Web, FTP, and Proxy
Web Service
Web Service Three major techniques in WWW (World Wide Web) System HTML – HyperText Markup Language Mark-up the text and define presentation effect by HTML Tags. http://www.w3.org/ HTTP – Hyper-Text Transfer Protocol Communication method between client and server, both browsers and web servers have to follow this standard. HTTPS – secured version URL – Uniform Resource Locator Describe how to access an object shared on the Internet Format Protocol :// [ [ username [ :password ] @ ] hostname [ :port ] ] [ /directory ] [ /filename ]
Web Service – The Client-Server Architecture Web Server: Answer HTTP request Web Client: Request certain page using URL Client Browser Web Server 2. 以 HTTP 協定送出 Request 4. 以 HTTP 協定回覆 Response 1. 以 URL 描述索取的資源位置向 Server 發送要求 3. 從 URL 描述的位置將 HTML 文件取出並回覆給 Client 5. 接收到 HTML 後由 Browser 解析後根據 HTML 描述定義將資料呈現出來
Web Service – The HTTP Protocol (1) HTTP: Hypertext Transfer Protocol RFCs: (HTTP 1.1) http://www.faqs.org/rfcs/rfc2068.html http://www.faqs.org/rfcs/rfc2616.html (Updated Version) Useful Reference: http://jmarshall.com/easy/http/ A network protocol used to deliver virtually all files and other data on the World Wide Web. HTML files, image files, query results, or anything else. Client-Server Architecture A browser is an HTTP client because it sends requests to an HTTP server (Web server), which then sends responses back to the client.
Web Service – The HTTP Protocol (2) Clients: ※ Send Requests to Servers Action “path or URL” Protocal Actions: GET, POST, HEAD Ex. GET /index.php HTTP/1.1 Headers Header_Name: value Ex. From: someuser@jmarshall.com (blank line) Data … Servers: ※ Respond to the clinets Status: 200: OK 404: Not Found … Ex. HTTP/1.1 200 OK Headers Same as clients Ex. Content-Type: text/html (blank line) Data…
Web Service – The HTTP Protocol (3) Example: nabsd [/home/chwong] -chwong- telnet nabsd.cs.nctu.edu.tw 80 Trying 140.113.17.215... Connected to nabsd.cs.nctu.edu.tw. Escape character is '^]'. GET / HTTP/1.0 Host: nabsd.cs.nctu.edu.tw HTTP/1.0 200 OK Content-Type: text/html Accept-Ranges: bytes ETag: "1897433431" Last-Modified: Tue, 29 May 2007 06:25:04 GMT Content-Length: 94 Date: Tue, 29 May 2007 06:25:06 GMT Server: lighttpd/1.4.15 X-Cache: HIT from nabsd.cs.nctu.edu.tw Via: 1.0 nabsd.cs.nctu.edu.tw:80 (squid/2.6.STABLE13) Connection: close <html> <body> <a href="http://nabsd.cs.nctu.edu.tw/~chwong/docs/"> haha </a> </body> </html> Connection closed by foreign host. status Headers Data action
Web Service – The HTTP Protocol (4) Get vs. Post (client side) Get: Parameters in URL GET http://nabsd.cs.nctu.edu.tw/get.php?a=1&b=3 HTTP/1.1 No data content Corresponding in HTML files Link URL: http://nabsd.cs.nctu.edu.tw/get.php?a=1&b=3 Using Form: <form method=“GET” action=“get.php”> … </form> Post: Parameters in Data Content POST http://nabsd.cs.nctu.edu.tw/post.php HTTP/1.1 <form method=“POST” action=“post.php”> … </form>
Web Service – The HTTP Protocol (5) HTTP Headers: What HTTP Headers can do? [Ref] http://www.cs.tut.fi/~jkorpela/http.html Content information (type, date, size, encoding, …) Cache control Authentication URL Redirection Transmitting cookies Knowing where client come from Knowing what software client use …
