Documentation/networking/ipvs-sysctl.txt

   1 /proc/sys/net/ipv4/vs/* Variables:
   2
   3 am_droprate - INTEGER
   4         default 10
   5
   6         It sets the always mode drop rate, which is used in the mode 3
   7         of the drop_rate defense.
   8
   9 amemthresh - INTEGER
  10         default 1024
  11
  12         It sets the available memory threshold (in pages), which is
  13         used in the automatic modes of defense. When there is no
  14         enough available memory, the respective strategy will be
  15         enabled and the variable is automatically set to 2, otherwise
  16         the strategy is disabled and the variable is  set  to 1.
  17
  18 backup_only - BOOLEAN
  19         0 - disabled (default)
  20         not 0 - enabled
  21
  22         If set, disable the director function while the server is
  23         in backup mode to avoid packet loops for DR/TUN methods.
  24
  25 conn_reuse_mode - INTEGER
  26         1 - default
  27
  28         Controls how ipvs will deal with connections that are detected
  29         port reuse. It is a bitmap, with the values being:
  30
  31         0: disable any special handling on port reuse. The new
  32         connection will be delivered to the same real server that was
  33         servicing the previous connection. This will effectively
  34         disable expire_nodest_conn.
  35
  36         bit 1: enable rescheduling of new connections when it is safe.
  37         That is, whenever expire_nodest_conn and for TCP sockets, when
  38         the connection is in TIME_WAIT state (which is only possible if
  39         you use NAT mode).
  40
  41         bit 2: it is bit 1 plus, for TCP connections, when connections
  42         are in FIN_WAIT state, as this is the last state seen by load
  43         balancer in Direct Routing mode. This bit helps on adding new
  44         real servers to a very busy cluster.
  45
  46 conntrack - BOOLEAN
  47         0 - disabled (default)
  48         not 0 - enabled
  49
  50         If set, maintain connection tracking entries for
  51         connections handled by IPVS.
  52
  53         This should be enabled if connections handled by IPVS are to be
  54         also handled by stateful firewall rules. That is, iptables rules
  55         that make use of connection tracking.  It is a performance
  56         optimisation to disable this setting otherwise.
  57
  58         Connections handled by the IPVS FTP application module
  59         will have connection tracking entries regardless of this setting.
  60
  61         Only available when IPVS is compiled with CONFIG_IP_VS_NFCT enabled.
  62
  63 cache_bypass - BOOLEAN
  64         0 - disabled (default)
  65         not 0 - enabled
  66
  67         If it is enabled, forward packets to the original destination
  68         directly when no cache server is available and destination
  69         address is not local (iph->daddr is RTN_UNICAST). It is mostly
  70         used in transparent web cache cluster.
  71
  72 debug_level - INTEGER
  73         0          - transmission error messages (default)
  74         1          - non-fatal error messages
  75         2          - configuration
  76         3          - destination trash
  77         4          - drop entry
  78         5          - service lookup
  79         6          - scheduling
  80         7          - connection new/expire, lookup and synchronization
  81         8          - state transition
  82         9          - binding destination, template checks and applications
  83         10         - IPVS packet transmission
  84         11         - IPVS packet handling (ip_vs_in/ip_vs_out)
  85         12 or more - packet traversal
  86
  87         Only available when IPVS is compiled with CONFIG_IP_VS_DEBUG enabled.
  88
  89         Higher debugging levels include the messages for lower debugging
  90         levels, so setting debug level 2, includes level 0, 1 and 2
  91         messages. Thus, logging becomes more and more verbose the higher
  92         the level.
  93
  94 drop_entry - INTEGER
  95         0  - disabled (default)
  96
  97         The drop_entry defense is to randomly drop entries in the
  98         connection hash table, just in order to collect back some
  99         memory for new connections. In the current code, the
 100         drop_entry procedure can be activated every second, then it
 101         randomly scans 1/32 of the whole and drops entries that are in
 102         the SYN-RECV/SYNACK state, which should be effective against
 103         syn-flooding attack.
 104
 105         The valid values of drop_entry are from 0 to 3, where 0 means
 106         that this strategy is always disabled, 1 and 2 mean automatic
 107         modes (when there is no enough available memory, the strategy
 108         is enabled and the variable is automatically set to 2,
 109         otherwise the strategy is disabled and the variable is set to
 110         1), and 3 means that that the strategy is always enabled.
 111
 112 drop_packet - INTEGER
 113         0  - disabled (default)
 114
 115         The drop_packet defense is designed to drop 1/rate packets
 116         before forwarding them to real servers. If the rate is 1, then
 117         drop all the incoming packets.
 118
 119         The value definition is the same as that of the drop_entry. In
 120         the automatic mode, the rate is determined by the follow
 121         formula: rate = amemthresh / (amemthresh - available_memory)
 122         when available memory is less than the available memory
 123         threshold. When the mode 3 is set, the always mode drop rate
 124         is controlled by the /proc/sys/net/ipv4/vs/am_droprate.
 125
 126 expire_nodest_conn - BOOLEAN
 127         0 - disabled (default)
 128         not 0 - enabled
 129
 130         The default value is 0, the load balancer will silently drop
 131         packets when its destination server is not available. It may
 132         be useful, when user-space monitoring program deletes the
 133         destination server (because of server overload or wrong
 134         detection) and add back the server later, and the connections
 135         to the server can continue.
 136
 137         If this feature is enabled, the load balancer will expire the
 138         connection immediately when a packet arrives and its
 139         destination server is not available, then the client program
 140         will be notified that the connection is closed. This is
 141         equivalent to the feature some people requires to flush
 142         connections when its destination is not available.
 143
 144 expire_quiescent_template - BOOLEAN
 145         0 - disabled (default)
 146         not 0 - enabled
 147
 148         When set to a non-zero value, the load balancer will expire
 149         persistent templates when the destination server is quiescent.
