Error while upgrading from v1.10.0 to 1.10.1

[Yoshyn]

Environment :

• Elixir (elixir -v):

Erlang/OTP 20 [erts-9.3.3.11] [source] [64-bit] [smp:2:2] [ds:2:2:10] [async-threads:10] [hipe] [kernel-poll:false]

Elixir 1.8.2 (compiled with Erlang/OTP 20)

Describe the bug

Error while upgrading from v1.10.0 to 1.10.1 with a minimal configuration (just the minimal on the readme) without changing anything except the following :

Changes :

1. Before update :

  defp deps do
    [{...,
     {:new_relic_agent, "~> 1.10"},
     {:new_relic_phoenix, "~> 0.2"},
   ]
  end

• Agent version (mix deps | grep new_relic_agent):

* new_relic_agent (Hex package) (mix)
  locked at 1.10.0 (new_relic_agent) 008a074a

2. After update :

  defp deps do
    [{...,
     {:new_relic_agent, "== 1.10.1"},
     {:new_relic_phoenix, "~> 0.2"},
   ]
  end

• Agent version (mix deps | grep new_relic_agent):

* new_relic_agent (Hex package) (mix)
  locked at 1.10.1 (new_relic_agent) e200b5d2

Generate the following error :

[error] GenServer NewRelic.Harvest.Collector.Metric.HarvestCycle terminating
** (ArgumentError) argument error
    (stdlib) :ets.lookup(NewRelic.Harvest.Collector.AgentRun, :data_report_period)
    (new_relic_agent) lib/new_relic/harvest/collector/agent_run.ex:108: NewRelic.Harvest.Collector.AgentRun.lookup/1
    (new_relic_agent) lib/new_relic/harvest/collector/harvest_cycle.ex:124: NewRelic.Harvest.Collector.HarvestCycle.trigger_
harvest_cycle/1
    (new_relic_agent) lib/new_relic/harvest/collector/harvest_cycle.ex:63: NewRelic.Harvest.Collector.HarvestCycle.handle_in
fo/2
    (stdlib) gen_server.erl:616: :gen_server.try_dispatch/4
    (stdlib) gen_server.erl:686: :gen_server.handle_msg/6
    (stdlib) proc_lib.erl:247: :proc_lib.init_p_do_apply/3
Last message: :harvest_cycle
State: %{harvest_cycle_key: :data_report_period, harvester: nil, name: NewRelic.Harvest.Collector.Metric.HarvestCycle, super
visor: NewRelic.Harvest.Collector.Metric.HarvesterSupervisor, timer: nil}
[warn] Logger dropped 180 OTP/SASL messages as it had 555 messages in its inbox, exceeding the amount of :discard_threshold
500 messages. The number of messages was reduced to 375 (75% of the threshold)

Any ideas ?
Can it be related to #138 ?

[Yoshyn]

As the versions has several commits, I tried to check by commit individually :

{:new_relic_agent, github: "newrelic/elixir_agent", ref: "ref_of_the_commit", override: true},

And the error occur at b742e39

So my it's not related to OTP (#138). I'll check to get a better understanding on what happen.

From the console :

iex(0)>NewRelic.Config.enabled?()
false
iex(1)>payload = NewRelic.Harvest.Collector.Connect.payload()
iex(2)>NewRelic.Harvest.Collector.Protocol.connect(payload)
{:error, :harvest_disabled}

Seem that this error is not handle

[binaryseed]

Thanks, I'll take a look today

[binaryseed]

Hi, I was able to reproduce this by running against OTP 20.

It's caused because AgentRun leverages handle_continue which was added to GenServer in OTP 21. In 20 that GenServer can't boot so the ETS table doesn't get created so trying to access it causes a crash...

I'll see what I can do to get the OTP version requirements front and center so this can't happen...

Can you upgrade your OTP version?

[binaryseed]

This PR will raise a helpful error if the app is started pre OTP 21 #151

[Yoshyn]

Thank you. I get a better understanding now. I wondering what was the handle_continue.
And the error was not clear. It's clearly better !

Can you upgrade your OTP version?

😱 😢 😨 I'm afraid but this will not be possible in my case.

I'll try to find an alternative solution (maybe fork the gem and reimplement the AgentRun to OTP 20).

[binaryseed]

FYI, the upcoming Elixir 1.10 release will also be OTP 21+ only.