]> git-server-git.apps.pok.os.sepia.ceph.com Git - cephmetrics.git/log
cephmetrics.git
8 years agoceph-rgw-workload.json: Set editable=False 148/head
Zack Cerza [Wed, 25 Oct 2017 16:54:42 +0000 (10:54 -0600)]
ceph-rgw-workload.json: Set editable=False

BZ1506132

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agoosd-node-detail.json: Don't hardcode cluster name
Zack Cerza [Wed, 25 Oct 2017 16:54:18 +0000 (10:54 -0600)]
osd-node-detail.json: Don't hardcode cluster name

BZ1506200

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agoMerge pull request #147 from ceph/wip-libsemanage-python
Boris Ranto [Tue, 24 Oct 2017 07:34:43 +0000 (09:34 +0200)]
Merge pull request #147 from ceph/wip-libsemanage-python

Require libsemanage-python on collector nodes

Reviewed-by: Boris Ranto <branto@redhat.com>
8 years agoRequire libsemanage-python on collector nodes 147/head
Zack Cerza [Fri, 20 Oct 2017 20:34:54 +0000 (14:34 -0600)]
Require libsemanage-python on collector nodes

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agoMerge pull request #141 from ceph/wip-fix-none-osd-colld
Zack Cerza [Thu, 12 Oct 2017 02:55:13 +0000 (20:55 -0600)]
Merge pull request #141 from ceph/wip-fix-none-osd-colld

collectors: No update on fetch osd_stats failure

8 years agoMerge pull request #143 from ceph/wip-pyyaml
Boris Ranto [Wed, 11 Oct 2017 19:43:36 +0000 (21:43 +0200)]
Merge pull request #143 from ceph/wip-pyyaml

Ensure PyYAML is installed on the Grafana node

Reviewed-by: Boris Ranto <branto@redhat.com>
8 years agoEnsure PyYAML is installed on the Grafana node 143/head
Zack Cerza [Wed, 11 Oct 2017 16:34:33 +0000 (10:34 -0600)]
Ensure PyYAML is installed on the Grafana node

It's needed by dashUpdater.py. BZ1497198

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agoMerge pull request #142 from ceph/ui-bz-fixes
Boris Ranto [Wed, 11 Oct 2017 10:18:03 +0000 (12:18 +0200)]
Merge pull request #142 from ceph/ui-bz-fixes

Ui bz fixes

Reviewed-by: Boris Ranto <branto@redhat.com>
8 years agoosd-information: disk size summary timings and metrics updated 142/head
Paul Cuzner [Wed, 11 Oct 2017 04:16:42 +0000 (17:16 +1300)]
osd-information: disk size summary timings and metrics updated

Update to the query interval needed on larger configurations to make sure
all samples are present and OSD size table units updated. BZ1496186

8 years agoosd-node-detail: updated to correct disk units
Paul Cuzner [Wed, 11 Oct 2017 04:14:25 +0000 (17:14 +1300)]
osd-node-detail: updated to correct disk units

All disk units now showing as decimal not binary values. In addition
help text updated on the raw capacity panel to better explain how the
value is derived. BZ1496186

8 years agoceph-pools : sort and selection fixes
Paul Cuzner [Wed, 11 Oct 2017 04:12:53 +0000 (17:12 +1300)]
ceph-pools : sort and selection fixes

sortByMaxima added to queries and filter (pool_name) added to the
recovery pool overview panel to work the same as the others. BZ1499734

8 years agoat-a-glance: updates to address OSD UP state issues
Paul Cuzner [Wed, 11 Oct 2017 04:11:02 +0000 (17:11 +1300)]
at-a-glance: updates to address OSD UP state issues

OSD Hosts and RGW Hosts panel changed to vonage status panel to better
reflect UP states BZ1498504

8 years agoMerge pull request #139 from ceph/wip-packaging
Zack Cerza [Tue, 10 Oct 2017 23:05:50 +0000 (17:05 -0600)]
Merge pull request #139 from ceph/wip-packaging

Finalize packaging changes

8 years agorpm: Move the rpm patch off the gitroot 139/head
Boris Ranto [Tue, 10 Oct 2017 22:24:27 +0000 (00:24 +0200)]
rpm: Move the rpm patch off the gitroot

