Bug #35681 | mysqld slave fails to replicate timestamps after daylight saving change | ||
---|---|---|---|
Submitted: | 30 Mar 2008 11:10 | Modified: | 5 May 2008 19:29 |
Reporter: | Raúl García | Email Updates: | |
Status: | Closed | Impact on me: | |
Category: | MySQL Server: Data Types | Severity: | S3 (Non-critical) |
Version: | 5.0.45-community-log | OS: | Any |
Assigned to: | CPU Architecture: | Any | |
Tags: | daylight saving, replication, slave, timestamp |
[30 Mar 2008 11:10]
Raúl García
[31 Mar 2008 10:43]
Raúl García
Edit: Some Linux slaves fail as well, so the problem is somewhere else. Master has ECT (USA's East Coast Time), and so do all working slaves. Non-working slaves have different time zones, such as CET (Central European Time), so what really happened is european slaves stopped working after time change in Europe (CET to CEST saving time).
[31 Mar 2008 17:40]
Sveta Smirnova
Thank you for the report. Please provide output of SELECT @@sql_mode on both master and slave.
[31 Mar 2008 18:07]
Raúl García
"SELECT @@sql_mode;" returns, in all masters and slaves, an empty record: mysql> SELECT @@sql_mode; +------------+ | @@sql_mode | +------------+ | | +------------+ 1 row in set (0.00 sec) Tomorrow I will try to set the timezone on the master to CEST, just as the failing slaves. If it's a problem related to that, failing slaves should begin working and working slaves should fail. I usually use INSERT with the function now() in them, such as "INSERT INTO table 1 now() AS ts;". Slaves always replicate this instruction, but even if they are hours slow from master, they still get the right Timestamp instead of setting it two hours later. Could this be related?
[31 Mar 2008 18:24]
Sveta Smirnova
Please also update to current version 5.0.51a. At least one related bug #29536 was fixed after version 5.0.45 has been released.
[31 Mar 2008 19:04]
Raúl García
I will do that this week, but although this may probably solve the problem in the future (current replication will stay wrong, so I'll have to change all timezones), there's still something else going on, why does the master let me insert with milliseconds (discarding millisecond information), but slaves don't? I am, in almost all cases, using a standard MySQL installation, and in all cases I get a blank sql_mode record.
[2 Apr 2008 20:23]
Sveta Smirnova
Problem is not milliseconds, but not existant date due daylight saving change. Strange why slave is affected. This is why I think you can be affected by bug #29536
[2 May 2008 23:00]
Bugs System
No feedback was provided for this bug for over a month, so it is being suspended automatically. If you are able to provide the information that was originally requested, please do so and change the status of the bug back to "Open".
[5 May 2008 8:50]
Raúl García
Problem was fixed after updating to latest MySQL 5. Thank you for your support. You may close this ticket.
[5 May 2008 19:29]
Sveta Smirnova
Thank you for the feedback. Closed regarding to last comment.