Attachment 787954 Details for Bug 1114090

sshd_config server configuration

sshd_config (text/plain), 7.05 KB, created by James Carter on 2018-10-31 06:13:59 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: James Carter

Created: 2018-10-31 06:13:59 UTC

Size: 7.05 KB

patch

obsolete

># sshd server system-wide configuration file.  See sshd(8) and sshd_config(5)
># for more information.  While formerly the default for all unset
># parms was shown, this is getting excessive, and only highlights are now shown.
># Revision history:
>#   2017-07-23	openssh-7.2p2	jimc	For Tumbleweed; rearrange ciphers
>#   2016-09-20	openssh-6.6p1	jimc	More Logjam (CVE-2015-4000)
>#   2015-07-17	openssh-6.2p2	jimc	Mitigate Logjam (CVE-2015-4000)
>#   2011-03-27	openssh-5.8p1	jimc	For SuSE 11.4
>#   (numerous revisions not recorded)
>#   2002-05-20	openssh-3.2.2p	jimc	Set up for Xena
>
># With multiple parm sources the first one wins (put host-specific settings
># earlier).  Sources are {command line, /etc/ssh/ssh_config }
>
># Network Parameters
># Port			22 (DFLT)
># AddressFamily		any (dflt) (or inet or inet6)
># ListenAddress		all (DFLT)
># Protocol 1 is crackable and there's a widespread virus that attacks it.
>Protocol		2
># ClientAlive uses the encrypted channel; TCPKeepAlive uses the TCP feature and
># can be spoofed.  Connection dies after Interval*CountMax.
>ClientAliveInterval	120
>ClientAliveCountMax	4
>TCPKeepAlive		no
># Compression		delayed (DFLT) (whether allowed if client requests it)
># UseDNS		yes (DFLT) (check if host's A rec <-> PTR rec)
>
># Cryptography (lists are comma separated, in preference order).
># jimc's tests in 2017: with an AES crypto engine, aes256-gcm is equal in speed
># to aes128-gcm, and twice as fast (+more secure) than aes256-ctr.  aes256-gcm
># is 2.5x faster than chacha, but without the engine chacha is 2 to 2.5x faster
># than aes128-gcm.  VMs don't have access to the engine.  So prefer aes256-gcm
># but allow chacha.  MD5 and SHA1 are at end of life.  Where is Keccak?  
>
># Conspiracy theorists believe that NIST/NSA approved elliptic curves got
># approved because the NSA has a way to attack them; jimc discounts that 
># threat.  Aaron Toponce (2nd ref below) has actual crypto complaints about 
># ecdh-sha2-nistp{256,384,521}.  He also mentions that Github will only use
># SHA1 (in 2015) so I'm keeping it.  
># References for openssh-6.6p1 revision (these are all from 2015)
>#   https://jbeekman.nl/blog/2015/05/ssh-logjam/ (for KEX)
>#   https://pthree.org/2015/01/12/reasonable-ssh-security-for-openssh-6-0-or-later/
>#   https://bbs.archlinux.org/viewtopic.php?id=188613
>
># Ciphers for data transfer.  aes* == Rijndael.  Most preferred first.  
>Ciphers		aes256-gcm@openssh.com,aes128-gcm@openssh.com,chacha20-poly1305@openssh.com,aes256-ctr,aes192-ctr,aes128-ctr,aes256-cbc
># Default for openssh-7.2p2 = chacha20-poly1305@openssh.com, aes128-ctr,
># 	aes192-ctr, aes256-ctr, aes128-gcm@openssh.com,
># 	aes256-gcm@openssh.com, aes128-cbc, aes192-cbc, aes256-cbc, 3des-cbc
>
># Message Authentication Codes.  "etm" means "encrypt then MAC" which is safer.
># Android (VX ConnectBot) offers hmac-sha1-96, hmac-sha1, hmac-md5-96, hmac-md5
>MACs		hmac-sha2-512-etm@openssh.com,hmac-sha2-256-etm@openssh.com,hmac-sha2-512,hmac-sha2-256,hmac-sha1
>
># KEX (key exchange) algos: elliptic curve Diffie-Hellman is more efficient and
># is not vulnerable to Logjam.  Beekman recommends this preference order.  
># Using diffie-hellman-group-exchange-*, 
># the server picks a prime field DH group from /etc/ssh/moduli, which you 
># should rebuild yourself with 2048 bit minimum (see ssh-keygen).  The client
># does what the server tells it and does not use /etc/ssh/moduli.  
># diffie-hellman-group14-sha1 uses a fixed 2048 bit prime, for fallback.  
