release notes | UserGuide: 4.2, 5.0, 5.1, 5.2, 6.0, 6.1, 6.2, 7.0, 7.1, 7.2 (unreleased), 8.0, 8.1, 8.2, 9.0, 9.1, 9.2, 10.0, 10.1, 10.2, 7.2 (unreleased) | Book: 4.2, 5.0, 5.1, 5.2, 6.0, 6.1, 6.2, 7.0, 7.1, 7.2 (unreleased), 8.0, 8.1, 8.2, 9.0, 9.1, 9.2, 10.0, 10.1, 10.2, 7.2 (unreleased) | Wiki | Q&A black_bg
Edit me on GitHub
Back to the contents page.

Chapter 25. Complex Network Configuration

Table of Contents

This chapter contains solutions for several non-trivial network configurations. The first section discusses the interoperation of dCache with firewalls and does not require any background knowledge about dCache other than what is given in the installation guide (Chapter 2, Installing dCache) and the first steps tutorial (Chapter 3, Getting to know your dCache. The following sections will deal with more complex network topologies, e.g. private subnets. Even though not every case is covered, these cases might help solve other problems, as well. Intermediate knowledge about dCache is required.

Warning

The TCP and UDP ports used for dCache internal communication (port 11111 by default) MUST be subject to firewall control so that only other dCache nodes can access them. Failure to do this will allow an attacker to issue arbitrary commands on any node within your dCache cluster, as whichever user the dCache process runs.

Firewall Configuration

The components of a dCache instance may be distributed over several hosts (nodes). Some of these components are accessed from outside and consequently the firewall needs to be aware of that. We contemplate two communication types, the dCache internal communication and the interaction from dCache with clients.

Since dCache is very flexible, most port numbers may be changed in the configuration. The command dcache ports will provide you with a list of services and the ports they are using.

Basic Installation

This section assumes that all nodes are behind a firewall and have full access to each other.

dCache internal.

  • As we assume that all nodes are behind a firewall and have full access to each other there is nothing to be mentioned here.

  • On the pool nodes the LAN range ports need to be opened to allow pool to pool communication. By default these are ports 33115-33145 (set by the properties dcache.net.lan.port.min and dcache.net.lan.port.max).

dCache communication with client.

  • The door ports need to be opened to allow the clients to connect to the doors.

  • The WAN/LAN range ports need to be opened to allow the clients to connect to the pools. The default values for the WAN port range are 20000-25000. The WAN port range is defined by the properties dcache.net.wan.port.min and dcache.net.wan.port.max.

Multi-Node with Firewalls

Multinode setup with firewalls on the nodes.

dCache internal.

  • The LocationManager server runs in the dCacheDomain. By default it is listening on UDP port 11111. Hence, on the head node port 11111 needs to be opened in the firewall to allow connections to the LocationManager. Remember to limit this so that only other dCache nodes are allowed access.

  • On the pool nodes the LAN range ports need to be opened to allow pool to pool communication. By default these are ports 33115-33145 (set by the properties dcache.net.lan.port.min and dcache.net.lan.port.max).

dCache communication with client.

  • The door ports need to be opened to allow the clients to connect to the doors.

  • The WAN/LAN range ports need to be opened to allow the clients to connect to the pools. The default values for the WAN port range are 20000-25000. The WAN port range is defined by the properties dcache.net.wan.port.min and dcache.net.wan.port.max.

More complex setups are described in the following sections.

GRIDFTP Connections via two or more Network Interfaces

Description

The host on which the GridFTP door is running has several network interfaces and is supposed to accept client connections via all those interfaces. The interfaces might even belong to separate networks with no routing from one network to the other.

As long as the data connection is opened by the GridFTP server (passive FTP mode), there is no problem with having more than one interface. However, when the client opens the data connection (active FTP mode), the door (FTP server) has to supply it with the correct interface it should connect to. If this is the wrong interface, the client might not be able to connect to it, because there is no route or the connection might be inefficient.

Also, since a GridFTP server has to authenticate with an X.509 grid certificate and key, there needs to be a separate certificate and key pair for each name of the host or a certificate with alternative names. Since each network interface might have a different name, several certificates and keys are needed and the correct one has to be used, when authenticating via each of the interfaces.

Solution

Define two domains, one for the internal and one for the external use. Start a separate SRM and GRIDFTP service in these domains.

