Details
-
Bug
-
Status: Closed (View Workflow)
-
Major
-
Resolution: Incomplete
-
10.2.6
-
CentOS 7.x x86_64
Description
The server is a brand new, dedicated mysql box, and here are the relevant variables in my.cnf
innodb_encryption_threads = 8
innodb_background_scrub_data_check_interval = 3600
innodb_background_scrub_data_compressed = 1
innodb_background_scrub_data_interval = 86400
innodb_background_scrub_data_uncompressed = 1
innodb_immediate_scrub_data_uncompressed = 1
innodb_scrub_log = 1
innodb_scrub_log_speed = 50000
After the service restarted, the total CPU usage is 0, and after a short while(several minutes, an hour, not quite sure), the total CPU usage grows to 15, and system load is up to 8. And after I comment the above variables, the system is back to normal.
Is there anything wrong with my configuration?
perf top -p ${pid}
Samples: 1M of event 'cycles:ppp', Event count (approx.): 278084696592
Overhead Shared Object Symbol
30.93% mysqld [.] fil_crypt_thread
29.82% mysqld [.] ut_delay
16.56% [kernel] [k] _raw_spin_lock
1.41% mysqld [.] PolicyMutex<TTASEventMutex<GenericPolicy> >::enter
Attachments
Issue Links
- relates to
-
MDEV-21870 Deprecate and ignore innodb_scrub_log and innodb_scrub_log_speed
-
- Closed
-
Activity
| Field | Original Value | New Value |
|---|---|---|
| Description |
The server is a brand new, dedicated mysql box, and here are the relevant variables in my.cnf
innodb_encryption_threads = 8 innodb_background_scrub_data_check_interval = 3600 innodb_background_scrub_data_compressed = 1 innodb_background_scrub_data_interval = 86400 innodb_background_scrub_data_uncompressed = 1 innodb_immediate_scrub_data_uncompressed = 1 innodb_scrub_log = 1 innodb_scrub_log_speed = 50000 After the service restarted, the total CPU usage is 0, and after a short while(several minutes, an hour, not quite sure), the total CPU usage grows to 15, and system load is up to 8. And after I comment the above variables, the system is back to normal. Is there anything wrong with my configuration? |
The server is a brand new, dedicated mysql box, and here are the relevant variables in my.cnf
innodb_encryption_threads = 8 innodb_background_scrub_data_check_interval = 3600 innodb_background_scrub_data_compressed = 1 innodb_background_scrub_data_interval = 86400 innodb_background_scrub_data_uncompressed = 1 innodb_immediate_scrub_data_uncompressed = 1 innodb_scrub_log = 1 innodb_scrub_log_speed = 50000 After the service restarted, the total CPU usage is 0, and after a short while(several minutes, an hour, not quite sure), the total CPU usage grows to 15, and system load is up to 8. And after I comment the above variables, the system is back to normal. Is there anything wrong with my configuration? perf top -p ${pid} Samples: 1M of event 'cycles:ppp', Event count (approx.): 278084696592 Overhead Shared Object Symbol 30.93% mysqld [.] fil_crypt_thread 29.82% mysqld [.] ut_delay 16.56% [kernel] [k] _raw_spin_lock 1.41% mysqld [.] PolicyMutex<TTASEventMutex<GenericPolicy> >::enter |
| Summary | scrub log cause high cpu load | scrub log cause high system load |
| Labels | need_feedback |
Also, when the CPU usage grows, does it become permanent, or does it raise and after a whole goes back to normal, and then the cycle repeats?
And when you are saying it's a brand new server, does it mean that the database is empty, there are no InnoDB users' tables, and there is no significant activity on the server before the spikes start happening?
| Component/s | Storage Engine - InnoDB [ 10129 ] | |
| Component/s | Server [ 13907 ] | |
| Fix Version/s | N/A [ 14700 ] | |
| Resolution | Incomplete [ 4 ] | |
| Status | Open [ 1 ] | Closed [ 6 ] |
| Link |
This issue relates to |
| Workflow | MariaDB v3 [ 81140 ] | MariaDB v4 [ 152283 ] |
Could you please attach the whole config file(s)?