Connection recovery works after login but fails on initial connect when API is offline (EMA 3.9.1.1)

We’re using Refinitiv EMA version 3.9.1.1 and running into an issue with initial connection recovery.

The traditional config options for reconnect handling (ReconnectAttemptLimit, ReconnectMinDelay, etc.) don’t seem to work in this version.

Scenario:
We’re simulating API loss by routing our EMA connection through a local proxy.

If the proxy is initially online and we successfully log in, everything works as expected: killing and restarting the proxy does not disrupt the connection permanently, EMA automatically restores the session.

However, if we start the application with the proxy offline, after about 45 seconds we see a login timeout, and starting the proxy after this point does not trigger reconnection. The API becomes effectively unusable.

Relevant logs during initial connection attempt (proxy offline):

13:21:39.737 [ema-api-1] TRACE com.refinitiv.ema.access.OmmConsumerImpl - loggerMsg
ClientName: ChannelCallbackClient
Severity: Trace
Text: Attempt to connect using
1] SOCKET
Channel name ALC_Primary
Instance Name PriceConsumer_1
RsslReactor @34ea22e1
InterfaceName
hostName localhost
port 14002
CompressionType None
tcpNodelay true
reconnectAttemptLimit -1
reconnectMinDelay 1000 msec
reconnectMaxDelay 5000 msec
guaranteedOutputBuffers 100
numInputBuffers 100
sysRecvBufSize 65535
sysSendBufSize 65535
connectionPingTimeout 30000 msec
initializationTimeout 5 sec

loggerMsgEnd

13:21:39.880 [ema-api-1] TRACE com.refinitiv.ema.access.OmmConsumerImpl - loggerMsg
ClientName: ChannelCallbackClient
Severity: Trace
Text: Received ChannelOpened on channel ALC_Primary
Instance Name PriceConsumer_1
loggerMsgEnd

13:21:39.882 [ema-api-1] TRACE com.refinitiv.ema.access.OmmConsumerImpl - loggerMsg
ClientName: ChannelCallbackClient
Severity: Trace
Text: Successfully created a Reactor and Channel(s)
Channel name(s) ALC_Primary
Instance Name PriceConsumer_1
loggerMsgEnd

13:21:39.887 [ema-api-1] WARN com.refinitiv.ema.access.OmmConsumerImpl - loggerMsg
ClientName: ChannelCallbackClient
Severity: Warning
Text: Received ChannelDownReconnecting event on channel ALC_Primary
RsslReactor @34ea22e1
RsslChannel @5312dd68
Error Id 0
Internal sysError 0
Error Location Reactor.processWorkerEvent
Error text Error initializing channel: errorId=-1 text=Connection refused: no further information
loggerMsgEnd

13:21:39.893 [ema-api-1] DEBUG com.yieldbroker.ema.processor.EmaPriceListener - [onStatusMsg] Received message: StatusMsg
streamId="1"
domain="Login Domain"
state="Open / Suspect / None / 'channel down'"
StatusMsgEnd

After the login timeout expires we get:

13:22:24.884 [ema-api-1] ERROR com.refinitiv.ema.access.OmmConsumerImpl - loggerMsg
ClientName: PriceConsumer_1
Severity: Error
Text: login failed (timed out after waiting 45000 milliseconds) for localhost:14002)
loggerMsgEnd13:22:24.887 [ema-api-1] ERROR com.refinitiv.ema.access.OmmConsumerImpl - loggerMsg
ClientName: ChannelCallbackClient
Severity: Error
Text: Received ChannelDown event on channel ALC_Primary
Instance Name PriceConsumer_1
RsslReactor Channel is null
Error Id 0
Internal sysError 0
Error Location null
Error text Reactor shutting down...
loggerMsgEnd

Is there a recommended way to handle initial login retries in EMA 3.9.1.1 without breaking the automatic reconnection that works after a successful login? Essentially, we want the client to keep retrying to log in until the proxy/API becomes available, but still leverage the built-in automatic recovery after login.

Thanks!

Find more posts tagged with

java

connection

recovery

ema-api

#technology

Accepted answers

Jirapongse

@dgoloughlin

Thank you for reaching out to us.

This is an expected behavior, as mentioned in the LoginRequestTimeOut configuration.

EMA will throw an exception if it doesn't receive a login response message from a server after this amount of time.

You can set this configuration to 0 to wait for a response indefinitely.

All comments

Jirapongse

@dgoloughlin

Thank you for reaching out to us.

This is an expected behavior, as mentioned in the LoginRequestTimeOut configuration.

EMA will throw an exception if it doesn't receive a login response message from a server after this amount of time.

You can set this configuration to 0 to wait for a response indefinitely.

dgoloughlin

Following up on my earlier question. I’ve successfully configured LoginRequestTimeout to 0 so that the consumer retries login indefinitely. That’s now working perfectly - thanks!

However, I’m seeing that only the first channel (the one referenced by the DefaultConsumer) is ever used. The other channels defined in the <ChannelList> are never tried, even if the first one is unavailable.

From my understanding, I expected that if the first channel failed (or was unreachable), EMA would automatically attempt to connect using the other channels defined in the configuration, but that doesn’t seem to happen.

Have I misunderstood the purpose of the <ChannelList>? Is there a way to configure EMA so that it automatically cycles through the defined channels until one succeeds?

Here's my config:

<EmaConfig>

    <ConsumerGroup>

        <DefaultConsumer value="PriceConsumer"/>

        <ConsumerList>

            <Consumer>

                <Name value="PriceConsumer"/>

                <Channel value="ALC_Primary"/>

                <Dictionary value="Dictionary_1"/>

                <XmlTraceToStdout value="0"/>

                <LoginRequestTimeOut value="0"/>

            </Consumer>

        </ConsumerList>

    </ConsumerGroup>


    <ChannelGroup>

        <ChannelList>

            <Channel>

                <Name value="ALC_Primary"/>

                <ChannelType value="ChannelType::RSSL_SOCKET"/>

                <Host value="localhost"/>

                <Port value="14002"/>

            </Channel>


            <Channel>

                <Name value="ALC_Secondary"/>

                <ChannelType value="ChannelType::RSSL_SOCKET"/>

                <Host value="xxx.xxx.xxx.xxx"/>

                <Port value="14002"/>

            </Channel>


            <Channel>

                <Name value="IPSec_Japan_1"/>

                <ChannelType value="ChannelType::RSSL_SOCKET"/>

                <Host value="xxx.xxx.xxx.xxx"/>

                <Port value="14002"/>

            </Channel>


            <Channel>

                <Name value="IPSec_Japan_2"/>

                <ChannelType value="ChannelType::RSSL_SOCKET"/>

                <Host value="xxx.xxx.xxx.xxx"/>

                <Port value="14002"/>

            </Channel>

        </ChannelList>

    </ChannelGroup>


    <DictionaryGroup>

        <DictionaryList>

            <Dictionary>

                <Name value="Dictionary_1"/>

                <DictionaryType value="DictionaryType::FileDictionary"/>

                <RdmFieldDictionaryFileName value="./etc/RDMFieldDictionary"/>

                <EnumTypeDefFileName value="./etc/enumtype.def"/>

            </Dictionary>

        </DictionaryList>

    </DictionaryGroup>

</EmaConfig>

dgoloughlin

https://community.developers.refinitiv.com/discussion/comment/128798#Comment_128798

I'll answer my own question - I was missing the <ChannelSet value="ALC_Primary, ALC_Secondary"/> from my config file.

Everything is now working as expected, thankyou again!

EXPLORE OUR SITES