I bet you are using Microsoft Access to view the data. From what I
remember, Access prints #Deleted if the record that you try to update
gets deleted (or if the primary key for the record gets updated) while
you are looking at it. In other words, if you are looking at the
record that has
custid = 317, cust_name = 'John P Smith'
and you type "John Q Smith" in the cust_name, field, then ACCESS tries
UPDATE table SET cust_name = 'John Q Smith' WHERE custid = 317
behind your back. If it finds that no rows were updated, then it
concludes that the row with custid = 317 must not exist anymore, and
prints #Deleted in all of the fields for that row. (It may also
periodically check to see that a record is still available, in which
case the #Deleted might appear without you trying to edit anything.)
Does this table ever get updated or deleted from? If so, check who else
is deleting or updating while you are looking at the data. You should
make a careful check to see if this is the case before assuming that
this is a problem with postgres.
>On Fri, 23 Apr 2004, Jerry Robertson wrote:
>>We have been running Postgre for over a year and are very pleased with its ease of use and performance.
>>We have encountered one problem that has been minor until today and that is:
>>Occasionally a row in a table get populated with #Deleted comments. The entire row can not be deleted. Generally
>>this does not get displayed, however, today it is. Can you help?
>Are these rows that maybe your application is doing this to, or are these
>rows that are getting deleted in postgresql but still somehow showing up?
>It sounds like you've either found a very rare occuring bug (no one else
>has reported anything like this that I've seen) or maybe are having some
>kind of hardware issue where bits aren't getting set quite right.
>When postgresql "deletes" a tuple prior to it being vacuumed, it simply
>sets a time stamp on it that vacuum can use to see it is "deleted" and no
>longer visible. If you are actually getting back tuples with the words
>#Deleted in them, then I would guess you have an application error.
>---------------------------(end of broadcast)---------------------------
>TIP 5: Have you checked our extensive FAQ?
In response to
pgsql-general by date
|Next:||From: Paul Tillotson||Date: 2004-04-28 03:18:22|
|Subject: Re: Arbitrary precision modulo operation|
|Previous:||From: Brent Wood||Date: 2004-04-28 02:15:45|
|Subject: help building datetime from varchars|