To totally unlock this section you need to Log-in
Remote Procedure Call (RPC) is an inter-process communication technique to allow client and server software to communicate on a network. The RPC protocol is based on a client/server model. The client makes a procedure call that appears to be local but is actually run on a remote computer. During this process, the procedure call arguments are bundled and passed through the network to the server. The arguments are then unpacked and run on the server. The result is again bundled and passed back to the client, where it is converted to a return value for the client's procedure call.
RPC is used by several components in Windows Server, such as the File Replication Service (FRS), Active Directory Replication, Certificate services, DCOM, domain join, DCPromo and RDP, NLB and Cluster, Microsoft Operations Master, Exchange and SQL Server.
The RPC Server
An RPC server is a communications interface provided by an application or service that allows remote clients to connect, pass commands, and transfer data using the RPC protocol. A typical example of an RPC server is Microsoft Exchange Server. Microsoft Exchange Server is an application running on a computer that supplies an RPC communications interface for an RPC client.
An application will register its RPC server with the operating system’s End Point Mapper (EPM) service so that the remote client can locate the RPC server. When the application registers with the EPM it will indicate the IP address and TCP port that it is listening on.
The RPC Client
An RPC client is an application running on any given computer that uses the RPC protocol to communicate with an RPC server. An example of a typical RPC client is the Microsoft Outlook application.
NOTE: In this article the terms RPC server and RPC client refer to the application running at both ends of an RPC communication.
RPC Quick Fixes
Common causes of RPC errors include:
- Errors resolving a DNS or NetBIOS name.
- The RPC service or related services may not be running.
- Number of connectivity Problems with network connectivity.
- File and printer sharing is not enabled.
Unable to resolve DNS or NetBIOS names in an Active Directory environment
Use the following commands to verify DNS is working for all DC's or specific DC's:
To get a DNS status for all DCs in forest, run the following command:
DCDIAG /TEST:DNS /V /E /F:<filename .log>
The "/e" switch runs the DNS test against all DCs in an Active Directory Forest.
To get DNS health on a single DC, run the command below:
DCDIAG /TEST:DNS /V /S:<dcname> /F:<filename .log>
The "/s:" switch runs the DNS test against a specified domain controller.
To verify that a domain controller can be located for a specific domain, run the command below.
NLTEST /DSGETDC:<netbios or DNS domain name>
- Servers and clients that are receiving the error should be checked to verify that they are configured with the appropriate DNS server.
- Servers should not be pointing to their ISP's DNS servers in the preferred or alternate DNS server portion of the TCP/IP settings.
- The ISP's DNS servers should only be used as forwarders in DNS.
Ensure that at least one correct DNS record is registered on each domain controller:
- To ensure that a correct DNS record is registered on each domain controller, find this server's Active Directory replication partners that run DNS.
- Open DNSManager and connect in turn to each of these replication partners.
- Find the host (A) resource record registration for this server on each of the other replication partner domain controllers.
- Delete those host (A) records that do not have IP addresses corresponding to any of this server's IP addresses.
- If a domain controller has no host (A) records for this server, add at least one that corresponds to an IP address on this server. (If there are multiple IP addresses for this server, add at least one that is on the same network as the domain controller you are updating.)
Name resolution may also fail with the RPC Server is unavailable error if NetBIOS over TCP/IP is disabled on the WINS tab in the advanced section of the TCP/IP properties. The NetBIOS over TCP/IP setting should be either enabled or default (use DHCP).
Verify that a single label domain name is not being configured. DNS names that do not contain a suffix such as .com, .corp, .net, .org or .local are considered to be single-label DNS names. Microsoft doesn't recommend using single label domain names because they cannot be registered with an Internet registrar and domain members do not perform dynamic updates to single-label DNS zones.
The RPC service or related services may not be started
Verify the status and startup type for the RPC and RPC locator services on the server that gets the error:
- By default, Windows server 2003 domain controllers (and beyond) and member servers all should have the RPC service started and set to Automatic startup and the RPC Locator service stopped and set to Manual Startup.
