975 reputation
1820
bio website lupolt.net
location London, United Kingdom
age 36
visits member for 2 years, 11 months
seen 3 hours ago

SQL Server DBA/developer


Dec
31
comment In SQL Server, can I guarantee an order without an explicit ORDER BY clause when an index seek is forced on a table with only a clustered index?
Tangential question: is there any performance or functionality benefit to having SELECT NULL vs SELECT * or SELECT 1 etc on the subquery inside the EXISTS condition, or was SELECT NULL chosen only for clarity?
Dec
30
comment Secondary Clustered DTC will not work when Primary goes offline - SQL Server Issues
To answer your other question, yes, if you wanted to make SQL Server automatically use DTC2 if and only if DTC1 failed, you'd have to delete DTC1 as well as any mappings of the SQL Server instance to DTC1 and make sure DTC2 is next in the order of evaluation listed in the documented I linked to. But I don't see any advantage in doing this versus using local DTC instances or configuring a DTC instance in each SQL Server group. With either of those options, Windows can be configured to bring up MSDTC if it fails, and you don't have to deal with telling SQL Server to use another DTC instance.
Dec
30
comment Secondary Clustered DTC will not work when Primary goes offline - SQL Server Issues
Yes, if a node failed, all resources from the failed node should come back online on the other node, and DTC should resume working.
Dec
30
revised Secondary Clustered DTC will not work when Primary goes offline - SQL Server Issues
Formatted list so that it doesn't appear as a run-on sentence
Dec
30
suggested approved edit on Secondary Clustered DTC will not work when Primary goes offline - SQL Server Issues
Dec
30
answered Secondary Clustered DTC will not work when Primary goes offline - SQL Server Issues
Dec
29
comment Secondary Clustered DTC will not work when Primary goes offline - SQL Server Issues
Which versions of Windows and SQL Server? Could you also explain the intended purpose of each DTC instance? For example, do you want each SQL Server instance associated with a different DTC instance?
Oct
12
comment How do I kill a rollback process when I want to drop the database?
I did something like that once and it turned out to be a Really Bad Idea. I'd rather just restart the instance than stomp on file handles and get potentially undefined behavior.
Oct
4
comment MySQL uses more than 500% of the CPU
The error message is about corrupt temporary tables due to filling up the MySQL tmpdir. I can't see any point in running a repair here.
Oct
1
awarded  Civic Duty
Sep
30
comment Will increased network latency cause table locks in MS SQL Server?
Interesting results. What client program/library is this with?
Sep
30
awarded  Explainer
Sep
29
comment why I can't delete old transaction log backup files with an existing scheduled job?
I'm not sure if I understand, but I think you need two maintenance plan cleanup tasks, 1 to delete transaction log backups (trn) and one to delete database backups (bak).
Sep
29
comment why I can't delete old transaction log backup files with an existing scheduled job?
In the maintenance plan, do you by any chance have the file extension to delete set to ".trn" instead of "trn"? If you include the period, it won't find any files to delete. This was a common annoyance with maintenance plans. Ola Hallengren's maintenance scripts are much easier to work with and generally more reliable.
Sep
28
revised Create failed for Availability Group 'AGN'
added 140 characters in body
Sep
28
answered Create failed for Availability Group 'AGN'
Sep
28
comment Create failed for Availability Group 'AGN'
Availability Groups only exist on SQL Server 2012 and up. I'm not sure why this is tagged with 2008.
Sep
28
comment How to get more information from dm_exec_query_stats
I'd suggest getting the value of last_execution_time from sys.dm_exec_query_stats. Then see what correlates with this in your website logs. You may need to enable logging of the query string (cs-uri-query) if it isn't already on. Searching the web server logs for the text that was injected probably won't work, as you found. One reason is that it might be encoded as hexadecimal.
Sep
28
comment MySQL Slave has unmanagbly higher IO load than master
I think your comment about seeks is probably looking in the right direction since random I/O is typically far slower than sequential. When I tested some SATA drives a few years ago, the maximum throughput for sustained 8 KB random writes on a 15k SATA RAID1 was under 5 MB/sec, and this was with the disks attached to a controller that had a write cache enabled, and on smaller disks that probably had a lower average seek time than the ones you're using.
Sep
27
comment Moved database (percona mysql) to larger disk, now everything is missing (though the files exist)
They shouldn't be empty if MySQL is starting. Can you log into MySQL? Does SHOW VARIABLES LIKE 'log_error' point to the name of an empty file? Do you get any error output if you try to restart MySQL?