Web Service – Static vs. Dynamic Pages Technologies of Dynamic Web Pages Client Script Language JavaScript, Jscript, VBScript Client Interactive Technology Java Applet, Flash, XMLHTTP,AJAX Server Side CGI Languages: Perl, ASP, JSP, PHP, C/C++, …etc. Static vs. Dynamic
Web Service – Virtual Hosting (1) Providing services for more than one domain-name (or IP) in one web server. IP-Based Virtual Hosting vs. Name-Based Virtual Hosting IP-Base – Several IPs (or ports) Name-Base – Singe IP, several hostnames Example (Apache configuration) NameVirtualHost 140.113.17.215 <VirtualHost 140.113.17.215> ServerName nabsd.cs.nctu.edu.tw DocumentRoot "/www/na" </VirtualHost> ServerName sabsd.cs.nctu.edu.tw DocumentRoot "/www/sa" <VirtualHost 140.113.17.215:80> DocumentRoot /www/nabsd ServerName nabsd.cs.nctu.edu.tw </VirtualHost> <VirtualHost 140.113.17.221:80> DocumentRoot /www/tphp ServerName tphp.cs.nctu.edu.tw
Web Service – Virtual Hosting (2) Q: How Name-Based Virtual Hosting works? A: It takes use of HTTP Headers. % telnet cswproxy.cs.nctu.edu.tw 80 Trying 140.113.235.111... Connected to cswproxy.cs.nctu.edu.tw. Escape character is '^]'. GET / HTTP/1.0 Host: www.cs.nctu.edu.tw HTTP/1.0 200 OK Date: Tue, 05 Jun 2007 13:50:34 GMT ………… <html> <head> <title>NCTU -- CS</title> <META HTTP-EQUIV="Pragma" CONTENT="no-cache"> <meta http-equiv="refresh" content="0; URL=chinese/doc/index.html"> </head> </html> Connection closed by foreign host. % telnet cswproxy.cs.nctu.edu.tw 80 Trying 140.113.235.111... Connected to cswproxy.cs.nctu.edu.tw. Escape character is '^]'. GET / HTTP/1.0 Host: www.csie.nctu.edu.tw HTTP/1.0 200 OK Date: Tue, 05 Jun 2007 13:51:01 GMT ………… <html> <head> <title>NCTU -- CSIE</title> <meta http-equiv="refresh" content="0; URL=http://www.cs.nctu.edu.tw/"> Connection closed by foreign host.
File Transfer Protocol FTP File Transfer Protocol
FTP FTP File Transfer Protocol Used to transfer data from one computer to another over the internet. Client-Server Architecture. Separated control/data connections. Modes: Active Mode, Passive Mode RFCs: RFC 959 – File Transfer Protocol RFC 2228 – FTP Security Extensions RFC 2640 – UTF-8 support for file name
FTP – Flow (1) Client Server Binding on port 21 Connect to server port 21 using port A. USER #### PASS ******** PORT h1,h2,h3,h4,p1,p2 Send some requests get return data from p1*256+p2 Quit Server Binding on port 21 Accepts connection from client, output welcome messages. 331 User name okay, need password. 230 User logged in, proceed. 200 PORT Command successful. Binding source port 20, connect to client port p1*256+p2, send data. …
FTP – Flow (2) Example Control Connection % telnet chonsilab.dyndns.org 21 Trying 140.113.215.86... Connected to chonsilab.dyndns.org. Escape character is '^]'. 220 Serv-U FTP-Server v2.5k for WinSock ready... USER test 331 User name okay, need password. PASS test 230 User logged in, proceed. PORT 140,113,17,215,39,19 200 PORT Command successful. LIST 150 Opening ASCII mode data connection for /bin/ls. 226 Transfer complete. quit 221 Goodbye! Connection closed by foreign host.