Signed-off-by: Boris Ranto <branto@redhat.com>
8 years agoMerge pull request #140 from ceph/wip-vonage-color
Zack Cerza [Mon, 9 Oct 2017 16:07:17 +0000 (10:07 -0600)]
Merge pull request #140 from ceph/wip-vonage-color

Tweak Vonage panel color in newer versions

8 years agocollectors: No update on fetch osd_stats failure 141/head
Boris Ranto [Sat, 7 Oct 2017 09:28:09 +0000 (11:28 +0200)]
collectors: No update on fetch osd_stats failure

We currently do not check that we were able to successfully fetch the
osd_stas for a given osd_id which can result in collectors being down
for a while. We should check that we were able to fetch them before we
actually try to update them.

Signed-off-by: Boris Ranto <branto@redhat.com>
8 years agoTweak Vonage panel color in newer versions 140/head
Zack Cerza [Fri, 6 Oct 2017 19:36:05 +0000 (13:36 -0600)]
Tweak Vonage panel color in newer versions

The plugin recently added a system for configurable colors, breaking our
earlier hack. Add a new one.

See: https://github.com/Vonage/Grafana_Status_panel/commit/8633ba23bb2f336342ba3f7002a07be597b6e816

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agorpm: ipaddr ansible filter requires python-netaddr
Boris Ranto [Fri, 6 Oct 2017 17:34:58 +0000 (19:34 +0200)]
rpm: ipaddr ansible filter requires python-netaddr

Signed-off-by: Boris Ranto <branto@redhat.com>
8 years agoansible: Disable devel_mode
Boris Ranto [Thu, 5 Oct 2017 20:39:17 +0000 (22:39 +0200)]
ansible: Disable devel_mode

Signed-off-by: Boris Ranto <branto@redhat.com>
8 years agoansible: Fix merge_vars.yml
Boris Ranto [Fri, 6 Oct 2017 10:00:25 +0000 (12:00 +0200)]
ansible: Fix merge_vars.yml

Currently, we override the variables in merge_vars. However, if we run
the script several times (e.g. a host is a grafana and a collectd node)
then vars[item] is defined but it can be that it is not a mapping. In
this case we redefine the value to empty string breaking the actual
values of the variables.

We redefine e.g. devel_mode or use_epel to empty string this way making
it false for the grafana server node that is both grafana and the
collectd node.

Signed-off-by: Boris Ranto <branto@redhat.com>
8 years agoMerge pull request #138 from ceph/dboard-fixes
Boris Ranto [Fri, 6 Oct 2017 07:16:40 +0000 (09:16 +0200)]
Merge pull request #138 from ceph/dboard-fixes

dashboard updates to address minor usability issues

Reviewed-by: Boris Ranto <branto@redhat.com>
8 years agoMerge pull request #137 from ceph/wip-graphite-migrate
Zack Cerza [Fri, 6 Oct 2017 01:54:45 +0000 (19:54 -0600)]
Merge pull request #137 from ceph/wip-graphite-migrate

Fix graphite migration

8 years agodashboard updates to address minor usability issues 138/head
Paul Cuzner [Fri, 6 Oct 2017 01:37:40 +0000 (14:37 +1300)]
dashboard updates to address minor usability issues

ceph-at-a-glance: Capacity util panel now links directly to the capacity
chart in the ceph-cluster dashboard BZ 1496198

ceph-osd-information: query timespan changed from the default 1hr to 1m
aligning to the OSDs panel query on the 'at-a-glance' dashboard
BZ 1498504

8 years agoUse --fake-initial for migrations if necessary 137/head
Zack Cerza [Thu, 5 Oct 2017 19:43:57 +0000 (13:43 -0600)]
Use --fake-initial for migrations if necessary

Migrations won't work correctly if the db already exists. This manifests
in an error like:
django.db.utils.OperationalError: table "django_content_type" already exists

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agoUse 'graphite-manage migrate' instead of 'syncdb'
Zack Cerza [Thu, 5 Oct 2017 19:42:50 +0000 (13:42 -0600)]
Use 'graphite-manage migrate' instead of 'syncdb'

syncdb is deprecated.