 150         This may be useful, when a user makes a destination server
 151         quiescent by setting its weight to 0 and it is desired that
 152         subsequent otherwise persistent connections are sent to a
 153         different destination server.  By default new persistent
 154         connections are allowed to quiescent destination servers.
 155
 156         If this feature is enabled, the load balancer will expire the
 157         persistence template if it is to be used to schedule a new
 158         connection and the destination server is quiescent.
 159
 160 ignore_tunneled - BOOLEAN
 161         0 - disabled (default)
 162         not 0 - enabled
 163
 164         If set, ipvs will set the ipvs_property on all packets which are of
 165         unrecognized protocols.  This prevents us from routing tunneled
 166         protocols like ipip, which is useful to prevent rescheduling
 167         packets that have been tunneled to the ipvs host (i.e. to prevent
 168         ipvs routing loops when ipvs is also acting as a real server).
 169
 170 nat_icmp_send - BOOLEAN
 171         0 - disabled (default)
 172         not 0 - enabled
 173
 174         It controls sending icmp error messages (ICMP_DEST_UNREACH)
 175         for VS/NAT when the load balancer receives packets from real
 176         servers but the connection entries don't exist.
 177
 178 pmtu_disc - BOOLEAN
 179         0 - disabled
 180         not 0 - enabled (default)
 181
 182         By default, reject with FRAG_NEEDED all DF packets that exceed
 183         the PMTU, irrespective of the forwarding method. For TUN method
 184         the flag can be disabled to fragment such packets.
 185
 186 secure_tcp - INTEGER
 187         0  - disabled (default)
 188
 189         The secure_tcp defense is to use a more complicated TCP state
 190         transition table. For VS/NAT, it also delays entering the
 191         TCP ESTABLISHED state until the three way handshake is completed.
 192
 193         The value definition is the same as that of drop_entry and
 194         drop_packet.
 195
 196 sync_threshold - vector of 2 INTEGERs: sync_threshold, sync_period
 197         default 3 50
 198
 199         It sets synchronization threshold, which is the minimum number
 200         of incoming packets that a connection needs to receive before
 201         the connection will be synchronized. A connection will be
 202         synchronized, every time the number of its incoming packets
 203         modulus sync_period equals the threshold. The range of the
 204         threshold is from 0 to sync_period.
 205
 206         When sync_period and sync_refresh_period are 0, send sync only
 207         for state changes or only once when pkts matches sync_threshold
 208
 209 sync_refresh_period - UNSIGNED INTEGER
 210         default 0
 211
 212         In seconds, difference in reported connection timer that triggers
 213         new sync message. It can be used to avoid sync messages for the
 214         specified period (or half of the connection timeout if it is lower)
 215         if connection state is not changed since last sync.
 216
 217         This is useful for normal connections with high traffic to reduce
 218         sync rate. Additionally, retry sync_retries times with period of
 219         sync_refresh_period/8.
 220
 221 sync_retries - INTEGER
 222         default 0
 223
 224         Defines sync retries with period of sync_refresh_period/8. Useful
 225         to protect against loss of sync messages. The range of the
 226         sync_retries is from 0 to 3.
 227
 228 sync_qlen_max - UNSIGNED LONG
 229
 230         Hard limit for queued sync messages that are not sent yet. It
 231         defaults to 1/32 of the memory pages but actually represents
 232         number of messages. It will protect us from allocating large
 233         parts of memory when the sending rate is lower than the queuing
 234         rate.
 235
 236 sync_sock_size - INTEGER
 237         default 0
 238
 239         Configuration of SNDBUF (master) or RCVBUF (slave) socket limit.
 240         Default value is 0 (preserve system defaults).
 241
 242 sync_ports - INTEGER
 243         default 1
 244
 245         The number of threads that master and backup servers can use for
 246         sync traffic. Every thread will use single UDP port, thread 0 will
 247         use the default port 8848 while last thread will use port
 248         8848+sync_ports-1.
 249
 250 snat_reroute - BOOLEAN
 251         0 - disabled
 252         not 0 - enabled (default)
 253
 254         If enabled, recalculate the route of SNATed packets from
 255         realservers so that they are routed as if they originate from the
 256         director. Otherwise they are routed as if they are forwarded by the
 257         director.
 258
 259         If policy routing is in effect then it is possible that the route
 260         of a packet originating from a director is routed differently to a
 261         packet being forwarded by the director.
 262
 263         If policy routing is not in effect then the recalculated route will
 264         always be the same as the original route so it is an optimisation
 265         to disable snat_reroute and avoid the recalculation.
 266
 267 sync_persist_mode - INTEGER
 268         default 0
 269
 270         Controls the synchronisation of connections when using persistence
 271
 272         0: All types of connections are synchronised
 273         1: Attempt to reduce the synchronisation traffic depending on
 274         the connection type. For persistent services avoid synchronisation
 275         for normal connections, do it only for persistence templates.
 276         In such case, for TCP and SCTP it may need enabling sloppy_tcp and
 277         sloppy_sctp flags on backup servers. For non-persistent services
 278         such optimization is not applied, mode 0 is assumed.
 279
 280 sync_version - INTEGER
 281         default 1
 282
 283         The version of the synchronisation protocol used when sending
 284         synchronisation messages.
 285
 286         0 selects the original synchronisation protocol (version 0). This
 287         should be used when sending synchronisation messages to a legacy
 288         system that only understands the original synchronisation protocol.
 289
 290         1 selects the current synchronisation protocol (version 1). This
 291         should be used where possible.
 292
 293         Kernels with this sync_version entry are able to receive messages
 294         of both version 1 and version 2 of the synchronisation protocol.