FTP – Flow (3) Example (contd.) Retrieving Data % cat server.pl #!/usr/bin/perl -w package MyPackage; use strict; use base qw(Net::Server::PreFork); MyPackage->run(port => $ARGV[0]); sub process_request { while (<STDIN>) { s/\r?\n$//; print STDERR "$_\n"; } Example (contd.) Retrieving Data Client must bind the random port %perl server.pl 10003 2007/06/06-13:16:08 MyPackage (type Net::Server::PreFork) starting! pid(4346) Binding to TCP port 10003 on host * Group Not Defined. Defaulting to EGID '1000 110 100 80 0 1000 1000' User Not Defined. Defaulting to EUID '1001' -rwxrwxrwx 1 user group 0 Sep 11 2005 AUTOEXEC.BAT -rwxrwxrwx 1 user group 209 Sep 11 2005 boot.ini -rwxrwxrwx 1 user group 213830 Mar 25 2005 bootfont.bin -rwxrwxrwx 1 user group 0 Sep 11 2005 CONFIG.SYS drwxrwxrwx 1 user group 0 Apr 8 17:30 Documents and Settings -rwxrwxrwx 1 user group 0 Sep 11 2005 IO.SYS -rwxrwxrwx 1 user group 0 Sep 11 2005 MSDOS.SYS -rwxrwxrwx 1 user group 47772 Mar 25 2005 NTDETECT.COM -rwxrwxrwx 1 user group 304752 Mar 25 2005 ntldr drwxrwxrwx 1 user group 0 May 21 23:30 Program Files drwxrwxrwx 1 user group 0 Aug 19 2006 RECYCLER drwxrwxrwx 1 user group 0 Feb 16 2006 System Volume Information drwxrwxrwx 1 user group 0 May 28 16:45 WINDOWS
FTP – commands, responses USER username PASS password LIST Return list of file in current dir. RETR filename Retrieves (gets) file. STOR filename Stores (puts) file onto server. PORT h1,h2,h3,h4,p1,p2 Set to active mode PASV Set to passive mode DELE Remove file on the server. QUIT Return Codes First code 1: Positive Preliminary reply 2: Positive Completion reply 3: Positive Intermediate reply 4: Transient Negative Completion reply 5: Permanent Negative Completion reply Second code 0: The failure was due to a syntax error 1: A reply to a request for information. 2: A reply relating to connection information 3: A reply relating to accounting and authorization. 5: The status of the Server file system
FTP – Active Mode vs. Passive Mode (1) FTP client bind a random port (>1023) and sends the random port to FTP server using “PORT” command. When the FTP server initiates the data connection to the FTP client, it binds the source port 20 and connect to the FTP client the random port sent by client. PORT h1,h2,h3,h4,p1,p2 Passive Mode FTP client sends “PASV” command to the server, make the server bind a random port (>1023) and reply the random port back. When initializing the data connection, the FTP client connect to the FTP Server the random port, get data from that port. PASV Server reply: 227 Entering Passive Mode (h1,h2,h3,h4,p1,p2) ※ IP:port (6bytes) h1,h2,h3,h4,p1,p2 Ex. 140.113.17.215:45678 140,113,17,215,178,110
FTP – Active Mode vs. Passive Mode (2)
FTP – When FTP meets NAT/Firewall (1) Firewall behavior Generally, the NAT/Firewall permits all outgoing connection from internal network, and denies all incoming connection from external network. Problem when FTP meets NAT/Firewall Due to the separated command/data connection, the data connections are easily blocked by the NAT/Firewall. Problem Cases: Active mode, NAT/Firewall on client side. Passive mode can solve this problem. Passive mode, NAT/Firewall on server side. Active mode can solve this problem. Both client side and server side have NAT/Firewall The real problem.
FTP – When FTP meets NAT/Firewall (2) Active mode, NAT/Firewall on client side. Passive mode can solve this problem. Client Server NAT/Firewall PORT IP, port Y Connect to port Y BLOCKED Active Mode Client Server NAT/Firewall PASV reply IP, port Z Connect to port Z PASS Passive Mode
FTP – When FTP meets NAT/Firewall (3) Passive mode, NAT/Firewall on Server side. Active mode can solve this problem. Client Server NAT/Firewall PASV reply IP, port Z Connect to port Z BLOCKED Passive Mode Client Server NAT/Firewall PORT IP, port Y Connect to port Y PASS Active Mode
FTP – When FTP meets NAT/Firewall (4) Real Problem: Firewall on both sides. Solution: ftp-proxy running on NAT/Firewall Client Server NAT/Firewall PORT IP, port Y Connect to port Y BLOCKED Active Mode Client Server NAT/Firewall PASV reply IP, port Z Connect to port Z BLOCKED Passive Mode
FTP – Security Security concern Solutions As we seen, FTP connections (both command and data) are transmitted in clear text. What if somebody sniffing the network? We need encryption. Solutions FTP over SSH So called secure-FTP. Both commands and data are encrypted while transmitting. Poor performance. FTP over TLS Only commands are encrypted while transmitting. Better performance.