># diffie-hellman-group1-sha1 uses a fixed 1024 bit prime, assumed compromised.
># curve25519-sha256@libssh.org is in >=vers. 6.5 (with bug) and is reliable
># in >=6.7; Beekman prefers it if available.  
>KexAlgorithms	curve25519-sha256@libssh.org,diffie-hellman-group-exchange-sha256,diffie-hellman-group-exchange-sha1,diffie-hellman-group14-sha1
>
>#	Minimum Diffie-Hellman prime field size.  RFC4419 and SSH default is 
>#	1024, which is being exploited in the wild.  
>KexDHMin		2048
>
># User Authentication
># AuthenticationMethods		(allow any single enabled method, dflt)
># ChallengeResponseAuthentication	yes (DFLT) (for smart cards)
># HostbasedAuthentication	no (DFLT) (SSH host identity, rhosts user auth)
># HostbasedUsesNameFromPacketOnly no (dflt) (omit reverse DNS lookup)
># GSSAPI -> honor cred from other end.    Kerberos -> create cred from passwd
>GSSAPIAuthentication		yes
># Kerberos -> for 'password' auth, use KDC rather than local password file.
># We UsePam which takes care of this, so turn this off in sshd.
>KerberosAuthentication		no
># KerberosOrLocalPasswd		yes (DFLT) (if krb fails try any legal auth)
># PubkeyAuthentication		yes (DFLT) (proto 2 only)
># You should turn off PasswordAuthentication if you UsePam.  However,
># SSH Secure Shell for Windows v2.4.0 proposes auth method "password",
># which is not recognized as UsePam (KeyboardInteractive).  So leave this on.
>PasswordAuthentication		yes
>UsePAM				yes
>
># User Restrictions (applied in the following order)
># without-password means root can only authenticate by a passwordless 
># mechanism, which would be publickey or GSSAPI (Kerberos).  Default is "yes".
>PermitRootLogin		without-password
>
># Login Procedure
># UseLogin		no (DFLT)
># LoginGraceTime	120 (DFLT) (abort login if not finished in this time)
># StrictModes		yes (DFLT) (user's files not world read/write)
># GSSAPICleanupCredentials	yes (DFLT)
># KerberosTicketCleanup		yes (DFLT) (if local Kerberos login)
># PrintLastLog		yes (DFLT)
># Mathnet's login_bridge prints the MOTD.
>PrintMotd		no
>PermitUserEnvironment	yes
>AcceptEnv		"LOGNAME LANG LC_*"
>
># Port and X11 Forwarding (forbidding forwarding does not tighten security
># because users can write their own forwarders.)
># AllowTcpForwarding	yes (DFLT)
># AllowAgentForwarding	yes (dflt)
># PermitOpen		any (dflt) (permit forwarding to only certain host:ports)
>PermitTunnel		yes
># GatewayPorts		no (DFLT) (can other hosts connect to forwarded ports)
>X11Forwarding		yes
># X11UseLocalhost	yes (DFLT) (DISPLAY=localhost:10)
># X11DisplayOffset	10 (DFLT) (X-server port is 6000 + this value)
>
># File Paths
># Dflt checks authorized_keys2 also.  Suppress this.
>AuthorizedKeysFile	.ssh/authorized_keys
># AuthorizedKeysCommand	(internal, dflt)
>#   For PKI, default is to look for the loginID in the cert's principals list.
>#   See TrustedUserCAKeys for which certs to believe in. 
># AuthorizedPrincipalsFile	(none)(DFLT)
>#   sshd believes in certs signed by the CA(s) in this file.  
># TrustedUserCAKeys	(none)(DFLT)
># HostKey		/etc/ssh/ssh_host{_,_rsa_,_dsa_}key (DFLT)
># HostCertificate	(none)(DFLT) -- matches HostKey secret key(s).
># PidFile		/var/run/sshd.pid (DFLT)
># RevokedKeys		(none)(dflt) -- list of revoked public keys
># XAuthLocation		/usr/bin/xauth (DFLT)
>
># Miscellaneous 
>#   /usr/lib/ssh/sftp-server exists but we're using the internal server:
>Subsystem		sftp internal-sftp
># MaxStartups softens denial of service attacks.  Counts connections where
># authentication has not finished (see LoginGraceTime).  Referring to S:R:M, if
># there are over S stuck connections, new connections are refused with
># probability R, increasing linearly to 100% when there are M stuck
># connections.
>MaxStartups		10:30:60
># LogLevel		INFO (DFLT)
># SyslogFacility	AUTH (DFLT)

Actions: View

Attachments on bug 1114090: 787952 | 787953 | 787954