NTP synchronization in 77.30 VSX cluster

[Kirill_Borozdin]

Hello everyone,

Hope you`re all doing well.

I need your help with one  question which might seem easy but I have some doubts and need your expert advice  before implementing this change on production.

I found that current time on one Multi-Domain management server and 77.30 VSX cluster which is managed by this Smart isn`t synchronized and has large difference (about 1 hour) to the real time of location. It`s due to wrong NTP server configured both on Smart and VSX cluster

I`m worried if so huge time change can lead to any problems with ClusterXL protocol and state synchronization between cluster members or not. I found that one guy had the same question on CPUG forum but got different opinion in answers and he also didn`t share results of change, was it successful or not:

How to configure NTP for existed Checkpoint cluster without downtime? 

Has someone met this situation before and what the correct way of time synchronization for such topology where both Smart and VSX has wrong time configured and this should be fixed. I also have an assumption that different time should not influence on  logging and ClusterXL but I decided to ensure this having your expert opinion.

Thanks in advance

BR,

Kirill Borozdin

[Vladimir]

Kirill,

I haven't done this myself in production so take this with the grain of salt:

If you have an HA cluster, you can take one member down (clusterXL_admin down), change time on the remaining member to correspond to that of Smart-1. If the time will advance, this should cause minimal interruptions to the traffic (sessions longer than defined values may drop). If it'll roll back, you may get bunch of session drops.

Change the time to ntp on this member.

Bring the second member's time in rough sync and change it to ntp.

Wait for clocks to sync completely and bring the second cluster member up.

If you are using a load sharing cluster, I'll suggest playing it safe and running this through TAC.