![]() Is there any trace flag that disables this new behaviour ? = SCRIPT = Perhaps this has something to do with the new « Concurrent PFS updates » of SQL Server 2019. We remove the RAM and replace it by another, with a different reference. recomputing statistics (UPDATE STATISTICS T_TIME_INTERVAL_TIV WITH FULLSCAN ).Some elements that seems to solve the problem: disabling hyperthreading (with SOFT NUMA enabled): on 90 tests only 1 is a failure.disabling hyperthreading (with SOFT NUMA disabled): on 10 tests only 1 is a failure.disabling SOFT NUMA (with hyperthreading enabled): the first 3 test are successes and last 7 are failure. ![]() Some elements that seems to decrease the occurrence of the phenomenon: While soft NUMA is enabled, we performed complementary tests by disabling SOFT NUMA, but it does not solve the problem.But no hardware information appears to be about the RAM in the servers (we performed a complete memory test). The « buffer latch » information likely indicates a memory fault.And the problem remains, even when we move the tempdb. Disk failure has been sweeped, because we create the database on the 3 different hard disk drives.Installing CU6 and after CU7 does not solve the problem.A time-out occurred while waiting for buffer latch time-out for buffer latch in pages of tempdbĪ time-out occurred while waiting for buffer latch - type 4, bp 00000292CE3D60C0, page 9:18634, stat 0x10b, database id: 2, allocation.But… information is recorded in the SQL Server event log: Of course, at no point did we implement the NOLOCK hint or the READ UNCOMMITTED isolation level. Msg 601, Level 12, State 1, Line … Could not continue scan with NOLOCK due to data movement. It is this last query (executed almost 10 times) which very often leads to an error with the following message: execution of the test request (calculation of aggregated time intervals – collapsing).enlarging the database files and switching to SIMPLE recovery mode.The test that generate the malfunctions is as follows: The problem appears systematically on the 2 new machines… 1 CPU : 1 Xeon Gold 6208U – 2.90 Ghz (16 cores x 2 – hyperthreading).Starting a new installation of SQL Server 2019 Enterprise (.5 / X64 on Windows Server 2019 Standard 10.0 / Build 17763) on 2 new physical machines (for AlwaysOn replicas). We have a very strange problem with a new SQL Server 2019 on 2 new physical machines:
0 Comments
Leave a Reply. |