ceph/doc/mgr/telemetry.rst

   1 .. _telemetry:
   2
   3 Telemetry Module
   4 ================
   5
   6 The telemetry module sends anonymous data about the cluster back to the Ceph
   7 developers to help understand how Ceph is used and what problems users may
   8 be experiencing.
   9
  10 This data is visualized on `public dashboards <https://telemetry-public.ceph.com/>`_
  11 that allow the community to quickly see summary statistics on how many clusters
  12 are reporting, their total capacity and OSD count, and version distribution
  13 trends.
  14
  15 Channels
  16 --------
  17
  18 The telemetry report is broken down into several "channels", each with
  19 a different type of information.  Assuming telemetry has been enabled,
  20 individual channels can be turned on and off.  (If telemetry is off,
  21 the per-channel setting has no effect.)
  22
  23 * **basic** (default: on): Basic information about the cluster
  24
  25     - capacity of the cluster
  26     - number of monitors, managers, OSDs, MDSs, object gateways, or other daemons
  27     - software version currently being used
  28     - number and types of RADOS pools and CephFS file systems
  29     - names of configuration options that have been changed from their
  30       default (but *not* their values)
  31
  32 * **crash** (default: on): Information about daemon crashes, including
  33
  34     - type of daemon
  35     - version of the daemon
  36     - operating system (OS distribution, kernel version)
  37     - stack trace identifying where in the Ceph code the crash occurred
  38
  39 * **device** (default: on): Information about device metrics, including
  40
  41     - anonymized SMART metrics
  42
  43 * **ident** (default: off): User-provided identifying information about
  44   the cluster
  45
  46     - cluster description
  47     - contact email address
  48
  49 * **perf** (default: off): Various performance metrics of a cluster, which can be used to
  50
  51     - reveal overall cluster health
  52     - identify workload patterns
  53     - troubleshoot issues with latency, throttling, memory management, etc.
  54     - monitor cluster performance by daemon
  55
  56 The data being reported does *not* contain any sensitive
  57 data like pool names, object names, object contents, hostnames, or device
  58 serial numbers.
  59
  60 It contains counters and statistics on how the cluster has been
  61 deployed, the version of Ceph, the distribution of the hosts and other
  62 parameters which help the project to gain a better understanding of
  63 the way Ceph is used.
  64
  65 Data is sent secured to *https://telemetry.ceph.com*.
  66
  67 Individual channels can be enabled or disabled with::
  68
  69   ceph telemetry enable channel basic
  70   ceph telemetry enable channel crash
  71   ceph telemetry enable channel device
  72   ceph telemetry enable channel ident
  73   ceph telemetry enable channel perf
  74
  75   ceph telemetry disable channel basic
  76   ceph telemetry disable channel crash
  77   ceph telemetry disable channel device
  78   ceph telemetry disable channel ident
  79   ceph telemetry disable channel perf
  80
  81 Multiple channels can be enabled or disabled with::
  82
  83   ceph telemetry enable channel basic crash device ident perf
  84   ceph telemetry disable channel basic crash device ident perf
  85
  86 Channels can be enabled or disabled all at once with::
  87
  88   ceph telemetry enable channel all
  89   ceph telemetry disable channel all
  90
  91 Please note that telemetry should be on for these commands to take effect.
  92
  93 List all channels with::
  94
  95   ceph telemetry channel ls
  96
  97   NAME      ENABLED    DEFAULT    DESC
  98   basic     ON         ON         Share basic cluster information (size, version)
  99   crash     ON         ON         Share metadata about Ceph daemon crashes (version, stack straces, etc)
 100   device    ON         ON         Share device health metrics (e.g., SMART data, minus potentially identifying info like serial numbers)
 101   ident     OFF        OFF        Share a user-provided description and/or contact email for the cluster
 102   perf      ON         OFF        Share various performance metrics of a cluster
 103
 104
 105 Enabling Telemetry
 106 ------------------
 107
 108 To allow the *telemetry* module to start sharing data::
 109
 110   ceph telemetry on
 111
 112 Please note: Telemetry data is licensed under the Community Data License
 113 Agreement - Sharing - Version 1.0 (https://cdla.io/sharing-1-0/). Hence,
 114 telemetry module can be enabled only after you add '--license sharing-1-0' to
 115 the 'ceph telemetry on' command.
 116 Once telemetry is on, please consider enabling channels which are off by
 117 default, such as the 'perf' channel. 'ceph telemetry on' output will list the
 118 exact command to enable these channels.
 119
 120 Telemetry can be disabled at any time with::
 121
 122   ceph telemetry off
 123
 124 Sample report
 125 -------------
 126
 127 You can look at what data is reported at any time with the command::
 128
 129   ceph telemetry show
 130
 131 If telemetry is off, you can preview a sample report with::
 132
 133   ceph telemetry preview
 134
 135 Generating a sample report might take a few moments in big clusters (clusters
 136 with hundreds of OSDs or more).
