Troubleshooting connection issues in Amazon Redshift - Amazon Redshift
Connecting from outside of Amazon EC2 and encountering a firewall timeout issueConnection is refused or failsClient and driver are incompatibleQueries appear to hang and sometimes fail to reach the clusterSetting the JDBC fetch size parameter

Troubleshooting connection issues in Amazon Redshift

If you have issues with connecting to your cluster from a SQL client tool, there are several things that you can check to narrow down the problem. If you are using SSL or server certificates, first remove this complexity while you troubleshoot the connection issue. Then add this back when you have found a solution. For more information, see Configuring security options for connections.

Important

Amazon Redshift has changed the way that SSL certificates are managed. If you have trouble connecting using SSL, you might need to update your current trust root CA certificates. For more information, see Transitioning to ACM certificates for SSL connections.

The following section has some example error messages and possible solutions for connection issues. Because different SQL client tools provide different error messages, this is not a complete list, but should be a good starting point for troubleshooting issues.

Connecting from outside of Amazon EC2 and encountering a firewall timeout issue

Your client connection to the database appears to hang or timeout when running long queries, such as a COPY command. In this case, you might observe that the Amazon Redshift console displays that the query has completed, but the client tool itself still appears to be running the query. The results of the query might be missing or incomplete depending on when the connection stopped.

Possible solutions

This issue happens when you connect to Amazon Redshift from a machine other than an Amazon EC2 instance. In this case, idle connections are terminated by an intermediate network component, such as a firewall, after a period of inactivity. This behavior is typical when you log on from a virtual private network (VPN) or your local network.

To avoid these timeouts, we recommend the following changes:

  • Increase client system values that deal with TCP/IP timeouts. Make these changes on the computer you are using to connect to your cluster. The timeout period should be adjusted for your client and network. For more information, see Change TCP/IP timeout settings.

  • Optionally, set keepalive behavior at the DSN level. For more information, see Change DSN timeout settings.

Change TCP/IP timeout settings

To change TCP/IP timeout settings, configure the timeout settings according to the operating system that you use to connect to your cluster.

  • Linux — If your client is running on Linux, run the following command as the root user to change the timeout settings for the current session: 

    /sbin/sysctl -w net.ipv4.tcp_keepalive_time=200 net.ipv4.tcp_keepalive_intvl=200 net.ipv4.tcp_keepalive_probes=5

    To persist the settings, create or modify the file /etc/sysctl.conf with the following values then reboot your system. 

    net.ipv4.tcp_keepalive_time=200
net.ipv4.tcp_keepalive_intvl=200
net.ipv4.tcp_keepalive_probes=5

  • Windows — If your client runs on Windows, edit the values for the following registry settings under HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters\:

    • KeepAliveTime: 30000

    • KeepAliveInterval: 1000

    • TcpMaxDataRetransmissions: 10

    These settings use the DWORD data type. If they do not exist under the registry path, you can create the settings and specify these recommended values. For more information about editing the Windows registry, refer to Windows documentation.

    After you set these values, restart your computer for the changes to take effect.

  • Mac — If your client is running on a Mac, run the following commands to change the timeout settings for the current session: 

    sudo sysctl net.inet.tcp.keepintvl=200000
sudo sysctl net.inet.tcp.keepidle=200000
sudo sysctl net.inet.tcp.keepinit=200000
sudo sysctl net.inet.tcp.always_keepalive=1

    To persist the settings, create or modify the file /etc/sysctl.conf with the following values: 

    net.inet.tcp.keepidle=200000
net.inet.tcp.keepintvl=200000
net.inet.tcp.keepinit=200000
net.inet.tcp.always_keepalive=1

    Restart your computer, and then run the following commands to verify that the values are set. 

    sysctl net.inet.tcp.keepidle
sysctl net.inet.tcp.keepintvl
sysctl net.inet.tcp.keepinit
sysctl net.inet.tcp.always_keepalive

Change DSN timeout settings

You can set keepalive behavior at the DSN level if you choose. You do this by adding or modifying the following parameters in the odbc.ini file:

KeepAlivesCount

The number of TCP keepalive packets that can be lost before the connection is considered broken.