The srm and the gridftp service have to be configured with the property listen, only to listen on the interface they should serve. The locations of the grid host certificate and key files for the interface have to be specified explicitly with the properties dcache.authn.hostcert.cert and dcache.authn.hostcert.key.

In this example we show a setup for two GridFTP doors serving two network interfaces with the hostnames door-internal (111.111.111.5) and door-external (222.222.222.5) which are served by two GRIDFTP doors in two domains.

[internalDomain]
listen=111.111.111.5
dcache.authn.hostcert.cert=PATH-ODE-ED/interface-cert-internal.pem
dcache.authn.hostcert.key=PATH-ODE-ED/interface-key-internal.pem
[internalDomain/srm]
srm.cell.name=srm-internal
srm.protocols.loginbroker=loginbroker-internal
srm.net.host=door-internal
[internalDomain/ftp]
ftp.authn.protocol = gsi
ftp.cell.name=GFTP-door-internal
dcache.service.loginbroker=loginbroker-internal

[externalDomain]
listen=222.222.222.5
dcache.authn.hostcert.cert=PATH-ODE-ED/interface-cert-external.pem
dcache.authn.hostcert.key=PATH-ODE-ED/interface-key-external.pem
[externalDomain/srm]
srm.cell.name=srm-external
srm.protocols.loginbroker=loginbroker-external
srm.net.host=door-external
[externalDomain/ftp]
ftp.authn.protocol = gsi
ftp.cell.name=GFTP-door-external
dcache.service.loginbroker=loginbroker-external

GRIDFTP with Pools in a Private Subnet

Description

If pool nodes of a dCache instance are connected to a secondary interface of the GridFTP door, e.g. because they are in a private subnet, the GridFTP door will still tell the pool to connect to its primary interface, which might be unreachable.

The reason for this is that the control communication between the door and the pool is done via the network of TCP connections which have been established at start-up. In the standard setup this communication is routed via the dCache domain. However, for the data transfer, the pool connects to the GridFTP door. The IP address it connects to is sent by the GridFTP door to the pool via the control connection. Since the GridFTP door cannot find out which of its interfaces the pool should use, it normally sends the IP address of the primary interface.

Solution

Tell the GridFTP door explicitly which IP it should send to the pool for the data connection with the ftp.net.internal property.

Example:

E.g. if the pools should connect to the secondary interface of the GridFTP door host which has the IP address 10.0.1.1, set

ftp.net.internal=10.0.1.1

in the /etc/dcache/dcache.conf file.

Using IPv6 with dCache

dCache does not require any special configuration to use IPv6 addressing. As long as all machines in an instance can be resolved cleanly through DNS, communication will work as expected. However, there are a few caveats that implementors should consider:

Use dual addressed interfaces for FTP

For machines reachable through both IPv4 and IPv6, FTP and GridFTP doors must be configured so that both address families use the same physical interface.

Adding addresses

When a new IPv6 address is added for a door, a restart of its domain is not required. Doors discover the available IP addresses as part of it publishing its details to SRM, frontend and info-provider. This is done periodically (every 5 seconds by default).

Links, Units and Groups

In an instance that was previously only used with IPv4 addressing, the default ‘world-net’ ugroup is likely to only match IPv4 connections. Consider the following situation:

(local) admin > \sp psu ls -a unit
*/*  (type=Protocol;canonical=*/*;uGroups=1)
 uGroupList :
   any-protocol  (links=0;units=1)
*@*  (type=Store;canonical=*@*;uGroups=1) (required=1; onlyOneCopyPer=[])
 uGroupList :
   any-store  (links=0;units=1)
0.0.0.0/0.0.0.0  (type=Net;canonical=0.0.0.0/0;uGroups=1)
 uGroupList :
   world-net  (links=14;units=1)

In order to properly match IPv6 connections, a IPv6 NET unit must be added to any ugroups that need to handle IPv6 traffic, in this example, only the world-net group. This can be done with two admin commands, targeting PoolManager:

psu create unit -net ::/0
psu addto ugroup world-net ::/0

This will update your system live, but these changes will be lost when you next restart the domain hosting poolmanager. Therefore you must make these changes persistent.

The file /var/lib/dcache/config/poolmanager.conf contains the poolmanager configuration.

If you maintain the contents of this file outside of dCache then you need to add the above two commands in the file. Take care to add them in the correct order (the `psu addto command must be after the ‘world-net’ ugroup and the ‘::/0’ unit are created).

If you don’t maintain the contents of this file yourself, then simply run the save admin command in poolmanager. This will update the file with the new content.