 137
 138 To protect your privacy, device reports are generated separately, and data such
 139 as hostname and device serial number is anonymized. The device telemetry is
 140 sent to a different endpoint and does not associate the device data with a
 141 particular cluster. To see a preview of the device report use the command::
 142
 143   ceph telemetry show-device
 144
 145 If telemetry is off, you can preview a sample device report with::
 146
 147   ceph telemetry preview-device
 148
 149 Please note: In order to generate the device report we use Smartmontools
 150 version 7.0 and up, which supports JSON output.
 151 If you have any concerns about privacy with regard to the information included in
 152 this report, please contact the Ceph developers.
 153
 154 In case you prefer to have a single output of both reports, and telemetry is on, use::
 155
 156   ceph telemetry show-all
 157
 158 If you would like to view a single output of both reports, and telemetry is off, use::
 159
 160   ceph telemetry preview-all
 161
 162 **Sample report by channel**
 163
 164 When telemetry is on you can see what data is reported by channel with::
 165
 166   ceph telemetry show <channel_name>
 167
 168 Please note: If telemetry is on, and <channel_name> is disabled, the command
 169 above will output a sample report by that channel, according to the collections
 170 the user is enrolled to. However this data is not reported, since the channel
 171 is disabled.
 172
 173 If telemetry is off you can preview a sample report by channel with::
 174
 175   ceph telemetry preview <channel_name>
 176
 177 Collections
 178 -----------
 179
 180 Collections represent different aspects of data that we collect within a channel.
 181
 182 List all collections with::
 183
 184   ceph telemetry collection ls
 185
 186   NAME                  STATUS                                               DESC
 187   basic_base            REPORTING                                            Basic information about the cluster (capacity, number and type of daemons, version, etc.)
 188   basic_mds_metadata    NOT REPORTING: NOT OPTED-IN                          MDS metadata
 189   basic_pool_usage      NOT REPORTING: NOT OPTED-IN                          Default pool application and usage statistics
 190   basic_usage_by_class  NOT REPORTING: NOT OPTED-IN                          Default device class usage statistics
 191   crash_base            REPORTING                                            Information about daemon crashes (daemon type and version, backtrace, etc.)
 192   device_base           REPORTING                                            Information about device health metrics
 193   ident_base            NOT REPORTING: CHANNEL ident IS OFF                  User-provided identifying information about the cluster
 194   perf_memory_metrics   NOT REPORTING: NOT OPTED-IN, CHANNEL perf IS OFF     Heap stats and mempools for mon and mds
 195   perf_perf             NOT REPORTING: NOT OPTED-IN, CHANNEL perf IS OFF     Information about performance counters of the cluster
 196
 197
 198 Where:
 199
 200 **NAME**: Collection name; prefix indicates the channel the collection belongs to.
 201
 202 **STATUS**: Indicates whether the collection metrics are reported; this is
 203 determined by the status (enabled / disabled) of the channel the collection
 204 belongs to, along with the enrollment status of the collection (whether the user
 205 is opted-in to this collection).
 206
 207 **DESC**: General description of the collection.
 208
 209 See the diff between the collections you are enrolled to, and the new,
 210 available collections with::
 211
 212   ceph telemetry diff
 213
 214 Enroll to the most recent collections with::
 215
 216   ceph telemetry on
 217
 218 Then enable new channels that are off with::
 219
 220   ceph telemetry enable channel <channel_name>
 221
 222 Interval
 223 --------
 224
 225 The module compiles and sends a new report every 24 hours by default.
 226 You can adjust this interval with::
 227
 228   ceph config set mgr mgr/telemetry/interval 72    # report every three days
 229
 230 Status
 231 --------
 232
 233 The see the current configuration::
 234
 235   ceph telemetry status
 236
 237 Manually sending telemetry
 238 --------------------------
 239
 240 To ad hoc send telemetry data::
 241
 242   ceph telemetry send
 243
 244 In case telemetry is not enabled (with 'ceph telemetry on'), you need to add
 245 '--license sharing-1-0' to 'ceph telemetry send' command.
 246
 247 Sending telemetry through a proxy
 248 ---------------------------------
 249
 250 If the cluster cannot directly connect to the configured telemetry
 251 endpoint (default *telemetry.ceph.com*), you can configure a HTTP/HTTPS
 252 proxy server with::
 253
 254   ceph config set mgr mgr/telemetry/proxy https://10.0.0.1:8080
 255
 256 You can also include a *user:pass* if needed::
 257
 258   ceph config set mgr mgr/telemetry/proxy https://ceph:telemetry@10.0.0.1:8080
 259
 260
 261 Contact and Description
 262 -----------------------
 263
 264 A contact and description can be added to the report.  This is
 265 completely optional, and disabled by default.::
 266
 267   ceph config set mgr mgr/telemetry/contact 'John Doe <john.doe@example.com>'
 268   ceph config set mgr mgr/telemetry/description 'My first Ceph cluster'
 269   ceph config set mgr mgr/telemetry/channel_ident true
 270