- Windows 2000 (and beyond) domain controllers should have the RPC and RPC Locator services both set to started and automatic startup, while Windows 2000 member servers should have the RPC service started and set to automatic startup while the RPC locator service should be started and set to manual startup.
- If you make any changes to the RPC service or to the RPC Locator service settings, restart the computer, and then test for the problem again.
- Additional Services that may result in "The RPC Server is Unavailable" errors are the TCP/IP NetBIOS helper service, Distributed File System service and Remote Registry service. These services should both be set to automatic and started. The Kerberos Key Distribution Center (KDC) should be Started and Automatic on Windows 2000, Windows 2003 DCs (and beyond). It should not be started and set to Disabled in all other cases.
Verify ports needed by RPC are open: verify that ports greater than 1024 are not blocked. Clients connect to RPC Endpoint Mapper on port 135. RPC Endpoint Mapper then tells the client which randomly assigned port between 1024-65535 a requested service is listening on.
Ports may be blocked by a hardware firewall or a software firewall. Software firewalls include Internet Connection Firewall on computers running Windows Server 2003 or Windows XP, and Windows Firewall on computers running Windows Vista, Windows 7, Windows Server 2008 and Windows Server 2008 R2.
A computer might also have third-party firewall software installed, or antivirus software with built-in firewall functionality. By default, port 135 TCP/UDP and ports 1024-65535 TCP must be open for RPC to work. You can restrict the ports greater than 1024 that RPC uses. However, RPC Endpoint Mapper is always on port 135.
File and Printer Sharing is not enabled
File and Printer sharing for Microsoft Networks will produce the error RPC Server is unavailable when you try to view or manage services on a remote computer using the Services snap-in. See the following example:
Unable to open service control manager database on \\
. Error 1722: The RPC server is unavailable.
This error message may occur if the File and Printer Sharing for Microsoft Networks component is not enabled on the remote computer.
The process of an RPC client connecting to an RPC server can be broken down into four phases. This troubleshooting guide will discuss the events that occur at each phase, how to test these events, and how to identify if the phase completed successfully.
- Phase 1: Name Resolution: Name resolution is the act of resolving a name to an IP address. This normally takes two forms: NetBIOS Name Resolution or the more common DNS Name Resolution.
- Phase 2: TCP session establishment: TCP session establishment is the act of establishing a TCP connection between the RPC client and the RPC server. TCP sessions will be initiated by the RPC client via a TCP 3-way handshake with the RPC server.
- Phase 3: RPC Discovery: When a client wants to connect to the RPC server supplied by the application it will contact the computer that hosts the RPC Server and discover how to connect to the RPC Server.
- Phase 4: RPC Communication: RPC Communication is the act of making RPC requests to the application endpoint and receiving RPC responses from this application.
Data needed to troubleshoot the issue:
Identify the client and server computers reporting the RPC error. Identify the DNS and WINS servers used by these computers. To do this:
- On each machine, open a command prompt and run ipconfig /all.
- Determine the IP address of both machines. If the server is part of a cluster get the cluster resource IP address as well. Identify the DNS servers and WINS servers that the RPC client is configured to use.
Note: You can also obtain this information by opening Control Panel\Network and Sharing Center, clicking Local Area Connection and selecting Properties.
- Identify the application(s) reporting RPC Server Unavailable.
- Simultaneous network traces (using Wireshark, Netmon, or a comparable network sniffer) from the machines hosting the RPC client and RPC Server while reproducing the task that results in a “RPC Server Unavailable” error.
- The network captures on both hosts should be started first.
- From a command prompt on the client run ipconfig /flushdns and nbtstat –R to clear the name resolution caches.
- Reproduce the error.
- Stop the traces and save them.
Name Resolution consists of one or possibly more NetBIOS or DNS queries to locate the IP address for the RPC Server. Troubleshooting this phase requires verifying that a response is received to the name resolution request and that the response contains the correct IP address for the RPC server. Compare the IP address reported by DNS or NetBIOS in the network trace for the server with the IP addresses you noted earlier. If it does not match then check DNS and WINS and note if there is a difference.
