Bug #79025 | Tablespace traversal/validation on recovery: performance regression 5.7 vs 5.6 | ||
---|---|---|---|
Submitted: | 29 Oct 2015 22:14 | Modified: | 19 Dec 2015 14:24 |
Reporter: | Szymon Komendera | Email Updates: | |
Status: | Can't repeat | Impact on me: | |
Category: | MySQL Server: InnoDB storage engine | Severity: | S5 (Performance) |
Version: | 5.7 | OS: | Any |
Assigned to: | CPU Architecture: | Any |
[29 Oct 2015 22:14]
Szymon Komendera
[16 Dec 2015 9:43]
MySQL Verification Team
Hello Szymon, Thank you for the report and test case. I'm not seeing that much difference at my end when comparing 5.6.28 with 5.7.10 start up time(binary tarball used on OL7 with default conf). Could you please try with latest GA, and provide test results along with configuration file used for tests? -- my local test results with default conf // 5.6.28 2015-12-16 08:57:41 17932 [Note] Plugin 'FEDERATED' is disabled. . 2015-12-16 08:57:41 17932 [Note] InnoDB: Starting crash recovery. 2015-12-16 08:57:41 17932 [Note] InnoDB: Reading tablespace information from the .ibd files... 2015-12-16 08:58:13 17932 [Note] InnoDB: Restoring possible half-written data pages 2015-12-16 08:58:13 17932 [Note] InnoDB: from the doublewrite buffer... InnoDB: Doing recovery: scanned up to log sequence number 937716651 2015-12-16 08:58:13 17932 [Note] InnoDB: Starting an apply batch of log records to the database... InnoDB: Progress in percent: 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 InnoDB: Apply batch completed 2015-12-16 08:58:15 17932 [Note] InnoDB: 128 rollback segment(s) are active. 2015-12-16 08:58:15 17932 [Note] InnoDB: Waiting for purge to start 2015-12-16 08:58:15 17932 [Note] InnoDB: 5.6.28 started; log sequence number 937716651 . Version: '5.6.28-enterprise-commercial-advanced' socket: '/tmp/mysql_ushastry.sock' port: 15000 MySQL Enterprise Server - Advanced Edition (Commercial) // 5.7.10 . 2015-12-16T08:15:51.237413Z 0 [Note] InnoDB: Starting crash recovery. 2015-12-16T08:15:51.244103Z 0 [Note] InnoDB: Starting an apply batch of log records to the database... InnoDB: Progress in percent: 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 2015-12-16T08:15:51.751037Z 0 [Note] InnoDB: Apply batch completed 2015-12-16T08:16:05.174772Z 0 [Note] InnoDB: Removed temporary tablespace data file: "ibtmp1" 2015-12-16T08:16:05.174809Z 0 [Note] InnoDB: Creating shared tablespace for temporary tables . . 2015-12-16T08:16:05.485620Z 0 [Note] bin/mysqld: ready for connections. Version: '5.7.10-enterprise-commercial-advanced' socket: '/tmp/mysql_ushastry.sock' port: 15000 MySQL Enterprise Server - Advanced Edition (Commercial) 2015-12-16T08:16:06.458616Z 0 [Note] InnoDB: Buffer pool(s) load completed at 151216 9:16:06 Thanks, Umesh
[19 Dec 2015 14:24]
Szymon Komendera
Hi Umesh, Thanks for spending your time on the repro. I re-ran the tests against 5.6.27, 5.7.9 and 5.7.10 (source distribution, compiled without debugging). - More powerful instance type: EC2 m4.2xlarge, 500GB / 5000 IOPS - Same exact OS installation as the last time (Debian Wheezy 7.9) I'm attaching a file "Bug79025_iteration2.zip" with all the data you should need. Unfortunately I've no access to a powerful enough bare-metal machine but let me know if you want me to use another OS or MySQL distribution on EC2. I did try out the binary distribution during previous iteration but I'm happy to do it again (whatever you need to avoid unnecessary legwork on your side). - Szymon
[19 Dec 2015 14:25]
Szymon Komendera
Bug79025_iteration2
Attachment: Bug79025_iteration2.zip (application/zip, text), 550.53 KiB.