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agoMerge pull request #136 from ceph/wip-db-wait
Zack Cerza [Wed, 4 Oct 2017 22:17:21 +0000 (16:17 -0600)]
Merge pull request #136 from ceph/wip-db-wait

Add a note about waiting to collect data

8 years agoAdd a note about waiting to collect data 136/head
Zack Cerza [Wed, 4 Oct 2017 21:23:41 +0000 (15:23 -0600)]
Add a note about waiting to collect data

Users might initially be confused that immediately after deployment, the
dashboard looks broken. This is because it doesn't yet have the data it
needs to function.

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agoMerge pull request #135 from ceph/wip-sanity
Zack Cerza [Wed, 4 Oct 2017 22:09:36 +0000 (16:09 -0600)]
Merge pull request #135 from ceph/wip-sanity

Check that all hosts are referred to by FQDN

8 years agoSkip ansible-lint on 'systemctl show' 135/head
Zack Cerza [Wed, 4 Oct 2017 21:18:09 +0000 (15:18 -0600)]
Skip ansible-lint on 'systemctl show'

The linter complains about this, but ansible doesn't have the required
functionality here.

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agoCheck that all hosts are referred to by FQDN
Zack Cerza [Tue, 3 Oct 2017 20:25:14 +0000 (14:25 -0600)]
Check that all hosts are referred to by FQDN

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agoSkip linting on sed call
Zack Cerza [Wed, 4 Oct 2017 21:14:26 +0000 (15:14 -0600)]
Skip linting on sed call

lineinfile won't necessarily work well for this.

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agoMerge pull request #132 from ceph/wip-rhsm
Boris Ranto [Tue, 3 Oct 2017 18:26:44 +0000 (20:26 +0200)]
Merge pull request #132 from ceph/wip-rhsm

ansible: Do not enable rhsm repos

Reviewed-by: Ken Dreyer <kdreyer@redhat.com>
8 years agoansible: Do not enable rhsm repos 132/head
Boris Ranto [Mon, 2 Oct 2017 09:29:01 +0000 (11:29 +0200)]
ansible: Do not enable rhsm repos

We are shipping as a regular product and as such, we cannot enable
additional repos via rhsm. The customers might even not have these repos
installed (especially the storage console repo might not be available to
them).

If we need any of the packages from these repos, we need to cross-ship
them in our product (as we already do downstream).

Signed-off-by: Boris Ranto <branto@redhat.com>
8 years agoMerge pull request #129 from ceph/wip-remove-refs-to-prod-repo
Zack Cerza [Fri, 29 Sep 2017 16:17:38 +0000 (10:17 -0600)]
Merge pull request #129 from ceph/wip-remove-refs-to-prod-repo

ansible: Remove references to old prod repo

8 years agoMerge pull request #125 from ceph/wip-collectd-vars
Zack Cerza [Thu, 28 Sep 2017 16:08:53 +0000 (10:08 -0600)]
Merge pull request #125 from ceph/wip-collectd-vars

Fix undefined collectd vars

8 years agoMerge pull request #128 from ceph/wip-prod-repos
Zack Cerza [Thu, 28 Sep 2017 16:08:33 +0000 (10:08 -0600)]
Merge pull request #128 from ceph/wip-prod-repos

Add support for custom repos

8 years agoMerge pull request #126 from ceph/selinux-fix
Boris Ranto [Thu, 28 Sep 2017 07:56:17 +0000 (09:56 +0200)]
Merge pull request #126 from ceph/selinux-fix

selinux policy update

8 years agoselinux: Update the policy 126/head
Boris Ranto [Wed, 27 Sep 2017 11:56:21 +0000 (13:56 +0200)]
selinux: Update the policy

This policy update should cover all the newly found avc denials.

Signed-off-by: Boris Ranto <branto@redhat.com>
8 years agoansible: Remove references to old prod repo 129/head
Boris Ranto [Wed, 27 Sep 2017 13:31:07 +0000 (15:31 +0200)]
ansible: Remove references to old prod repo

This fixes the build failure after the patch that removed the original
repo file.

Signed-off-by: Boris Ranto <branto@redhat.com>
8 years agoAdd support for custom repos 128/head
Zack Cerza [Mon, 25 Sep 2017 23:49:58 +0000 (17:49 -0600)]
Add support for custom repos

This is to enable testing packages built from non-master branches with
devel_mode=False

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agoMerge pull request #124 from ceph/wip-remove-prod-repo
Zack Cerza [Tue, 26 Sep 2017 00:03:32 +0000 (18:03 -0600)]
Merge pull request #124 from ceph/wip-remove-prod-repo

ansible: Remove old production repo

8 years agoFix undefined collectd vars 125/head
Zack Cerza [Fri, 22 Sep 2017 21:32:43 +0000 (15:32 -0600)]
Fix undefined collectd vars

We regressed by moving the definition of a couple of these vars. Set
them in a separate task file to avoid this in the future.

