From: | "Ed L(dot)" <pgsql(at)bluepolka(dot)net> |
---|---|
To: | pgsql-general(at)postgresql(dot)org |
Subject: | DROP INDEX performance/locking |
Date: | 2006-11-03 17:46:18 |
Message-ID: | 200611031046.18144.pgsql@bluepolka.net |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
We're taking some politically expensive downtime to drop an index
on a very large, heavily used table because to do it while live
blocks too many users for too long, requiring tens of seconds or
more on a system doing 200 transactions/second. That's due to
the fact that nearly every user interaction touches this
particular table.
It seems like one ought to be able to drop an index in a much
shorter amount of time and/or not block queries while it's
dropped. But I don't know the code well enough to say how hard
this would be.
Would a faster drop or non-blocking drop be worthy of a todo
item?
Ed
From | Date | Subject | |
---|---|---|---|
Next Message | Richard Broersma Jr | 2006-11-03 17:50:04 | Re: I can not add up columns when using a left outer join. Any ideas as to why? |
Previous Message | Bill Moran | 2006-11-03 17:24:09 | Re: fetching unique pins in a high-transaction |