Document Type | Troubleshooting
Category | Installation
Applicable Product Versions | 7FS02, 7FS02PS
Document Number | TINTS018
Issue
During installation, an Internal Error with condition '!"DB DOWN: all online log members missing"' (tc_lgwr.c:2658) error occurs, causing the Tibero Instance to go down.
Cause
_LGWR_USE_O_DIRECT parameter (which uses o_direct during redo logfile open, default is Y) when present, attempts Direct I/O to write directly to disk without using the OS buffer. During this process, it checks whether the Disk Logical Sector Size matches the Redo LogFile Block Size (default 512 bytes). If these sizes do not match, the Direct I/O for the Redo LogFile fails, resulting in the error.
# fdisk -l /dev/sda
Disk /dev/sda: 2.2 TiB, 2399812976640 bytes, 585891840 sectors
Units: sectors of 1 * 4096 = 4096 bytes
Sector size (logical/physical): 4096 bytes / 4096 bytes != redo logfile block size (512byte)
I/O size (minimum/optimal): 16777216 bytes / 16777216 bytes
Solutions
Case 1. Change the _LOG_BLOCK_SIZE parameter to match the Disk Logical Sector Size
If the value of the _LOG_BLOCK_SIZE parameter, which specifies the Redo LogFile Block Size, matches the Disk Logical Sector Size, the Direct I/O for the Redo LogFile succeeds and the error does not occur.
$ vi $TB_HOME/config/$TB_SID.tip _LOG_BLOCK_SIZE=4096 _LGWR_USE_O_DIRECT=Y Restart after applying
Case 2. Change the Disk Logical Sector Size to 512 Bytes and then proceed with Tibero installation
Change the Disk Logical Sector Size to 512 Bytes, which is the default Redo LogFile Block Size, using the method recommended by the disk manufacturer, and then proceed with the installation.