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agoansible: Remove old production repo 124/head
Boris Ranto [Fri, 22 Sep 2017 09:20:12 +0000 (11:20 +0200)]
ansible: Remove old production repo

We no longer use download.ceph.com as a production repo. We should
be removing instead of installing it.

Signed-off-by: Boris Ranto <branto@redhat.com>
8 years agoMerge pull request #121 from ceph/fix-rgw
pcuzner [Tue, 19 Sep 2017 00:15:31 +0000 (12:15 +1200)]
Merge pull request #121 from ceph/fix-rgw

rgw: update glob pattern used to detect rgw sockets (issue #115 )

8 years agorgw: update glob pattern used to detect rgw sockets 121/head
Paul Cuzner [Tue, 19 Sep 2017 00:10:26 +0000 (12:10 +1200)]
rgw: update glob pattern used to detect rgw sockets

8 years agoMerge pull request #120 from ceph/wip-poolname
pcuzner [Thu, 14 Sep 2017 23:04:25 +0000 (11:04 +1200)]
Merge pull request #120 from ceph/wip-poolname

Mon._get_df_stats(): Sanitize pool names

8 years agoMon._get_df_stats(): Sanitize pool names 120/head
Zack Cerza [Thu, 14 Sep 2017 22:03:00 +0000 (16:03 -0600)]
Mon._get_df_stats(): Sanitize pool names

_get_pool_stats() already replaces '.' with '_'; let's copy that
behavior

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agoMerge pull request #114 from ceph/wip-health-flap
pcuzner [Thu, 14 Sep 2017 21:46:34 +0000 (09:46 +1200)]
Merge pull request #114 from ceph/wip-health-flap

alert-status: Use max() for health alert condition

8 years agoMerge pull request #119 from ceph/db-changes
Zack Cerza [Thu, 14 Sep 2017 21:10:38 +0000 (15:10 -0600)]
Merge pull request #119 from ceph/db-changes

dashboard changes

8 years agoat-a-glance: MDS and OSDS panel changes 119/head
Paul Cuzner [Tue, 12 Sep 2017 23:59:04 +0000 (11:59 +1200)]
at-a-glance: MDS and OSDS panel changes

the MDS panel has been changed to report more detail using the status-panel
plugin (replacing singlestat).

The OSDs panel queries have been changed to better calculate the out/down
values of OSDs

8 years agoiscsi-overview: multiple panel fixes
Paul Cuzner [Tue, 12 Sep 2017 23:57:41 +0000 (11:57 +1200)]
iscsi-overview: multiple panel fixes

Values were shown in correctly in environments where the iscsi config had
been dropped and recreated. This update addresses issues in the following
panels; path summary, unused LUNs, defined capacity. In addition the
client charts only show entries for clients with i/o or load > 0.

8 years agonetwork/backend: dashboards updated to use the network interfaces by prefix
Paul Cuzner [Tue, 12 Sep 2017 23:55:20 +0000 (11:55 +1200)]
network/backend: dashboards updated to use the network interfaces by prefix

Before hand the network stats uses *, and therefore included metrics for
'lo' etc, which was observed to skew reporting in some environments.

8 years agoMerge pull request #116 from ceph/wip-graphite-tz
pcuzner [Tue, 12 Sep 2017 21:56:51 +0000 (09:56 +1200)]
Merge pull request #116 from ceph/wip-graphite-tz

Set timezone in Graphite's config

8 years agoMerge pull request #117 from ceph/wip-tb
pcuzner [Tue, 12 Sep 2017 21:54:08 +0000 (09:54 +1200)]
Merge pull request #117 from ceph/wip-tb

Tell collectd to log any Python tracebacks

8 years agoTell collectd to log any Python tracebacks 117/head
Zack Cerza [Thu, 27 Jul 2017 18:48:45 +0000 (11:48 -0700)]
Tell collectd to log any Python tracebacks

So that users can generate more useful bug reports.

