CentOS 4.5 and MySQL issues
Hello List, I have a weird problem with the above server using MySQL (4.1): 1. Some tables on specific database are loosing the Auto Inc pointer, making it start from 0 every time there is a restart to the server itself. 2. It seems that the MySQL server have a big delay (of a minute and sometimes more) between the time I update/delete records using a program (written in Java 1.6 and mysql connector), until the change is actually taking place. I thought that it was a cache issue, but it does seems that I was able to disable the query caching: mysql SHOW VARIABLES LIKE '%cach%'; +--++ | Variable_name| Value | +--++ | bdb_cache_size | 8388600| | binlog_cache_size| 32768 | | have_query_cache | YES| | key_cache_age_threshold | 300| | key_cache_block_size | 1024 | | key_cache_division_limit | 100| | max_binlog_cache_size| 4294967295 | | query_cache_limit| 1048576| | query_cache_min_res_unit | 4096 | | query_cache_size | 0 | | query_cache_type | OFF| | query_cache_wlock_invalidate | OFF| | table_cache | 64 | | thread_cache_size| 0 | +--++ I can not figure out how to make have_query_cache as no, but as I said I don't think there is a cache set, after I disabled query_cache_type. Any hints, tips or RTFM that I might have missed in this subjects that can help ? Thanks, Ido -- http://ik.homelinux.org/ = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: CentOS 4.5 and MySQL issues
On Tue, 2007-12-11 at 18:00 +0200, ik wrote: Hello List, I have a weird problem with the above server using MySQL (4.1): You need to provide more information about MySQL - what engine your tables are using, any replication or clustering taking place, etc'. 1. Some tables on specific database are loosing the Auto Inc pointer, making it start from 0 every time there is a restart to the server itself. Can't see any reason for this to happen unless something actually resets it (it can be done). Are you using NDB on these tables by any chance ? 2. It seems that the MySQL server have a big delay (of a minute and sometimes more) between the time I update/delete records using a program (written in Java 1.6 and mysql connector), until the change is actually taking place. As I mentioned earlier, this is due to transaction isolation between multiple users on the same table - the records are updated immediately (or at least as soon as they are committed) but the user doing the select only sees it after its own transaction completes. -- Oded = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: CentOS 4.5 and MySQL issues
On Dec 12, 2007 2:03 AM, Oded Arbel [EMAIL PROTECTED] wrote: On Tue, 2007-12-11 at 18:00 +0200, ik wrote: Hello List, I have a weird problem with the above server using MySQL (4.1): You need to provide more information about MySQL - what engine your tables are using, any replication or clustering taking place, etc'. The table that is effected is InnoDB. I do not use clustering at all, and it happens on every restart of the system. 1. Some tables on specific database are loosing the Auto Inc pointer, making it start from 0 every time there is a restart to the server itself. Can't see any reason for this to happen unless something actually resets it (it can be done). Are you using NDB on these tables by any chance ? I first noticed it, after the server was taken down after adding it some PCI hardware, but when I researched it before, I did found earlier indications to that issue. I did made then update on the auto inc field to make it indicate the right value. And no, I do not use NDB. 2. It seems that the MySQL server have a big delay (of a minute and sometimes more) between the time I update/delete records using a program (written in Java 1.6 and mysql connector), until the change is actually taking place. As I mentioned earlier, this is due to transaction isolation between multiple users on the same table - the records are updated immediately (or at least as soon as they are committed) but the user doing the select only sees it after its own transaction completes. The problem is, that the select arrived after I already closed the connection back to the connection pool of Java. And it is set to auto commit. Another issue, is that this happens only on specific machine and after a restart of the machine, while prior to that, things acted as expected, so there must be something that changed, but I cant put my finger on what was changed. -- Oded Ido -- http://ik.homelinux.org/ = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]