KeepAlivesIdle

The number of seconds of inactivity before the driver sends a TCP keepalive packet.

KeepAlivesInterval

The number of seconds between each TCP keepalive retransmission.

If these parameters don't exist, or if they have a value of 0, the system uses the keepalive parameters specified for TCP/IP to determine DSN keepalive behavior. On Windows, you can find the TCP/IP parameters in the registry in HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters\. On Linux and macOS, you can find the TCP/IP parameters can be found in the sysctl.conf file.

Connection is refused or fails

When your connection if refused or fails, you may receive an error similar to one of the following.

  • "Failed to establish a connection to <endpoint>."

  • "Could not connect to server: Connection timed out. Is the server running on host '<endpoint>' and accepting TCP/IP connections on port '<port>'?"

  • "Connection refused. Check that the hostname and port are correct and that the postmaster is accepting TCP/IP connections."

Possible solutions

Generally, when you receive an error message indicating that there is a failure to establish a connection, it is an issue with permission to access the cluster or with network traffic reaching the cluster.

To connect to the cluster from a client tool outside of the network that the cluster is in, you add a inbound rule to the cluster's security group. The rule configuration depends on whether the Amazon Redshift cluster is created in a virtual private cloud (VPC):

  • If you created the Amazon Redshift cluster in a virtual private cloud (VPC) based on Amazon VPC, add an inbound rule to the VPC security group that specifies the client CIDR/IP address, in Amazon VPC. For more information about configuring the VPC security groups for your cluster and publicly accessible options, see Redshift resources in a VPC.

  • If you created your Amazon Redshift cluster outside a VPC, add your client CIDR/IP address to the cluster security group in Amazon Redshift. For more information about configuring cluster security groups, see Amazon Redshift security groups.

If you attempt to connect to the cluster from a client tool that runs on an Amazon EC2 instance, you also add an inbound rule. In this case, add a rule to the cluster security group. The rule must specify the Amazon EC2 security group associated with the client tool's Amazon EC2 instance.

In some cases, you might have a layer between your client and server, such as a firewall. In these cases, make sure that the firewall accepts inbound connections over the port that you configured for your cluster.

Client and driver are incompatible

If your client and driver are incompatible, you may receive an error that says, "The specified DSN contains an architecture mismatch between the Driver and Application."

Possible solutions

When you attempt to connect and get an error about an architecture mismatch, this means that the client tool and the driver aren't compatible. This occurs because their system architecture doesn't match. For example, this can happen if you have a 32-bit client tool but have installed the 64-bit version of the driver. Sometimes 64-bit client tools can use 32-bit drivers, but you can't use 32-bit applications with 64-bit drivers. Make sure that the driver and client tool are using the same version of the system architecture.

Queries appear to hang and sometimes fail to reach the cluster

You experience an issue with queries completing, where the queries appear to be running but hang in the SQL client tool. Sometimes the queries fail to appear in the cluster, such as in system tables or the Amazon Redshift console.

Possible solutions

This issue can happen due to packet drop. In this case, there is a difference in the maximum transmission unit (MTU) size in the network path between two Internet Protocol (IP) hosts. The MTU size determines the maximum size, in bytes, of a packet that can be transferred in one Ethernet frame over a network connection. In AWS, some Amazon EC2 instance types support an MTU of 1500 (Ethernet v2 frames) and other instance types support an MTU of 9001 (TCP/IP jumbo frames).

To avoid issues that can occur with differences in MTU size, we recommend doing one of the following:

  • If your cluster uses the EC2-VPC platform, configure the Amazon VPC security group with an inbound custom Internet Control Message Protocol (ICMP) rule that returns Destination Unreachable. The rule thus instructs the originating host to use the lowest MTU size along the network path. For details on this approach, see Configuring security groups to allow ICMP "destination unreachable".

  • If your cluster uses the EC2-Classic platform, or you can't allow the ICMP inbound rule, disable TCP/IP jumbo frames so that Ethernet v2 frames are used. For details on this approach, see Configuring the MTU of an instance.

