Heartbeat failures

  • Using basic mirroring configuration, within local VLAN. Heartbeat set at default (10 seconds). NIC cards software have been updated to latest version, using SQL Server 2008, Windows 2008. No network errors getreported, but Witness and/or mirror still drops out of quorum, briefly, on rare occasions. Management is asking me to develop an explanation; prior to opening a ticket with Microsoft, what should I be examining to determine why the Witness fails to respond for at least 10 seconds at a time. Do we need to establish a monitoring heartbeat under 10 seconds to capture a transient problem condition, or is there some other means of identifying correlated events to explain the failures? It seems that only one member of the mirror loses connectivity at any one time, so the mirror doesn't fail completely, and doesn't even fail over (which was happening before the NIC upgrades).

    Database is small with low activity/usage. Fewer than 150 tables, small number of records in any given table.

  • Some things to check ...

    Is the ping reponse < 10ms for SYNC mirror set up ? If yes there may network latency.

    Usually we have 30ms thrshold set for SYNC with no witness in our ennvironment, you try that as well in yours.

    Cheers

    Sat

    Cheer Satish 🙂

  • Satish,

    Most of the time the mirror works normally and the ping responds in minimum time value. It's an intermittent outage, which makes it very difficult to pin down. We are trying to determine longitudinally whether there is any consistent time to the outage, which would indicate a transient traffic condition. Lately, the Mirror has been logging the most errors, but the Witness as indicating unavailable (but the Principal does not report any such errors). Would that imply a problem between Mirror and Witness, or a problem on Mirror? Less frequent reports occur between Witness and Principal. Again, network monitoring and logs indicate no unusual events which would seem to imply the issue is not in the network.

Viewing 3 posts - 1 through 2 (of 2 total)

You must be logged in to reply to this topic. Login to reply