FTP – Pure-FTPd (1) Introduction A small, easy to set up, fast and secure FTP server Support chroot Restrictions on clients, and system-wide. Verbose logging with syslog Anonymous FTP with more restrictions Virtual Users, and Unix authentication FXP (File eXchange Protocol) FTP over TLS UTF-8 support for file names
FTP – Pure-FTPd (2) Installation Ports: /usr/ports/ftp/pure-ftpd Options
FTP – Pure-FTPd (3) Startup: Other options WITH_CERTFILE for TLS Default: /etc/ssl/private/pure-ftpd.pem WITH_LANG Change the language of output messages Startup: Add pureftpd_enable=“YES” into /etc/rc.conf
FTP – Pure-FTPd Configurations(1) File: /usr/local/etc/pure-ftpd.conf Documents Configuration sample: /usr/local/etc/pure-ftpd.conf.sample All options are explained clearly in this file. Other documents See /usr/local/share/doc/pure-ftpd nabsd [/usr/local/share/doc/pure-ftpd] -chwong- ls AUTHORS README README.MySQL THANKS CONTACT README.Authentication-Modules README.Netfilter pure-ftpd.png COPYING README.Configuration-File README.PGSQL pureftpd.schema HISTORY README.Contrib README.TLS NEWS README.LDAP README.Virtual-Users
FTP – Pure-FTPd Configurations(2) # Cage in every user in his home directory ChrootEveryone yes # If the previous option is set to "no", members of the following group # won't be caged. Others will be. If you don't want chroot()ing anyone, # just comment out ChrootEveryone and TrustedGID. TrustedGID 0 # PureDB user database (see README.Virtual-Users) PureDB /etc/pureftpd.pdb # If you want simple Unix (/etc/passwd) authentication, uncomment this UnixAuthentication yes # Port range for passive connections replies. - for firewalling. PassivePortRange 30000 50000 # This option can accept three values : # 0 : disable SSL/TLS encryption layer (default). # 1 : accept both traditional and encrypted sessions. # 2 : refuse connections that don't use SSL/TLS security mechanisms, # including anonymous sessions. # Do _not_ uncomment this blindly. Be sure that : # 1) Your server has been compiled with SSL/TLS support (--with-tls), # 2) A valid certificate is in place, # 3) Only compatible clients will log in. TLS 2 # UTF-8 support for file names (RFC 2640) # Define charset of the server filesystem and optionnally the default charset # for remote clients if they don't use UTF-8. # Works only if pure-ftpd has been compiled with --with-rfc2640 FileSystemCharset big5 # ClientCharset big5
FTP – Pure-FTPd Problem Shooting Logs Location In default, syslogd keeps ftp logs in /var/log/xferlog Most frequent problem pure-ftpd: (?@?) [ERROR] Unable to find the 'ftp' account It’s ok, but you may need it for Virtual FTP Account. pure-ftpd: (?@?) [ERROR] Sorry, but that file doesn't exist: [/etc/ssl/private/pure-ftpd.pem] If you set TLS = 2, then this file is needed. How to generate a pure-ftpd.pem? See README.TLS
FTP – Pure-FTPd Tools pure-* pure-ftpwho pure-pw List information of users who use the FTP server now. pure-pw To create Virtual Users using PureDB man pure-pw See README.Virtual-Users
FTP – PF: Issues with FTP (1) Reference: http://www.openbsd.org/faq/pf/ftp.html FTP Client Behind the Firewall Problem Clients cannot use active mode Use ftp-proxy Use inetd to start ftp-proxy man ftp-proxy In pf.conf nat-anchor “ftp-proxy/*” rdr-anchor “ftp-proxy/*” rdr on $int_if proto tcp from any to any port 21 -> 127.0.0.1 \ port 8021 anchor “ftp-proxy/*”
FTP – PF: Issues with FTP (2) PF “Self-Protecting” an FTP Server Problem Clients cannot use passive mode Open holes so that clients can connect into the data channel In pf.conf pass in on $ext_if proto tcp from any to any port 21 keep state pass in on $ext_if proto tcp from any to any port > 49151 keep state
FTP – PF: Issues with FTP (3) FTP Server Protected by an External PF Firewall Running NAT Problem Clients cannot use passive mode Use ftp-proxy Need some flags of ftp-proxy man ftp-proxy In pf.conf nat-anchor “ftp-proxy/*” nat on $ext_if inet from $int_if -> ($ext_if) rdr-anchor “ftp-proxy/*” pass in on $ext_if inet proto tcp to $ext_ip port 21 flags S/SA keep state pass out on $int_if inet proto tcp to $ftp_ip port 21 user proxy flags S/SA keep state anchor “ftp-proxy/*”