Configuring security groups to allow ICMP "destination unreachable"

When there is a difference in the MTU size in the network between two hosts, first make sure that your network settings don't block path MTU discovery (PMTUD). PMTUD enables the receiving host to respond to the originating host with the following ICMP message: Destination Unreachable: fragmentation needed and DF set (ICMP Type 3, Code 4). This message instructs the originating host to use the lowest MTU size along the network path to resend the request. Without this negotiation, packet drop can occur because the request is too large for the receiving host to accept. For more information about this ICMP message, go to RFC792 on the Internet Engineering Task Force (IETF) website.

If you don't explicitly configure this ICMP inbound rule for your Amazon VPC security group, PMTUD is blocked. In AWS, security groups are virtual firewalls that specify rules for inbound and outbound traffic to an instance. For information about Amazon Redshift cluster security group, see Amazon Redshift security groups. For clusters using the EC2-VPC platform, Amazon Redshift uses VPC security groups to allow or deny traffic to the cluster. By default, the security groups are locked down and deny all inbound traffic. For information about how to set inbound and outbound rules for EC2-Classic or EC2-VPC instances, see Differences between instances in EC2-Classic and a VPC in the Amazon EC2 User Guide.

For more information about how to add rules to VPC security groups, see VPC security groups. For more information about specific PMTUD settings required in this rule, see Path MTU discovery in the Amazon EC2 User Guide.

Configuring the MTU of an instance

In some cases, your cluster might use the EC2-Classic platform or you can't allow the custom ICMP rule for inbound traffic. In these cases, we recommend that you adjust the MTU to 1500 on the network interface (NIC) of the EC2 instances you connect to your Amazon Redshift cluster from. This adjustment disables TCP/IP jumbo frames to ensure that connections consistently use the same packet size. However, this option reduces your maximum network throughput for the instance entirely, not just for connections to Amazon Redshift. For more information, see the following procedures.

To set MTU on a Microsoft Windows operating system

If your client runs in a Microsoft Windows operating system, you can review and set the MTU value for the Ethernet adapter by using the netsh command.

  1. Run the following command to determine the current MTU value: 

    netsh interface ipv4 show subinterfaces

  2. Review the MTU value for the Ethernet adapter in the output.

  3. If the value is not 1500, run the following command to set it: 

    netsh interface ipv4 set subinterface "Ethernet" mtu=1500 store=persistent

    After you set this value, restart your computer for the changes to take effect.

To set MTU on a Linux operating system

If your client runs in a Linux operating system, you can review and set the MTU value by using the ip command.

  1. Run the following command to determine the current MTU value: 

    $ ip link show eth0

  2. Review the value following mtu in the output.

  3. If the value is not 1500, run the following command to set it: 

    $ sudo ip link set dev eth0 mtu 1500
To set MTU on a Mac operating system

  • Follow instructions on the MacOS support site about How to change the MTU for troubleshooting purposes. For more information, search the support site.

Setting the JDBC fetch size parameter

By default, the JDBC driver collects all the results for a query at one time. As a result, when you attempt to retrieve a large result set over a JDBC connection, you might encounter a client-side out-of-memory error. To enable your client to retrieve result sets in batches instead of in a single all-or-nothing fetch, set the JDBC fetch size parameter in your client application.

Note

Fetch size is not supported for ODBC.

For the best performance, set the fetch size to the highest value that does not lead to out of memory errors. A lower fetch size value results in more server trips, which prolong execution times. The server reserves resources, including the WLM query slot and associated memory, until the client retrieves the entire result set or the query is canceled. When you tune the fetch size appropriately, those resources are released more quickly, making them available to other queries.

Note

If you need to extract large datasets, we recommend using an UNLOAD statement to transfer the data to Amazon S3. When you use UNLOAD, the compute nodes work in parallel to speed up the transfer of data.

For more information about setting the JDBC fetch size parameter, go to Getting results based on a cursor in the PostgreSQL documentation.