DNS Name Resolution
To identify DNS Name Resolution in a network trace use the following filter in Network Monitor or Wireshark: dns. DNS resolution will be occurring at the client so open the network trace taken from the RPC client machine. You will be looking for one packet that is the query from the client to the DNS server and then the response packet from the DNS server. It will look similar to this:
If the trace shows the correct IP address for the RPC server was returned by the DNS server proceed to TCP Session Establishment.
NetBIOS Name Resolution
NetBIOS queries come in two forms, WINS or NetBIOS Broadcasts. WINS will consist of a unicast query to a WINS server and a response from the WINS server.
NetBIOS broadcasts are queries broadcast to all hosts on the local subnet so name resolution is limited to only hosts on the subnet. The host with the name listed in the NetBIOS Broadcast will respond with its IP address.
To identify NetBIOS Name Resolution in a network trace, use the following filter in Network Monitor - “nbtns”. For Wireshark, use the following filter - nbns”. If the trace shows a successful resolution using WINS or NetBIOS queries proceed to TCP Session Establishment.
TCP Session Establishment
TCP Sessions always begin with a TCP 3-way handshake. The handshake should look similar to what is shown below. The RPC Client will send the first packet, known as the SYN packet. The computer hosting the RPC Server will send a SYN/ACK response, and then the RPC Client will send an ACK packet.
Scenarios that may cause the TCP session to fail:
If a firewall or network problem is the culprit, it is likely a failure will occur during this phase. To diagnose this you will want to look at the network traces taken from the RPC Client and RPC Server. If a firewall or other network device is causing a problem it will usually manifest as a retransmit of the TCP SYN packet by the RPC Client about 3 seconds after the first TCP SYN is sent.
This can be seen in a Netmon network trace using the display filter specification of “tcpsynretransmit==1”. In other cases, firewalls will allow the 3-way handshake to succeed but may block the RPC packets due to the contents of the packet at a higher level. In these cases it is possible to see the retransmit of the RPC packet within half a second of the original packet being sent.
To identify this condition in a Netmon network trace use the display filter specification of “tcpretransmit==1”.
To see either of these retransmit conditions in a trace taken using Wireshark use the display filter specification of “tcp.analysis.retransmission”.
The RPC Server is not actively listening
It was noted earlier that an RPC Server will register itself and listen on a particular port and IP address of the host computer. If for some reason that fails the TCP layer will answer the SYN packet from the client with a Reset packet.
A device in the middle between the RPC Client and RPC Server will be resetting the connection attempt.
In the client side trace it will appear as if the server sent the TCP Reset while the trace from the server indicates the client is the source of the TCP Reset.
For both these scenarios, check for the presence of a Reset packet in the TCP three way handshake by using the display filter specification of “TCP.flags.reset==1”.
If the 3-way handshake is successful, continue to the RPC Discovery phase.
The RPC Discovery phase will occur one of two ways. In both methods the client will know the identifier for the RPC Server it wants to contact and will supply that to the computer hosting the RPC Server and ask for information on how to contact the RPC Server. The identifier is different depending on which method is used and the RPC client will know ahead of time which method it wishes to use.
Discovery - RPC Over TCPIP
This method is a two-step process. First the RPC client will contact the End Point Mapper (EPM) on the machine hosting the RPC Server to find out what port and IP address that Server is listening on. Upon successful completion of this the RPC client will contact the RPC Server directly on the indicated IP address and Port. Below is a sample of what this would look like and a step by step explanation below it.
This step depends on the successful TCP session establishment twice, first to the EPM and then to the RPC Server.
- The RPC Client will open a TCP session with TCP port 135 on the computer hosting RPC Server of interest. This can be picked out using the following filter syntax in Netmon or Wireshark: “tcp.port==135”.
- The RPC Client will send an RPC Bind request using the UUID of the End Point Mapper and the RPC EPM should respond with a Bind ACK packet.
- The RPC Client will make a MAP request to the EPM to locate the IP address and port of the RPC Server of interest, identifying the RPC Server based on its UUID.
- The EPM will send back a MAP Response that indicates the IP and port the RPC Server is listening on.
