Details
-
Bug
-
Status: Closed (View Workflow)
-
Critical
-
Resolution: Fixed
-
10.2(EOL), 10.3(EOL), 10.4(EOL), 10.5, 10.6
-
None
Description
2020-12-15 20:38:03 0 [ERROR] InnoDB: Invalid log block checksum. block: 811904837 checkpoint no: 10978 expected: 3929835436 found: 881240600
|
This causes user confusion because it is a transient error that can be retried and overcome. Per Marko: {
--backup will invoke the function in a loop:
for (int retries= 0; retries < 100; retries++) {
|
if (log_sys.log.read_log_seg(&lsn, end_lsn)
|
|| lsn != start_lsn) {
|
break;
|
}
|
msg("Retrying read of log at LSN=" LSN_PF, lsn);
|
my_sleep(1000);
|
}
|
If the function log_t::files::read_log_seg() fails, it will display the message:
ib::error() << "Invalid log block checksum."
|
<< " block: " << block_number
|
<< " checkpoint no: "
|
<< log_block_get_checkpoint_no(buf)
|
<< " expected: " << crc
|
<< " found: " << cksum;
|
If you are also seeing the "Retrying read of log" message, there should be nothing to worry about. Due to the inter-process communication via the file system, Mariabackup must be prepared to retry reading data pages (innodb_page_size bytes) or log blocks (512 bytes) that appear corrupted.
}
So could this be changed to a [Warning] until it stops retrying?