vsphere input field type conflict since telegraf version 1.10

[jstiops]

Relevant telegraf.conf:

#realtime stats
 [[inputs.vsphere]]
   ## List of vCenter URLs to be monitored. These three lines must be uncommented
   ## and edited for the plugin to work.
   vcenters = [ "https://redacted/sdk" ]
   username = "redacted"
   password = "redacted"


interval = "60s"
  insecure_skip_verify = true
  force_discover_on_init = true

  # Exclude all historical metrics
  datastore_metric_exclude = ["*"]
  cluster_metric_exclude = ["*"]
  datacenter_metric_exclude = ["*"]

  collect_concurrency = 1
  discover_concurrency = 1

  ## timeout applies to any of the api request made to vcenter
  timeout = "10s"
   
#historical stats
 [[inputs.vsphere]]
   vcenters = [ "https://redacted/sdk" ]
   username = "redacted"
   password = "redacted"

  interval = "300s"
  insecure_skip_verify = true
  force_discover_on_init = true
  ## the interval before (re)discovering objects subject to metrics collection (default: 300s)
  object_discovery_interval = "600s"
  host_metric_exclude = ["*"] # Exclude realtime metrics
  vm_metric_exclude = ["*"] # Exclude realtime metrics
  cluster_metric_exclude = ["*"] # Exclude cluster metrics
  datacenter_metric_exclude = ["*"] # Exclude datacenter metrics

  #max_query_metrics = 256
  collect_concurrency = 1

  ## timeout applies to any of the api request made to vcenter
  timeout = "120s"  

System info:

telegraf 1.10 on Windows Server 2016
vSphere 6.7 update1
InfluxDB 1.7.4

Steps to reproduce:

upgraded telegraf.exe on windows from 1.9.4 tot 1.10.

Expected behavior:

no change in behavior.

Actual behavior:

seeing following entries in telegraf.log, alot of them for alot of metrics

2019-03-10T16:33:03Z E! [outputs.influxdb]: when writing to [https://redacted:8086]: received error partial write: field type conflict: input field "used_summation" on measurement "vsphere_vm_cpu" is type float, already exists as type integer dropped=921; discarding points
2019-03-10T16:34:02Z E! [outputs.influxdb]: when writing to [https://redacted:8086]: received error partial write: field type conflict: input field "datastoreWriteOIO_latest" on measurement "vsphere_host_datastore" is type float, already exists as type integer dropped=872; discarding points
2019-03-10T16:34:02Z E! [outputs.influxdb]: when writing to [https://redacted:8086]: received error partial write: field type conflict: input field "maxlimited_summation" on measurement "vsphere_vm_cpu" is type float, already exists as type integer dropped=948; discarding points

Additional info:

had to revert back to 1.9.4 to get my metrics back.

[jstiops]

Ah, probably related issues and PR described in #5563

[einhirn]

Had the same problem, added "converter" module to just make everyting a "float" and move the data to a new DB because InfluxDB always reverted to the old schema even though I deleted the measurement...

[einhirn]

Had the same problem, added "converter" module to just make everyting a "float" and move the data to a new DB because InfluxDB always reverted to the old schema even though I deleted the measurement...

Offtopic: how would I go on and try to change influxdb schema when no longer using the said workaround?

[danielnelson]

In 1.10.1, we will restore the integer types which will provide compatibility with earlier releases. If you have already moved your InfluxDB schema to floats you will need to set use_int_samples = false to stick with 1.10 style.

Closed in #5563

[danielnelson]

@einhirn I wrote up a method for migrating between schema's using Kapacitor when we needed to change the mysql plugin, I think the same method could be used here.

[danielnelson]

@jstiops @einhirn Would you be able to help test the fix? Here are some builds:

• telegraf_1.10.0~5970053b-0_amd64.deb
• telegraf-1.10.0~5970053b-0.x86_64.rpm
• telegraf-1.10.0~5970053b_windows_amd64.zip

[jstiops]

i've tested with the latest windows_am64 nightly (as i write this) instead of linked builds, seems to run ok!