| Bug #87657 | Trailing # comment on !includedir / !include directive causes error | ||
|---|---|---|---|
| Submitted: | 4 Sep 2017 12:56 | Modified: | 5 Sep 2017 14:11 |
| Reporter: | Oli Sennhauser | Email Updates: | |
| Status: | Verified | Impact on me: | |
| Category: | MySQL Server: Options | Severity: | S3 (Non-critical) |
| Version: | 5.7 | OS: | Any |
| Assigned to: | CPU Architecture: | Any | |
[4 Sep 2017 12:56]
Oli Sennhauser
[4 Sep 2017 13:50]
Peter Laursen
Maybe I am missing something, but I don't see how mysqld should be able to write at all (to error log or any file) when it is not running. What do you suggest: it should then ignore the erroneous option, start, write to log and next abort? Or "inject/pass on" some information to the system (systemctrl or whatever) that the system next can/will log? -- Peter -- not a MySQL/Oracle person
[4 Sep 2017 16:29]
Terje Røsten
Seems to be dup of https://bugs.mysql.com/bug.php?id=87087: Bug#87087 mysqld completely silent even in failure if log-error is not writable
[4 Sep 2017 16:42]
Oli Sennhauser
I am not so sure if this is the same problem. Because MySQL is/was writeable to MySQL error log file. It somehow looks more like the error message was not redirected to log/systemd error trap. I think I have seen this already in older version but was too lazy to file a but.
[5 Sep 2017 10:30]
Terje Røsten
Hi Oli! I can't reproduce: $ grep -v \# /etc/my.cnf [mysqld] datadir=/var/lib/mysql socket=/var/lib/mysql/mysql.sock symbolic-links=0 log-error=/var/log/mysqld.log pid-file=/var/run/mysqld/mysqld.pid !includedir /invalid/mydir $ systemctl start mysqld Job for mysqld.service failed because the control process exited with error code. See "systemctl status mysqld.service" and "journalctl -xe" for details. $ journalctl -xe |grep invalid|head -3 Sep 05 12:18:09 vide27 mysqld_pre_systemd[10535]: my_print_defaults: Can't read dir of '/invalid/mydir' (Errcode: 2 - No such file or directory) Sep 05 12:18:09 vide27 mysqld_pre_systemd[10535]: my_print_defaults: Can't read dir of '/invalid/mydir' (Errcode: 2 - No such file or directory) Sep 05 12:18:09 vide27 mysqld[10563]: mysqld: Can't read dir of '/invalid/mydir' (Errcode: 2 - No such file or directory) This is with MySQL 5.7.19 RPMS on EL7.
[5 Sep 2017 12:23]
Oli Sennhauser
Hello Terje I tried to reproduce as well an I can confirm your finding. Error message is displayed with journalctl -xe. So either I did not look well or I did something else wrong. I will investigate further... Never the less: The error message is wrong. This is my test case: # Fresh MySQL 5.7 installation via MySQL repository systemctl stop mysqld cat >> /etc/my.cnf << _EOF [mysqld] !includedir /etc/my.cnf.d/ # this directory exists! log_bin = bin-log _EOF systemctl start mysqld Sep 05 14:03:42 centos7tpl mysqld[2496]: mysqld: Can't read dir of '/etc/my.cnf.d/ # this directory exists!' Sep 05 14:03:42 centos7tpl mysqld[2496]: mysqld: [ERROR] Fatal error in defaults handling. Program aborted! This directory is perfectly readable. The problem is the comment AFTER the !includedir directive... --> this works !includedir /etc/my.cnf.d --> this does not work !includedir /etc/my.cnf.d # some comment here Regards, Oli
[5 Sep 2017 14:11]
Terje Røsten
Using a trailing # comment in my.cnf on !includedir or !include directives breaks parsing or makes directive silently ignored. Example 1): Adding this line !includedir /etc/my.cnf.d # comment breaks parsing in my.cnf breaks everything after '!includedir ' is regarded as argument: $ my_print_defaults mysqld my_print_defaults: Can't read dir of '/etc/my.cnf.d # comment breaks parsing' (Errcode: 2 - No such file or directory) Example 2) With the line !include /etc/foo.cnf # comment breaks parsing in my.cnf MySQL looks for file named 'etc/foo.cnf # comment breaks parsing' don't find it and ignores the error.
