Tiny topology shows '0' for all stats.

[Niels Basjes]

Hi,

I have a very small topology here.

In fact this is a thing that generates synthetic data and puts it into Kafka.

When looking at the web UI I see that all counters (i.e. Bytes received, Records received, Bytes sent,  Records sent) all remain 0.

I verified and I'm seeing thousands of records arriving into Kafka.

Is this a bug in Flink or am I misinterpreting the meaning of these numbers?

--

[Ufuk Celebi]

> On 14 Dec 2015, at 16:25, Niels Basjes <[hidden email]> wrote:
>
> Hi,
>
> I have a very small topology here.
> In fact this is a thing that generates synthetic data and puts it into Kafka.
> When looking at the web UI I see that all counters (i.e. Bytes received, Records received, Bytes sent,  Records sent) all remain 0.
> I verified and I'm seeing thousands of records arriving into Kafka.
>
> Is this a bug in Flink or am I misinterpreting the meaning of these numbers?

Sources and sinks do not show the number of received or sent records, because of the internals of how these numbers are collected. I agree that this is confusing. Big +1 to improve this.

You actually don’t see any numbers, because the operators are chained and hence you only have one task, which acts as both source and sink.

You should see some sent and received numbers if you break up the chain (env.disableOperatorChaining()). Can you confirm this?

– Ufuk

[Stephan Ewen]

To sum this up, the web dashboard stats are Flink network stats. If you job has no network communication, its all zero.

On Mon, Dec 14, 2015 at 5:03 PM, Ufuk Celebi <[hidden email]> wrote:

> On 14 Dec 2015, at 16:25, Niels Basjes <[hidden email]> wrote:
>
> Hi,
>
> I have a very small topology here.
> In fact this is a thing that generates synthetic data and puts it into Kafka.
> When looking at the web UI I see that all counters (i.e. Bytes received, Records received, Bytes sent,  Records sent) all remain 0.
> I verified and I'm seeing thousands of records arriving into Kafka.
>
> Is this a bug in Flink or am I misinterpreting the meaning of these numbers?

Sources and sinks do not show the number of received or sent records, because of the internals of how these numbers are collected. I agree that this is confusing. Big +1 to improve this.

You actually don’t see any numbers, because the operators are chained and hence you only have one task, which acts as both source and sink.

You should see some sent and received numbers if you break up the chain (env.disableOperatorChaining()). Can you confirm this?

– Ufuk

[Niels Basjes]

Hi,

@Ufuk: I added the env.disableOperatorChaining() and indeed now I see two things on the screen and there are numbers counting what has happened.

@Stephan: Yes, I understand these numbers now.

I found that this is already a jira ticket to add what I was looking for:

https://issues.apache.org/jira/browse/FLINK-2944

https://issues.apache.org/jira/browse/FLINK-3130

Niels

On Mon, Dec 14, 2015 at 5:03 PM, Ufuk Celebi <[hidden email]> wrote:

> On 14 Dec 2015, at 16:25, Niels Basjes <[hidden email]> wrote:
>
> Hi,
>
> I have a very small topology here.
> In fact this is a thing that generates synthetic data and puts it into Kafka.
> When looking at the web UI I see that all counters (i.e. Bytes received, Records received, Bytes sent,  Records sent) all remain 0.
> I verified and I'm seeing thousands of records arriving into Kafka.
>
> Is this a bug in Flink or am I misinterpreting the meaning of these numbers?

Sources and sinks do not show the number of received or sent records, because of the internals of how these numbers are collected. I agree that this is confusing. Big +1 to improve this.

You actually don’t see any numbers, because the operators are chained and hence you only have one task, which acts as both source and sink.

You should see some sent and received numbers if you break up the chain (env.disableOperatorChaining()). Can you confirm this?

– Ufuk

--

[Nick Dimiduk]

For my own understanding, are you suggesting the FLINK-2944 (or a subtask) is the appropriate place to implement exposure of metrics such as bytes, records in, out of Streaming sources and sinks?

On Tue, Dec 15, 2015 at 5:24 AM, Niels Basjes <[hidden email]> wrote:

Hi,

@Ufuk: I added the env.disableOperatorChaining() and indeed now I see two things on the screen and there are numbers counting what has happened.

@Stephan: Yes, I understand these numbers now.

I found that this is already a jira ticket to add what I was looking for:

https://issues.apache.org/jira/browse/FLINK-2944

https://issues.apache.org/jira/browse/FLINK-3130

Niels

On Mon, Dec 14, 2015 at 5:03 PM, Ufuk Celebi <[hidden email]> wrote:

> On 14 Dec 2015, at 16:25, Niels Basjes <[hidden email]> wrote:
>
> Hi,
>
> I have a very small topology here.
> In fact this is a thing that generates synthetic data and puts it into Kafka.
> When looking at the web UI I see that all counters (i.e. Bytes received, Records received, Bytes sent,  Records sent) all remain 0.
> I verified and I'm seeing thousands of records arriving into Kafka.
>
> Is this a bug in Flink or am I misinterpreting the meaning of these numbers?

Sources and sinks do not show the number of received or sent records, because of the internals of how these numbers are collected. I agree that this is confusing. Big +1 to improve this.

You actually don’t see any numbers, because the operators are chained and hence you only have one task, which acts as both source and sink.

You should see some sent and received numbers if you break up the chain (env.disableOperatorChaining()). Can you confirm this?

– Ufuk

--

Best regards / Met vriendelijke groeten,

Niels Basjes

[Niels Basjes]

@Nick: That is what I read in the current description and comments.

On Tue, Dec 15, 2015 at 6:25 PM, Nick Dimiduk <[hidden email]> wrote:

For my own understanding, are you suggesting the FLINK-2944 (or a subtask) is the appropriate place to implement exposure of metrics such as bytes, records in, out of Streaming sources and sinks?

On Tue, Dec 15, 2015 at 5:24 AM, Niels Basjes <[hidden email]> wrote:

Hi,

@Ufuk: I added the env.disableOperatorChaining() and indeed now I see two things on the screen and there are numbers counting what has happened.

@Stephan: Yes, I understand these numbers now.

I found that this is already a jira ticket to add what I was looking for:

https://issues.apache.org/jira/browse/FLINK-2944

https://issues.apache.org/jira/browse/FLINK-3130

Niels

On Mon, Dec 14, 2015 at 5:03 PM, Ufuk Celebi <[hidden email]> wrote:

> On 14 Dec 2015, at 16:25, Niels Basjes <[hidden email]> wrote:
>
> Hi,
>
> I have a very small topology here.
> In fact this is a thing that generates synthetic data and puts it into Kafka.
> When looking at the web UI I see that all counters (i.e. Bytes received, Records received, Bytes sent,  Records sent) all remain 0.
> I verified and I'm seeing thousands of records arriving into Kafka.
>
> Is this a bug in Flink or am I misinterpreting the meaning of these numbers?

Sources and sinks do not show the number of received or sent records, because of the internals of how these numbers are collected. I agree that this is confusing. Big +1 to improve this.

You actually don’t see any numbers, because the operators are chained and hence you only have one task, which acts as both source and sink.

You should see some sent and received numbers if you break up the chain (env.disableOperatorChaining()). Can you confirm this?

– Ufuk

--

Best regards / Met vriendelijke groeten,

Niels Basjes

--