Scheduling LTM config changes
Hi, I am looking for best (management and performance wise) to achieve setup for scenario below. Wonder if it could be done just via iRule or maybe iCall or iControl REST. In advance I have to say that for whatever reason there is no way to use GTM here. Scenario: Two DC, primary and backup During normal operation only servers from primary should respond to request When server from primary DC is down there are two options: VS type 1 - traffic should not be automatically switched to servers in backup DC. Mail or sms notification should be generated and based on decision of person responsible for service traffic should be switched or not. VS type 2 - traffic should be automatically switched to servers in backup DC. After switching to backup DC VS should change mode to type 1 - so no automatic fallback to primary DC after servers there become again up VS in primary DC can be for some time type 2 for other type 1. It should be possible to set schedule (with minute precision, possibly spanning few days) for VS's so at given time it will switch from type 1 to type 2 and at the end of scheduled period switches back from type 2 to type 1. It should as well take in account fact that switching types can occur only when VS is using servers from primary DC, when VS is using servers from backup DC (primary servers failed) it has always be in mode type 1 - mode change (even if scheduled) should be turned off until VS will be again using servers from primary DC. My idea (with fully manual management) is to configure system like that: VS type 1 has pool containing pool members from primary DC and backup DC, members from backup DC are set to Disabled or Forced Offline (I guess Disabled is enough, and if I am not wrong in this case members are still checked by configured monitor). In case of failure either: Members from primary DC are set to Forced Offline, pool members from backup DC are set to Enabled After resolving issue in primary DC configuration is restored to original state VS type 2. has pool with members from both DCs. Members from primary DC has higher priority than members from backup DC. Priority Group Activation (PGA) is configured. In case of failure: Traffic is automatically switched to backup DC members via PGA config, members from primary DC are set to Disabled so VS is becoming type 1 - no automatic fallback in case members from primary DC are back online When operator decides to use members from primary DC and change VS back to type 2 then members from primary DC are enabled - traffic should after some time completely move from backup members to primary If operator decides that VS should stay as type 1 primary members are enabled, backup members are disabled - so after no new connections accepted for backup members and after some time all connections will be only to primary members Now there is a question if above makes sense, or if maybe there is better way and finally which technology mentioned at the beginning of the post should be used. As side question I wonder what is simplest way to set mail notifications for member down/up events. Could it be set on BIG-IP device ot rather on some external monitoring system using SNMP traps? Other requirement I am not sure how can be set directly on BIG-IP is triggering execution of script on external machines before switching traffic between DCs and after successful switch. Piotr