Thoughts On Whether The SSIS (service/agent) Is Really Necessary?
Jul 12, 2007
I just learned I can deploy and schedule jobs to run SSIS packages (via job/sqlagent) without the Integration Service (agent) itself actually running alongside (or on) the server. (Double-click on manifest, deploy IS package to server, create job/job step to run IS package, watch it run even when integration service is completely disabled)
Other than convenient viewing, configuring, and RMC running w/in SQL Server Mgmt Studio 2005, why then do I need the integration service running on a production box at all? When do I really need the IS service itself?
In our (finance) world only (a) an act of God or (b) a DBA can touch production databases/servers. Allowing anyone to connect to yet another service - in this case, an integration service - to meddle w/ a package would be a no no, so...
1) Could I trouble someone for a concrete, critical reason why the DBA should enable it on a production server. Speed? Caching? Peace of mind knowing everything is piled onto and neatly running on the server?
2) On a more minor note, if I'm deploying a package to be housed solely w/in MSDB, is there anyway to prevent the prompt of a file location during deployment, i.e. the creation of an empty directory that would otherwise hold package dependencies if I were running it as a file?
We'd like to deploy only to MSDB (I know all the pros/cons w/r/t saving dtsxs to files v. msdb) and keep deployments clean (read: all in one place). DR is via SAN-to-SAN replication with, among other things, msdb cleanly getting replicated. We would very much like to avoid having to worry about (more) file/directories sitting out on a server share to be replicated to DR (it seems the default is to allow deployments to directories on the SQL server instance itself..ugh) Any architectural insights on this would be appreciated.
what is considered best practice for privileges etc on the sql agent service account and long term need for that account to run ssis packages? I tried to understand and appreciate the article at http://www.microsoft.com/technet/prodtechnol/sql/2005/newsqlagent.mspx but felt like either it was overkill or I wasnt getting it.
I would appreciate any thoughts/ideas on the following use case for the distributed service broker application we plan to migrate from our existing proprietary tcp based message protocol using database tables for reliability.
There are two ssb services running in separate sql server instances, each on a different server machine. For simplicity, let us assume the ssb endpoint names are SSBA, SSBB. SSBB is the Initiator of the Dialog while SSBA is the Target. Now the requirement is that if the underlying network communication between the two ssb endpoints(SSBA and SSBB) is broken or if the critical service SSBB is down, then processing of any incoming message into SSBA's queue from a third service broker service (say SSBEXPR) running within a SqlExpress instance should be delayed until SSBB is alive and network communication between SSBA and SSBB is established. In our existing implementation (wherein SSBA, SSBB and SSBEXPR are windows services) we use a combination of TCP socket disconnects and Heartbeat messages between SSBA and SSBB to determine the health of network connection and that of the SSBB service.
Now my understanding of how the underlying network connection for a ssb dialog works is that if there is no activity on a dialog for a certain amount of time then the underlying network connection is closed. Is there a way to specify the amount of time to say infinite value or something and thus change this behavior? My other question is how can one query the underlying network connection (i.e. a row from sys.dm_broker_connections) associated with a particular conversation? If none of this is possible, then any other patterns/ideas/approach is welcome.
Hi all, I do understand that it is highly recomended to have aserprate user (perfered a domain user account) for each of the SQL Server service and SQL Agent service. What is the reason behind that? (Someone told me to not run the service with an account that has a powerul privilegs! - I don't undrstanmd this point can you explain it please?) What is the diffrent between: 1- Local System account 2 -Network Service account
Whenever i change the SA password to anything else, my SQL Agent services like jobs, alerts starts falling, and if i attempt to stop and then restart sql agent service, it will not start again.
My server uses sql authentication for everything...
Hello, I am having a problem with my sql server agent service not working. I have done some research from earlier postings and still having no luck with getting it up and running.
Problem: "Could not start the SQLSEVERAGENT service on local computer. could be an internal Windows errow or service error.