- The RPC Client will then open a TCP session with the IP and port it received in the EPM MAP response.
- The client will send an RPC Bind Request to the RPC Server specifying the UUID of the RPC Server application and should get back a Bind ACK from the RPC Server.
- There will be an RPC Alter Context Request/Response in which authentication will take place. If an error is noted here then see the following section for help determining why the error is occurring - Authentication
- Perform some RPC operations…(Go to RPC Communication phase)
Discovery - RPC Over SMB
The second method an RPC Client may use to contact an RPC Server is RPC over SMB. This method depends upon first establishing an SMB session with the computer hosting the RPC Server and then using the Named Pipes protocol to communicate using RPC. So in effect there are several levels of encapsulation – RPC over Named Pipes over SMB over TCP. We will not address the SMB session setup in this document and the TCP session establishment has already been discussed.
With a successfully opened TCP and SMB session, next:
- The RPC Client will issue a SMB TreeConnectAndX for the tree name “IPC$”. This is a special hidden share for inter-process communication. It should get a positive response from the computer hosting the RPC Server.
- The RPC Client will then issue an SMB NTCreateAndX for the name of the PIPE of the RPC Server Application and should get back a positive response.
Some examples are:
EVENTLOG = The Event log service winreg = Remote Registry svcctl = Service Control Manager srvsvc = Server Service
Next there is a Bind handshake. This is to “bind” the RPC client to the RPC server. There are a total of four packets involved:
- The RPC Client bind request containing the UUID of the desired RPC Server.
- A Write AndX response from the RPC Server.
- A Read AndX request from the RPC Client.
- A Bind ACK response from the RPC Server.
- At this time a RPC request to the RPC server component is expected.
At this point RPC communication is occurring between the RPC Client and RPC Server.
Active Directory RPC Issues Symptoms
If you are experiencing replication problems and getting RPC server is unavailable errors as is reported in repadmin /showreps below, use Portqry or Network Monitor to determine if RPC traffic is being blocked is the first step when attempting to troubleshoot RPC Server is unavailable errors.
[Replications Check,DC2] A recent replication attempt failed: From DC1 to DC2 Naming Context: CN=Schema,CN=Configuration,DC=xl The replication generated an error (1722): The RPC server is unavailable. The failure occurred at 2003-10-30 11:59.47. The last success occurred at 2003-10-28 20:50.22. 26 failures have occurred since the last success. [DC1] DsBind() failed with error 1722, The RPC server is unavailable.. The source remains down. Please check the machine. Bermuda\DC1 via RPC objectGuid: 28c78c72-3c95-499a-bcda137a250f069f Last attempt @ 2003-10-30 11:58.15 failed, result 1722: The RPC server is unavailable.
If you are blocking all ICMP traffic between separate AD sites, you will receive the errors below in the output of DCDIAG when trying to replicate inter-site:
Testing server: contoso\DC1 Starting test: Replications * Replications Check [Replications Check,DC1] A recent replication attempt failed: From DC2 to DC1 Naming Context: CN=Schema,CN=Configuration,DC=litware,DC=com The replication generated an error (1722): The RPC server is unavailable. The failure occurred at 2003-08-24 23:00.51. The last success occurred at (never). 553 failures have occurred since the last success. [DC2] DsBind() failed with error 1722, The RPC server is unavailable.. The source remains down. Please check the machine. REPLICATION LATENCY WARNING DC1: A full synchronization is in progress from DC2 to DC1 Replication of new changes along this path will be delayed. [DC2] LDAP connection failed with error 58, The specified server cannot perform the requested operation.
To resolve this issue, remove the ICMP traffic restriction between domain controllers. When establishing an RPC session prior to AD replication, ICMP traffic is used. If the ICMP fails, so does the RPC session establishment, and hence AD replication also fails.
ISA 2004 (and newer solutions) can prevent ICMP traffic with the exception of computers specified in the Remote Management Computers computer set which can be configured in system policy.
The following error will appear when attempting to connect to the computer:
computer < \\servername.domain.local> cannot be managed. The network path was not found. RPC server is unavailable.
