This shows you the differences between two versions of the page.
Both sides previous revision Previous revision Next revision | Previous revision Last revision Both sides next revision | ||
app-notes:supervision [2015/04/05 13:46] hess |
app-notes:supervision [2015/04/20 09:42] fachet [Introduction] |
||
---|---|---|---|
Line 1: | Line 1: | ||
====== Supervision ====== | ====== Supervision ====== | ||
===== Introduction ===== | ===== Introduction ===== | ||
- | This note descibes which provisions are taken by NetModule Routers to guarantee always-on and always-connected functionality. | + | {{:app-notes:24-7.png?nolink&200 |}} |
- | {{ :pictures:supervision.png?nolink |}} | + | This note describes which provisions are taken by NetModule Routers to guarantee always-on and always-connected |
+ | functionality. | ||
+ | |||
+ | The NetModule Routers are designed for high reliability and 24x7 operations. To guarantee best connection availability | ||
+ | several levels and methods of supervision are used. | ||
+ | |||
+ | Depending on your hardware model, WAN links can be made up of either Wireless Wide Area Network (WWAN), | ||
+ | Wireless LAN (WLAN), Ethernet or PPP over Ethernet (PPPoE) connections. | ||
+ | |||
+ | The highest level of monitoring is the Ping Supervision which is individually configurable for each active (up) | ||
+ | link as well generic for any link based on the routing table. The user can define up to two host which are pinged | ||
+ | every "ping interval" seconds. If there is no response of at least one of the two hosts within "ping timeout" | ||
+ | milliseconds a retry is done using the "retry interval" time up to "max. number of trial fails". | ||
+ | If still there is no answer the defined emergency action is executed. It's recommended to set "restart link services" | ||
+ | on each link pinging the next hop e.g. the access point for a WiFi client link. | ||
+ | Restarting the link service means, that it will be tried to re-establish the link e.g. redial on a modem connection. | ||
+ | Emergency action "none" means that the link is defined as (inactive) down. | ||
+ | If you are using a VPN you should also enable the dead pear detection or keep alive mechanisms. | ||
+ | "ANY" is a shortcut for all WAN interfaces. | ||
+ | |||
+ | The** WWAN Manager** who is responsible for the control of the mobile network modems checks every 15 seconds | ||
+ | if the connection is still established. Different informations like configurable signal strength, | ||
+ | network registration status of service type are used to determine the status of the modem links. | ||
+ | |||
+ | The task of the **Link Manager** is to verify if a link is up or down and to select highest priority | ||
+ | link as the so-called hotlink which holds the default route for outgoing packets. | ||
+ | Every 5 seconds he checks for each active link if the following conditions are met: | ||
+ | |||
+ | ^Condition^ WWAN^ WLAN^ ETH^ PPPoE^ | ||
+ | |Modem is registered | X | | | | ||
+ | |Registered with valid service type | X | | | | | ||
+ | |Valid SIM state | X | | | | | ||
+ | |Sufficient signal strength | X | X | | | | ||
+ | |Client is associated | X | | | | | ||
+ | |Client is authenticated | X | | | | | ||
+ | |Valid DHCP address retrieved | X | X | X | X | | ||
+ | |Link is up and holds address | X | X | X | X | | ||
+ | |Ping check succeeded | X | X | X | X | | ||
+ | |||
+ | If at least one condition of a link is not met, the link manager counts the number of subsequent failures. | ||
+ | WWAN links are marked as down after 5 subsequent failures, PPPoE after three, WiFi and Ethernet after one. | ||
+ | The highest priority link which is up will become the so-called hotlink which holds the default route for | ||
+ | outgoing packets. Of course the link manager will change the hotlink as soon a higher priority interface comes up. | ||
+ | |||
+ | Finaly a **Hardware Watchdog** is available which will restart the device if he is not retriggered after x seconds. | ||
+ | The Router Suppervision deamon checks permanently if all components of the router application are still alive. | ||
+ | So in the unlikely case that the operating systems or the router software has an internal error the system restarts | ||
+ | completely by rebooting. This is the only condition where the system will reboot except "reboot system" | ||
+ | is configured in the link supervision. | ||
+ | |||
+ | ===== Summary ===== | ||
+ | - Level 4: VPN dead peer detection/keep alive | ||
+ | * Is VPN link over WAN link healthy? See also [[sdk:vpn-supervision|Supervision of a VPN Tunnel]]. | ||
+ | - Level 3: Link level Ping Supervision | ||
+ | * Is link to next network node healthy? | ||
+ | - Level 2: Modem/WiFi Supervision | ||
+ | * Is modem ok, device registered and a sufficient signal available ? | ||
+ | - Level 1: Link Supervison | ||
+ | * Is physical link up or physical connection ok? | ||
+ | - Level 0: Hardware Watchdog | ||
+ | * Is router firmware and operating system running ok? | ||
+ | |||
+ | |||
+ | ===== Configuration ===== | ||
+ | {{ :pictures:supervision.png?nolink |}} |