Speed up redundant failover

Discussion about recent product features & solutions!
Search

Post Reply
5 posts • Page 1 of 1
walterd
Posts: 38
Joined: Wed Mar 13, 2013 10:39 pm

Speed up redundant failover

Post by walterd »

Hello,
is there a way to speed up the fail over from the main to there secondary server ? I have a redundant system running with 3.16 on Linux servers, currently the time to failover to the second server is approx 8-9 seconds. Is that possible to have a faster failover ? I was looking to config options such as aliveTimeout or similar. Our specs calls for 5 sec or less failover time.

Thank You,

User avatar
leoknipp
Posts: 2926
Joined: Tue Aug 24, 2010 7:28 pm

Re: Speed up redundant failover

Post by leoknipp »

The time for the redundancy switch needs to take into account 2 parts:
-- Time to detect that the redundant server was lost
-- Time to execute the redundancy switch

Before we can tell you how to speed up the redundancy switch in your case we need to have more detailed information.
Possibly in that special case it is easier if you get in contact with your common WinCC OA support.

Best Regards
Leopold Knipp
Senior Support Specialist

walterd
Posts: 38
Joined: Wed Mar 13, 2013 10:39 pm

Re: Speed up redundant failover

Post by walterd »

The failover is quite fast is the first part, the detection that is causing problems. The moment the network is lost I takes 8 sec approx to start the failover, during this time the system is irresponsive.
The aliveTimeout is one of the parameter I was looking into, I have attached my config for reference
Thanks
Attachments
config.zip
(943 Bytes) Downloaded 396 times

walterd
Posts: 38
Joined: Wed Mar 13, 2013 10:39 pm

Re: Speed up redundant failover - SOLVED

Post by walterd »

FYI,
Problem solved changing in the config file , aliveTimeout = -3 (or 3 ,see the help for the difference) in [data] and [event]

User avatar
leoknipp
Posts: 2926
Joined: Tue Aug 24, 2010 7:28 pm

Re: Speed up redundant failover

Post by leoknipp »

For a detection of a lost connection between redundant servers you have to set the aliveTimeout for the [event] and [redu] section.

An entry in the [redu] section is necessary as the Redu Manager is also sending alive message to the redundant partner.
Please take into account that you must not use a timeout shorter than 3 seconds.
Before decreasing the timeouts you have to ensure that have a reliable and fast network with low latency times. If the network quality is not good, e.g. delayed transmission of packages, the system will possibly detect a loss of connection even if the network connection was not broken.

For the Data Manager [data] you must not use such a short timeout. Possibly during the recovery (when huge file information is transmitted via WinCC OA messages) the system will detect a timeout.

Best Regards
Leopold Knipp
Senior Support Specialist

Post Reply
5 posts • Page 1 of 1