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agoSet timezone in Graphite's config 116/head
Zack Cerza [Thu, 7 Sep 2017 20:52:06 +0000 (14:52 -0600)]
Set timezone in Graphite's config

Oddly, graphite defaults to using a hardcoded timezone rather than the
system's. This throws off queries, so let's configure it.

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agoalert-status: Use max() for health alert condition 114/head
Zack Cerza [Thu, 7 Sep 2017 20:09:47 +0000 (14:09 -0600)]
alert-status: Use max() for health alert condition

last() appears to invite flapping.

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agoMerge pull request #111 from ceph/wip-health-db
pcuzner [Wed, 6 Sep 2017 20:16:22 +0000 (08:16 +1200)]
Merge pull request #111 from ceph/wip-health-db

Don't forget to deploy the new health dashboard

8 years agoMerge pull request #112 from ceph/wip-df-stats
pcuzner [Wed, 6 Sep 2017 20:15:51 +0000 (08:15 +1200)]
Merge pull request #112 from ceph/wip-df-stats

Collect more pool stats, and display capacity in dashboard

8 years agoAdd pool capacity chart to ceph-cluster.json 112/head
Zack Cerza [Tue, 5 Sep 2017 22:31:45 +0000 (16:31 -0600)]
Add pool capacity chart to ceph-cluster.json

This chart will be helpful if the user wants to inspect pool capacity
over time, as opposed to the current state.

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agoAdd pool capacity to ceph-pools dashboard
Zack Cerza [Wed, 30 Aug 2017 19:25:29 +0000 (13:25 -0600)]
Add pool capacity to ceph-pools dashboard

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agoceph-grafana: Optionally update alert dashboard
Zack Cerza [Fri, 1 Sep 2017 16:16:53 +0000 (10:16 -0600)]
ceph-grafana: Optionally update alert dashboard

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agodashUpdater.py: Allow updating alert dashboard
Zack Cerza [Thu, 31 Aug 2017 16:00:12 +0000 (10:00 -0600)]
dashUpdater.py: Allow updating alert dashboard

This may end up only being useful for testing, but it seems important to
have the option to update the alert dashboard.

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agoAdd alert for pool capacity
Zack Cerza [Fri, 1 Sep 2017 15:47:43 +0000 (09:47 -0600)]
Add alert for pool capacity

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agoceph-grafana: Add dashboards tag
Zack Cerza [Wed, 30 Aug 2017 18:06:08 +0000 (12:06 -0600)]
ceph-grafana: Add dashboards tag

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agomon: Add 'ceph df' pool stats
Zack Cerza [Mon, 28 Aug 2017 21:12:58 +0000 (15:12 -0600)]
mon: Add 'ceph df' pool stats

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agoMerge pull request #110 from ceph/wip-eventurl
Zack Cerza [Tue, 29 Aug 2017 18:18:47 +0000 (12:18 -0600)]
Merge pull request #110 from ceph/wip-eventurl

Set EventURL in cephmetrics.conf

8 years agoDon't forget to deploy the new health dashboard 111/head
Zack Cerza [Tue, 29 Aug 2017 18:17:53 +0000 (12:17 -0600)]
Don't forget to deploy the new health dashboard

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agoSet EventURL in cephmetrics.conf 110/head
Zack Cerza [Fri, 25 Aug 2017 22:37:50 +0000 (16:37 -0600)]
Set EventURL in cephmetrics.conf

So that events get submitted to graphite

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agoMove some defaults into a cephmetrics-common role
Zack Cerza [Fri, 25 Aug 2017 22:36:41 +0000 (16:36 -0600)]
Move some defaults into a cephmetrics-common role

This is so that ceph-collectd and ceph-grafana can share some defaults

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agoMerge pull request #108 from ceph/support-svc-restarts
Zack Cerza [Thu, 24 Aug 2017 22:37:28 +0000 (16:37 -0600)]
Merge pull request #108 from ceph/support-svc-restarts

Support svc restarts

8 years agoMerge pull request #109 from ceph/wip-pipelining
Zack Cerza [Thu, 24 Aug 2017 22:36:57 +0000 (16:36 -0600)]
Merge pull request #109 from ceph/wip-pipelining

Use SSH pipelining

8 years agoUse SSH pipelining 109/head
Zack Cerza [Thu, 24 Aug 2017 22:27:58 +0000 (16:27 -0600)]
Use SSH pipelining

See http://docs.ansible.com/ansible/latest/become.html#becoming-an-unprivileged-user

Signed-off-by: Zack Cerza <zack@redhat.com>
8 years agomon: simplify the admin_socket read logic 108/head
Paul Cuzner [Wed, 23 Aug 2017 21:14:44 +0000 (09:14 +1200)]
mon: simplify the admin_socket read logic

The initial commit placed logic in each area that called the admin
socket. This patch separates the admin socket call out to a separate
method, so it gets checked in one place.

