch apply" returns success even when no OSDs are ...
we might need some more in-depth validation of drive groups here. Sebastian Wagner
03:10 PM Bug #44270 (Can't reproduce): Under certain circumstances, "ceph orch apply" returns success even...
On a single-node cluster, the "cephadm bootstrap" command deploys 1 MGR and 1 MON.
On very recent versions of mast...
Nathan Cutler
07:17 PM Bug #44273 (Need More Info): Getting "stray daemon osd.3 on host admin not managed by cephadm" on...
This should have been fixed by 607263224c26... can you reproduce this with debug_mgr = 20 and attach a log? Sage Weil
03:46 PM Bug #44273 (Can't reproduce): Getting "stray daemon osd.3 on host admin not managed by cephadm" o...
I typically test the most simple deployment imaginable: single node, 1 MGR, 1 MON, and 4 OSDs. The deployment is done... Nathan Cutler
07:13 PM Feature #43867 (Resolved): cephadm: progress item for upgrade
Sage Weil
04:16 PM Feature #43673 (Need More Info): ceph-ansible playbook: pivot to cephadm
Sebastian Wagner
03:38 PM Bug #44272 (Resolved): on SUSE, crash daemon starts but then always stops a couple minutes later
Recently cephadm/orchestrator started deploying crash daemon on all cluster nodes.
On SUSE (at least), the crash d...
Nathan Cutler
01:33 PM Feature #44205: cephadm: push/apply config.yml
to sum up our discussion from Friday:
* What about doing all calls synchronously and only return async completions...
Sebastian Wagner
01:30 PM Feature #44205 (In Progress): cephadm: push/apply config.yml
Sebastian Wagner
 

Also available in: Atom