Hi All,
I keep on having this on my cluster, in effect ndb on the nodes shuts down:
Status: Permanent error, external action needed
Message: Max number of open files exceeded, please increase
MaxNoOfOpenFiles (Resource configuration error)
Error: 2806
Error data:
Error object: Ndbfs::createAs
Thanks for your reply, Mike.
Yes, 13419851 rows were added to rmpdata1. However, 10 minutes seem to
be too long. I run the same join by using SQL procedure in a statistical
software called SAS on a similar machine. It only takes 1 minute and 3
seconds.
Yes, it is a 1:1 relationship between t
Again, please forgive my total ignorance.
My ERD shows that the web links (URL table) are connected, via the
sub-categories (SubCat table),
to the main categories (Categories table). Is this correct for what I
am trying to achieve ?
Or should I also link the URL table to the Categories
Jia,
Yes, it is a 1:1 relationship between table RItime and MVtime. However, I
don't get your suggestion, "I'd recommend joining the two tables into 1
table so you don't have to join them in the first place." Could you
elaborate that?
Sure but first I have to relate it to my own experience.
Hello All,
I'm reposting this since I didn't get much response the last time, so I'm
hoping to reach out again. My correlated update query (see below) was
running for 9 days before I killed it. Here is my original question:
I have a legacy application which was written using a compound prima
Hi Mike,
Thanks for your detailed answer. Now, I understand what you mean. And,
yes, I agree with you that keeping all data in one table works better
for a bunch of 1:1 relationship tables. Actually, this is what I was
trying to do with that query.
Since you mention "They all had a 1:1 rela
Jia,
The code you sent seems to be able to get the job done.
You could try something simpler by executing 2 sql statements instead of
using one. Something like:
create table rmpdata1
select ri.*, mv.* from RItime as ri left join MVtime as mv on
(ri.code=mv.code and ri.ndate=mv.ndate));
I'm looking for a way (better command line under linux) to reformat/
indent/beautify sql code like that (great) online (free) tool:
http://www.dpriver.com/pp/sqlformat.htm
or better, something to have that directly in emacs.
any help?
thank you in advance
bye bye
MAS!
--
MySQL General M
So I've disabled all the keys on the "item_trans" table -- since I am
updating every row, it wouldn't (shouldn't) be using the index anyway.
You can't disable unique indexes or primary keys. They are always active.
You can only deactivate non-unique indexes.
Here are a couple of suggestions
On Sun, Sep 6, 2009 at 6:01 PM, mos wrote:
>
> So I've disabled all the keys on the "item_trans" table -- since I am
>> updating every row, it wouldn't (shouldn't) be using the index anyway.
>>
>
> You can't disable unique indexes or primary keys. They are always active.
> You can only deactivat
10 matches
Mail list logo
