I posted this already at Database Administration (MOSC) but maybe it is seen by more people here:
On April 19th we had a problem for the fourth time that the Oracle service on a Windows server terminated suddenly. The Windows event log says: “ID 7034: The OracleServiceDWH1TCDB service terminated unexpectedly. It has done this … time(s).” Fortunately, it is only a test server so there is no impact on our business processes.
In all cases the last entry in the alert log before the crash was "Warning: VKTM detected a forward time drift."
A second common feature of the crashes is that they always happened a few minutes before the half hour. The exact timestamps are as follows:
06.06.19 01:27:22,803000000 +02:00 (WINDOWS DB BUNDLE PATCH 12.2.0.1.170831(64bit):26499442)
22.09.19 01:26:14,748000000 +02:00 (WINDOWS DB BUNDLE PATCH 12.2.0.1.190530(64bit):29632752)
01.12.19 09:27:08,322000000 +01:00 (WINDOWS DB BUNDLE PATCH 12.2.0.1.190530(64bit):29632752)
19.04.20 05:27:47,196000000 +02:00 (WINDOWS DB BUNDLE PATCH 12.2.0.1.191015(64bit):30150416)
As you can see, I have added the respective patch level, the database had at the time of the crash.
The database was installed in September 2017 but we didn’t experience such problems in the first year and a half but then about once a quarter. The database is running on a Windows 2016 server in a VMware ESXi 6.5 two-node cluster. But there are four more quite similar Oracle 12.2 servers on this cluster, that don’t have this special problem.
On the affected server every hour at the minutes :26 and :56 an Oracle database job starts, which refreshes several materialized views via a database link from another 12.2 Oracle server. Normally this job takes less than 20 seconds. The same job is running on a production server, which doesn’t experience this problem.
I found a few entries on the internet and some notes on Oracle support, which link the error "VKTM detected a forward time drift" to server crashes, but it seems to me that this information concerns older database versions and the bugs should be fixed in our version.
Last year I started a service request at Oracle support, but it didn’t lead to a real solution and at some point, I gave up. The last answers in October 2019 were: “Note 2313175.1 is the solution for this and applying the latest bundle patch.” and then “Based on the provided data so far we have already given possible causes and fixes.”
I will probably start a new SR again, but first I want to ask here if anyone has similar experiences or a good hint.