SDS_LOGCHECK configuration parameter
Use the SDS_LOGCHECK configuration parameter to set the number of seconds to delay the secondary server from taking over the role of the primary server. If the secondary server detects that the primary server is generating log records during the delay period, then the failover is prevented. The delay can prevent an unnecessary failover if network communication between the primary and secondary servers is temporarily unavailable.
- onconfig.std value
- SDS_LOGCHECK 0
- values
0
= Do not detect log activity; allow immediate failover.n = Wait up to n seconds. If log activity is detected from the primary server, failover is prevented; otherwise, failover is allowed.
- units
- Seconds
- takes effect
- When shared disk functionality is enabled on the primary server
Usage
For example, if the SDS_LOGCHECK configuration parameter is set to 10, and the primary server fails, the SD secondary server waits up to 10 seconds to either detect that the primary server is generating log records (in which case failover is prevented), or the SD secondary server detects that the primary is not generating log records and failover occurs.
An unnecessary failover can result in two primary servers that are both receiving input from applications and writing to the same chunks, which can cause unrepairable data corruption.
Set the SDS_LOGCHECK configuration parameter to a value greater than zero if you do not have I/O fencing configured and your system consists of a primary server and one or more SD secondary servers.
If your system has I/O fencing configured, and if an SD secondary server becomes a primary server, the I/O fencing script must prevent the failed primary server from updating any of the shared disks. If the system does not have I/O fencing configured, the SDS_LOGCHECK configuration parameter prevents the occurrence of multiple primary servers by not failing over to the SD secondary server if the original primary server is generating log records.