2006-05-09 14:23:55 - ! [298] SQLServer Error: 6, Specified SQL server not found. [SQLSTATE 08001] 2006-05-09 14:23:55 - ! [298] SQLServer Error: 11001, ConnectionOpen (Connect()). [SQLSTATE 01000] 2006-05-09 14:23:55 - ! [000] Unable to connect to server '\PROBLEMTRACK'; SQLServerAgent cannot start 2006-05-09 14:23:55 - ? [098] SQLServerAgent terminated (normally)
I have reset the password for the 'sa' account and running it and got error: Error 18456: login failed for user Specify a different account to be used. But when I try it only provides me with the 'sa' account.
My primary goal is to run some DMP ion the databases that I have on the server.
We came across a biggg issue with SQL Agent service on one of our SQL Server 2012 production servers. Suddenly, it stopped working. So we just thought of disabling and enabling the service, but once we disabled and try to set automatic (before starting it), gives us error message "The Remote Procedure Call Failed [0x800706be]".
Sql server service has been started with a normal user who do not have any permission but he is a domain user. The sql server service can be started successfuly in cluster
But if the Same user is used to start the SQL Server Agent service there is an error.What is the mininumum priviledge to start a sql server agent in cluster. Why it cannot the domain user cannot be used to start the sql agent service.
I have an SSIS package (TransAgentMaster) that I recently modified to include a call to a child package via the file system. The child package creates a text file. When I run the package in dev studio then the child package/text file is produced.
I then imported the TransAgentMaster as a stored packagesfilesystem package into SQL SSIS and executed the package. The child package produced the text file.
I then ran the SQL Server Agent to see if the child package would work and it did not generate the text file. Thus after updating a SSIS package importing the package into SSIS the job that calls the package will not call the child package. Please not that the TransAgentMaster package calls 7 children packages €¦ just not my new one.
Any thoughts why the agent will not run the child newly crated childe package?
I switched my SQLSERVERAGENT service to use the Local System Account.
I stopped and restarted the service.
Both the Services window and Enterprise Manager show the SQL Server Agent is running. But when I try kicking off a job I get the following message: Error 22022: SQLServerAgent is not currently running so it cannot be notified of this action. Any clues as to why the agent could be running, but not know that it is running?
I'm new to sql server administration so please show some understanding!
I installed sql server 2000 personal edition (upgrated to sp4) on an windows xp pro OS, but I cannot get the sql server agent service running. I get a message " the service did not start due to a logon failure". The server is my local pc, I'm using windows authentication and the login is "domain/loginname" where domain is my pc name.
Does it have to be installed on a server OS to have this service running?
When I run sp_delete_job to delete the whole job inside, will the job steps, job schedules and job servers of that job will be deleted automatically also?
so, do i need to run sp_delete_job only? or need sp_delete_jobstep, sp_delete_jobschedule and sp_delete_jobserver also?
If we were to assign permissions to a backup agent such as Backup Exec to backup the databases on the SQL server, what role would give the least amount but sufficient permissions to perform the backup? I know domain admin would make the agent a local admin and therefore allow it to back up the database but is there a role available to allow backup only?
Please note that I'm referring to a domain account used by Backup Exec to directly backup the databases rather than sql server agent.
I'm thinking of using SQL Server Agent Service for my PDA app. But, I want to use different accounts for SQL Server and SQL Server Agent Service. How can we do this in SQL Server 2005? Do we do this when installing it? Thanks
To run a remainder service that shoots a mail once everyday to all users of a subscribed service satisfying a pre-determined criteria, I'm planning to use a Sql Server Agent Job. In the first step of the Job, I'll have a stored procedure that gets all the records matching that criteria & store it in a table.
In the next step, I want a Windows Service (as there need not be any interface & my project requires easy deployment & minimal manual intervention) to fetch the records from the above table & shoot mails to users of the subscribed service. Several thousand mails may have to be sent this way. In order not to strain Sql server & also avoid extended stored procedures due to safety reasons, I'm considering using this windows service.
I want to know
1) how to make the Sql Server Agent execute the Windows service?
2) abt code or resources on how to write a vb.net Windows service that will work through a recordset & send mails.
3) from a perfomance point of view, if this method is feasible? If not, what are the other efficient alternatives.
I have a Win 2003 server running 3 instances of SQL 2000 Enterprise. The problem is I can only start 2 of the 3 Job Agents at a time, as the 3rd gets stuck in a starting state. Has anyone struck this problem before?
What sort of things would it interact with to justify having this ticked as the sqlserver.exe seems to be consuming all the CPU constantly and some suggestions have said that having this ticked can cause a performance issue.
