We help IT Professionals succeed at work.

sql query question

gopher_49 asked
Medium Priority
Last Modified: 2012-06-21
I'm using an application called GFI Mail Archiver.  All emails coming in and out of our network are dumped to a SQL database that is searchable via a .NET website.  When first implementing this it archived thousands of undesired emails.  I would like to perform a query to find these emails and delete them.  Below are a series of commands I found that will do the trick.  What will the 'where full_subject like' command find?  Can I use a wildcard with this command like:

where full_subject like 'Bayesian' (any email with Bayesian in the subject will be found in the query?  Below is the series of commands...

declare @tbl table
(id int)
-- filter messages to be deleted by subject
insert into @tbl
select id from arc
where full_subject like 'j.bloggs@domian.com - Bayesian filter'
delete from arc_sources where id in (select id from @tbl)
delete from arc_bodies where id in (select id from @tbl)
delete from arc_att where id in (select id from @tbl)
delete from arc_add where id in (select id from @tbl)
delete from arc_owners where id in (select id from @tbl)
delete from arc where id in (select id from @tbl)
delete from arc_threads
where NOT EXISTS(select id from arc where thread_id=arc_threads.thread_id)

Watch Question

where full_subject like '%Bayesian%'
might be slightly more efficient to use charindex instead of like:

where charindex('Bayesian', full_subject) > 0

Not the solution you were looking for? Getting a personalized solution is easy.

Ask the Experts


when using where charindex('Bayesian', full_subject) > 0 - any subject with 'Bayesian' listed somewhere in the subject will be deleted?  If so, this will work just fine.
If 'Bayesian Filter' is always the last 2 words in the full subject field then use this:

where right(full_subject,15) = 'Bayesian filter'
Yes, these two statements are functionally identical.  I prefer charindex because i am not a huge fan of Like

where FieldName like '%TEXT%'

where charindex('TEXT',FieldName) > 0


I'll try it this evening.  I'll get back with you in the morning.



The keywords I'll be using will fall in different parts of the subject line.  Below is an example of a subject that will have the word 'Bayesian' in it.  Will the where charindex('TEXT',FieldName) > 0 work?

user@domain.com - Bayesian Filter detected spam - Match the Whole Family in Cozy Sleepwear, Plus 15% Off


great.  I plan to to delete  a TON of emails based on this.  I would say around 200,000 or so.  I'll let you know if and when my server fries!  I'll get back with you first thing in the morning.



I have one more question.  What if the subject has the phrase 'sending mail server'.  And I'm using 'sending mail server' in the where charindex command.  Will it require the subject to have those words in that order?  That is what I would prefer to void deleting emails that shouldn't be deleted.  The only emails that will have 'sending mail server' in the subject are emails that have been marked as spam and I would like to delete them from the db.


the charindex('Bayesian', full_subject) > 0 command worked incredibly well in conjunction with the syntax mentioned above.  I cleared up over 250,000 spam emails.  I now have exclusions enabled to void archiving emails that are marked as spam.  The above syntax got rid of the junk that was entered into the archive before me getting a chance to enable the exclusion.  Anyway, it worked perfect.

Access more of Experts Exchange with a free account
Thanks for using Experts Exchange.

Create a free account to continue.

Limited access with a free account allows you to:

  • View three pieces of content (articles, solutions, posts, and videos)
  • Ask the experts questions (counted toward content limit)
  • Customize your dashboard and profile

*This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.


Please enter a first name

Please enter a last name

8+ characters (letters, numbers, and a symbol)

By clicking, you agree to the Terms of Use and Privacy Policy.