FTP – More Tools /usr/ports/ftp/pftpx /usr/ports/ftp/lftp FileZilla Another ftp proxy daemon /usr/ports/ftp/lftp A powerful functional client Support TLS FileZilla An FTP Client for Windows
Proxy
Reply (using cached result) Proxy Proxy A proxy server is a server which services the requests of its clients by: Making requests to other servers Caching some results for further same requests Goals: Performance Stability Central Control …etc. Roles: Forward Proxy Reverse Proxy Targets Web/FTP Pages TCP/IP Connections Request Reply Reply (using cached result) client Proxy Server Original Server
Proxy – The Forward Proxy Proxy the outgoing requests, for the reason of Bandwidth saving Performance Central control When objects requested are In cache, return the cached objects Otherwise, proxy server requests object from origin server, then cache it and return to client Request Reply Reply (using cached result) client Proxy Server Original Server
Proxy – The Reverse Proxy Proxy the incoming requests, for the reason of Reducing Server Load (by caching) Load Balance Fault Tolerant Reverse proxy acts as the original server, accept incoming requests, reply corresponding result. SEAMLESS for clients! Request Reply client Reverse Proxy Server Server1 Internet
Proxy – SQUID A web proxy server & cache daemon. Supports HTTP, FTP Limited support for TLS, SSL, Gopher, HTTPS Latest stable version: 2.6-STABLE13, 2007/5/11 Port install: /usr/ports/www/squid Startup: /etc/rc.conf squid_enable="YES" squid_config="/usr/local/etc/squid/squid.conf" squid_user="squid" /usr/local/etc/rc.d/squid start Configuration Sample/Documents: /usr/local/etc/squid/squid.conf.default
Proxy – SQUID Configuration (1) Listen Port Service Port http_port 3128 Neighbored Communication icp_port 3130 Logs access_log access_log /var/log/squid/access.log squid cache_log cache_log /var/log/squid/cache.log cache_store_log cache_store_log /var/log/squid/store.log
Proxy – SQUID Configuration (2) Access Control acl – define an access control list Format: acl acl-name acl-type data acl all src 0.0.0.0/0.0.0.0 acl NCTU srcdomain .nctu.edu.tw acl YAHOO dstdomain .yahoo.com acl allowhost src “/usr/local/etc/squid.squid.allow” http_access – define the control rule Format: http_access allow|deny acl-name http_access allow NCTU http_access allow allowhost http_access deny all
Proxy – SQUID Configuration (3) Proxy Relationship Protocol: ICP (Internet Cache Protocol) RFC 2186 2187, using UDP Related Configuration cache_peer hostname type http_port icp_port [options] cache_peer_domain cache-host domain [domain …] cache_peer_access cache-host allow|deny acl-name
Proxy – SQUID Configuration (4) Cache Control cache_mem 256 MB cache_dir ufs /usr/local/squid/cache 100 16 256 cache_swap_low 93 cache_swap_high 98 maximum_object_size 4096 KB maximum_object_size_in_memory 8 KB
Proxy – SQUID Configuration (5) Sample: Proxy Configuration http_port 3128 icp_port 3130 cache_mem 32 MB cache_dir ufs /usr/local/squid/cache 100 16 256 access_log /var/log/squid/access.log squid cache_log /var/log/squid/cache.log cache_store_log /var/log/squid/store.log pid_filename /usr/local/squid/logs/squid.pid visible_hostname nabsd.cs.nctu.edu.tw acl allowhosts src "/usr/local/etc/squid/squid.allow“ http_access allow allowhosts http_access deny all
Proxy – SQUID Configuration (6) Sample: Reverse Proxy Configuration http_port 80 vhost icp_port 3130 cache_mem 32 MB cache_dir ufs /usr/local/squid/cache 100 16 256 access_log /var/log/squid/access.log squid cache_log /var/log/squid/cache.log cache_store_log /var/log/squid/store.log pid_filename /usr/local/squid/logs/squid.pid visible_hostname nabsd.cs.nctu.edu.tw url_rewrite_program /usr/local/squid/bin/redirect.pl acl cswww dstdomain csws1 csws2 http_access allow all cswww always_direct allow cswww
Proxy – SQUID Configuration (7) % cat /usr/local/squid/bin/redirect.pl #!/usr/bin/perl $|=1; # use non-blocking I/O while(<STDIN>){ if (/^http:\/\/www\.cs\.nctu\.edu\.tw\/([^\s]*)/) { my $ran = int(rand(2)+1); print "http://csws$ran.cs.nctu.edu.tw/$1\n"; next; } print "\n";