Some tidy up and comments added too.

8 years agorgw: add warning when >1 rgw admin socket is detected
Paul Cuzner [Wed, 23 Aug 2017 21:11:52 +0000 (09:11 +1200)]
rgw: add warning when >1 rgw admin socket is detected

8 years agorgw: look for the admin_socket on each call
Paul Cuzner [Wed, 23 Aug 2017 02:58:34 +0000 (14:58 +1200)]
rgw: look for the admin_socket on each call

The admin_socket name for rgw is not fixed, unlike mon/osds. Therefore
to account for svc restarts and name changes the socket name is
determined at each get_stats cycle. If the socket isn't there, the
collector just passes back the version of radosgw to the caller and
will send stats again once a socket is detected on the host

8 years agomon: account for null dict from _admin_socket
Paul Cuzner [Wed, 23 Aug 2017 02:56:37 +0000 (14:56 +1200)]
mon: account for null dict from _admin_socket

the _admin_socket method could return a null dict if the
socket is not there (i.e. ceph-mon is down). By checking for the
empty dict, the collector can remain active while ceph-mon is
stopped and restarted during normal maintenance processes on a
host.

8 years agoiscsi: trigger stats only when iscsi is active
Paul Cuzner [Wed, 23 Aug 2017 02:54:15 +0000 (14:54 +1200)]
iscsi: trigger stats only when iscsi is active

look for the iscsi dir in sysfs to determine when to
send the iscsi stats. If the iscsi base dir is not there
the collector will just send the version of gwcli

8 years agocommon: catch stderr output into the stdout pipe
Paul Cuzner [Wed, 23 Aug 2017 02:52:33 +0000 (14:52 +1200)]
common: catch stderr output into the stdout pipe

Ensure stdout and stderr output is returned to the caller

8 years agocephmetrics: simplified the probe logic
Paul Cuzner [Wed, 23 Aug 2017 02:51:52 +0000 (14:51 +1200)]
cephmetrics: simplified the probe logic

probe logic changed in the base class, so the code here can
change to take advantage

8 years agobase: detect ceph role based on installed binaries
Paul Cuzner [Wed, 23 Aug 2017 02:51:02 +0000 (14:51 +1200)]
base: detect ceph role based on installed binaries

In addition get_version simplified, and the _admin_socket method logic
tightened up to account for missing socket files.

8 years agoMerge pull request #100 from ceph/event-info
Zack Cerza [Tue, 22 Aug 2017 18:37:14 +0000 (12:37 -0600)]
Merge pull request #100 from ceph/event-info

Add bluestore performance metrics and enhanced ceph health information

8 years agoMerge pull request #103 from ceph/wip-ubuntu-gweb
Zack Cerza [Tue, 22 Aug 2017 18:34:31 +0000 (12:34 -0600)]
Merge pull request #103 from ceph/wip-ubuntu-gweb

ansible: Use graphite-web instead of graphite-api on Ubuntu

8 years agominor fixes during PR review 100/head
Paul Cuzner [Tue, 22 Aug 2017 00:08:56 +0000 (12:08 +1200)]
minor fixes during PR review

8 years agoceph-cluster: fix queries for rgw and iscsi version tables
Paul Cuzner [Mon, 21 Aug 2017 04:58:04 +0000 (16:58 +1200)]
ceph-cluster: fix queries for rgw and iscsi version tables

8 years agodashboard query update to filter out old OSDs
Paul Cuzner [Mon, 21 Aug 2017 04:57:16 +0000 (16:57 +1200)]
dashboard query update to filter out old OSDs

Old OSDs will still exist in the TSDB, and could show as out or down.
The update uses transformNull to pick out osds with null values and
filter them out of the results shown.