Or when viewing the properties of the remote computer you will receive the error:
"Win32: The RPC server is unavailable".
Computer management is one of the better tools for testing RPC connectivity. When RPC traffic is being blocked, connections to other computers using the computer management console will fail.
When attempting to promote an additional domain controller in an Active Directory domain while the RPC service is blocked or not running, the following error will appear:
The domain "domain.local" is not an Active Directory domain, or an Active Directory domain controller for the domain could not be contacted.
Check if the system is on the same network segment, if a proper DNS server has been configured to translate pre-existing domain name and to contact an existing domain controller of the domain to which we are attempting to add the new one.
Connections to computers via Remote Desktop may fail if RPC connectivity cannot be established: when attempting to logon on to the domain via Remote Desktop the following error will be produced in the form of a popup error message if RPC connectivity is the root of the problem:
"The system cannot log you on due to the following error: The RPC server is unavailable.”
You may also see the following errors on the Terminal server:
Error 1727: The remote procedure call failed and did not execute Error 1722: The RPC server is unavailable. Error 1723: The RPC server is too busy to complete this operation. Error 1721: Not enough resources are available to complete this operation.
Event ID 5719: Source: NetLogon Description: No Windows NT Domain Controller is available for domain domain_name. The following error occurred: There are currently no logon servers available to service the logon request.
Event ID: 1219: Source: Winlogon Details: Logon rejected for CONTOSO\
. Unable to obtain Terminal Server User Configuration. Error: The RPC server is unavailable.
These errors can be a result of the TCP/IP NetBIOS Helper service being disabled on the Terminal server or NetBIOS over TCP/IP being disabled on one of the NIC's used to access the Terminal server.
You should also verify that the Client for Microsoft networks is bound to the adapter used to access the Terminal server. You can tell if this is happening by looking at a Netdiag /v from the box for the following output:
Testing redirector and browser... Failed NetBT transports test. . . . . . . : Failed List of NetBt transports currently configured: [FATAL] No NetBt transports are configured. Redir and Browser test . . . . . . : Failed List of transports currently bound to the Redir NetBIOSSmb [FATAL] The redir isn't bound to any NetBt transports. List of transports currently bound to the browser [FATAL] The browser isn't bound to any NetBt transports.
You can use the Portqry tool to verify that the required ports are open. You should run the Portqry tool on a computer that is not receiving any RPC errors against a computer that is receiving RPC errors by using the -n switch. To this, follow these steps:
- Click "Start", click "Run", type "cmd" in the "Open" box, and then click OK.
portqry -n <problem_server> -e 135
The output will appear similar to the following examples:
Querying target system called:
Attempting to resolve name to IP address...
Name resolved to 169.254.1.1
TCP port 135 (epmap service): LISTENING
Using ephemeral source port
Querying Endpoint Mapper Database...
UUID: f5cc59b4-4264-101a-8c59-08002b2f8426 NtFrs Service ncacn_ip_tcp:18.104.22.168 UUID: e3514235-4b06-11d1-ab04-00c04fc2dcd2 MS NT Directory DRS Interface ncacn_ip_tcp:22.214.171.124 UUID: e3514235-4b06-11d1-ab04-00c04fc2dcd2 MS NT Directory DRS Interface ncacn_http:126.96.36.199 UUID: e3514235-4b06-11d1-ab04-00c04fc2dcd2 MS NT Directory DRS Interface ncacn_http:188.8.131.52
If port 135 is blocked, the following will appear:
TCP port 135 (epmap service): NOT LISTENING
However, for these RPC Endpoint Mapper errors it is likely that ports greater than 1024 are blocked, and not port 135.From the output, you know the DC is using port 1094 for FRS and 1025, 1029, and 6004 for Active Directory replication. You can use the Portqry tool again to check those ports. For example, you can test all the ports at the same time by using the Portqry tool with the -o switch. For example, type:
portqry -n <problem_server> -o 1094,1025,1029,6004
If the ports all respond as "LISTENING," it's likely that blocked ports are not causing this problem. If any ports respond as "NOT LISTENING," the ports are probably blocked.
The following link will let you download PortQryV2 tool.