table and HA

Hi Kai,

Thanks a lot. I am not sure if I understand this correctly:

"StateMirror.MirrorSessions" controls this behavior and yes it's enabled by default and imediatly starts to replicate your session information once you've created the cluster group with valid session mirroring configuration. No need to enable VS specific settings...

So what exactly enabling or disabling connection mirroring for VS or SNAT is doing? If I understand correctly what you stated above whole session table is mirrored right away - no matter if connection mirroring is enabled on any object - is that correct? Or by session information you means only keys stored in sub/table by iRules?

Based on article you pointed me to (appreciate - it explains exactly what I was not sure about) I have to assume that table data can't be accessed cross TG 😞 but it was in the pre 11.4 versions

I was looking for fast and reliable way to exchange table data between separate devices/VE. My idea was to create HA config (but not exactly to achieve HA) to spread load between two (or more devices) keeping ability to reference same sub/table keys. Sure mod operations on subtables could be performed only in one TG but other TGs could read keys from this sub/tables. That way decisions could be made on given device (standby for TG) based on info stored by another one (active for TG).

Sure can be done via some sideband or VS directed requests but that seems to be much more complicated and not as fast and reliable.

Well, in relation to SOL13478 I am not talking about tcp session data but table keys. And actually I am not sure what failback means in "Note: Only FastL4 and SNAT connections are re-mirrored after failback" - so device A active - sub/table data copied to deice b, then device A fails so device B is active and using/updating table data, then device a back online - assume that now sub/table data is copied to device A - or I am wrong?. Then I am doing manual failover to device A - I assume that no sub/table data should be lost? So in what situation and what data is lost during "failback"?

Piotr