8 years agobase: handle admin_socket connection failures gracefully
Paul Cuzner [Mon, 21 Aug 2017 04:52:22 +0000 (16:52 +1200)]
base: handle admin_socket connection failures gracefully

Connection errors are reported to the parent object and the cephmetrics
log file

8 years agorgw: handle the scenario where the socket doesn't respond
Paul Cuzner [Mon, 21 Aug 2017 04:51:18 +0000 (16:51 +1200)]
rgw: handle the scenario where the socket doesn't respond

The perf dump to the socket could fail, so this change handles this
scenario by just reporting the issue avoiding connectionerror
problems

8 years agonode-detail: minor fix to template for osd_ids
Paul Cuzner [Mon, 21 Aug 2017 01:38:48 +0000 (13:38 +1200)]
node-detail: minor fix to template for osd_ids

osd_id template had to be updated to avoid issues with the move of
num_osds and ceph_version metrics.

8 years agorelationships diagram updated to reflect new ceph-health dashboard
Paul Cuzner [Mon, 21 Aug 2017 01:36:52 +0000 (13:36 +1200)]
relationships diagram updated to reflect new ceph-health dashboard

8 years agoceph-pools: fix yaxis in pool iops chart
Paul Cuzner [Mon, 21 Aug 2017 01:36:29 +0000 (13:36 +1200)]
ceph-pools: fix yaxis in pool iops chart

8 years agoosd-information: fixes for null entries and time windows used on pie-charts
Paul Cuzner [Mon, 21 Aug 2017 01:36:01 +0000 (13:36 +1200)]
osd-information: fixes for null entries and time windows used on pie-charts

osd's that fail result in nulls in the data series, so queries updated to
account for this gaps. In addition, a time window if 2 mins used to
restrict the obs that have to be grabbed from graphite for the pie charts

8 years agoceph-cluster: health info removed, ceph version tables added
Paul Cuzner [Mon, 21 Aug 2017 01:33:59 +0000 (13:33 +1200)]
ceph-cluster: health info removed, ceph version tables added

exploits the updates to the basecollector class which now provide ceph
version which is presented in table form, by role.

8 years agoMerge pull request #97 from ceph/wip-branto
Gregory Meno [Fri, 18 Aug 2017 17:39:13 +0000 (10:39 -0700)]
Merge pull request #97 from ceph/wip-branto

rpm: Fix spec file source versioning

8 years agorpm: Fix spec file source versioning 97/head
Boris Ranto [Wed, 16 Aug 2017 11:11:00 +0000 (13:11 +0200)]
rpm: Fix spec file source versioning

Signed-off-by: Boris Ranto <branto@redhat.com>
8 years agoansible: Use graphite-web instead of graphite-api on Ubuntu 103/head
David Galloway [Tue, 15 Aug 2017 21:17:02 +0000 (17:17 -0400)]
ansible: Use graphite-web instead of graphite-api on Ubuntu

Because of https://github.com/brutasse/graphite-api/issues/222

Signed-off-by: David Galloway <dgallowa@redhat.com>
8 years agoMerge pull request #95 from ceph/iscsi-fix
Zack Cerza [Tue, 8 Aug 2017 22:42:05 +0000 (15:42 -0700)]
Merge pull request #95 from ceph/iscsi-fix

iscsi.py: fix to defer the import of rtslib_fb

8 years agoiscsi.py: fix to defer the import of rtslib_fb 95/head
Paul Cuzner [Tue, 8 Aug 2017 21:48:55 +0000 (09:48 +1200)]
iscsi.py: fix to defer the import of rtslib_fb

the goal of the parent module cephmetrics is to be generic across the
different ceph roles. By deferring the import of rtslib to the instantiation
of the first (and only!) ISCSIGateway object cephmetrics can import this
iscsi module without a problem regardless of the runtime environment.

8 years agoMerge pull request #92 from ceph/iscsi-support
Zack Cerza [Tue, 8 Aug 2017 18:10:13 +0000 (11:10 -0700)]
Merge pull request #92 from ceph/iscsi-support

Add iSCSI gateway support

8 years agoiscsi.py: updates to address code review comments 92/head
Paul Cuzner [Mon, 7 Aug 2017 23:44:16 +0000 (11:44 +1200)]
iscsi.py: updates to address code review comments