Am trying to run SQL Server Agent with a service account which is not in the Administrators group. Have done the following - 1. Removed the service account from the Administrators group on the machine 2. Assigned sysadmin privileges to the service account 3. Added it to the SQLServer2005SQLAgentUser$ComputerName$MSSQLSERVER role 4. Through SQL Configuration Manager assigned this account to the SQL Server Agent service However, this does not start the Agent as a service. What is it that is missing?
I am trying to upgrade my default instance from SQL Server 2000 Standard to SQL Server 2005 Standard. I get to the service authentication window and hit my first problem. On that window, there is a dropdown box where you should be able to choose the default instance service AND the SQL Server Agent service and set the authentication information for both of those services. That's the way it is in my test environment. However, on my production server, when I click that dropdown box, only the instance appears, and not the agent. Why?
I tried the upgrade this past weekend and ran into this problem. After researching it and not finding anything out about it online, I decided to just go forward anyway since I had lots of backups and copies of everything. The upgrade appeared to go ok, but the agent service wouldn't start. It gave the usual message when you try to start a service that doesn't have anything to do, like the Performance Logs and Alerts service. When I looked at the agent in SQL Server Management Studio, it had (Agent XPs Disabled) next to it. So I researched this and found how to enable it. For some people online, this allowed them to start the agent service, but of course that didn't work for me. Finally in my research online I found that some individuals had a lot of missing system tables in their MSDB and Master databases after the upgrade due to some of the SQL scripts that are part of the upgrade process not running properly. So I compared my system tables and SP's against another SQL 2005 instance I have, and sure enough, there appears to be a lot of missing tables. It was at that point I decided to just restore everything and try again later.
My point in writing that second paragraph is to ask if that was all due to the fact that the SQL Server Agent wasn't listed in the upgrade wizard like it should have been? Or maybe the fact that it wasn't in that dropdown box was evidence of a deeper problem? Is it likely that if I do whatever it is I need to do to make the agent appear in the upgrade wizard that my missing system tables and agent service not starting problems would probably not be an issue anymore? And why would the agent not be appearing in the upgrade wizard?
I have six stored procedures that I have to run hourly during the day. When I schedule them into separate SQL jobs (they must run concurrently), the percent CPU of Sqlagent90.exe process goes way up, eventually pegging out the overal CPU usage on the server. When I run the six stored procedures concurrently from query windows in SQL Server Management Sudio, the overall CPU usage stays very low.
Is there something about the SQL Server Agent that inherently adds CPU overheard? Thanks in advance ....
hi, can anyone explain, how to execute package with sql server agent:
the problem is::
i made a integration service package (in visual studio 2005). When I 'm in visual studio and do "execute" the package it works well. Then I build a dtsx file. When i double click on it, it opens and if i do execute, it works well.
I want to execute the package every week automatically. I write a job in SQL server agent, in Steps I choosed Type=SQL server Integration service packege, then General->packacge source -> file system, and then wrote where the dtsx file is (which was built). "Run as" I choose my user name.
then i'm trying to start a job, but it fails, and writes "The package execution failed."
For intranet development, our DBAs are asking web developers to use fixed domain NT ID accounts instead of SQL accounts to connect to backend databases in all web applications. We are: Windows XP workstations in a 2003 Active Directory Topology.
I don’t think that this is a good idea (I could say more:) but I have found very little to no information on this subject. So I ask you guys... What are your thoughts? Why or why not?
Hi,I'm in the process of designing a DB (typical management system DB; 2transaction tables and about 5 look-up ones )for one of the departmentsin our company. The user wants this DB and thusly the client (forms,reports..etc.) solely for his department. However, I do expect soonafter deployment that other users want similar DBs and clients,therefore, facing problem of integrating such DBs, if companymanagement wants to implement it as enterprise DB. My question (may beyou could also tell me about other groups specialized in these kind ofissues), how should I create such DB? Should I create extra look-uptable for departments and have each one with its own ID and link it tothe main transaction oneMTIA,Grawsha
I'm still a database newbie so I would like to solicit thoughts aboutthe smartest way to do something in sqlserver.My company has a web application that we customize for each client.We can do this because everything is database driven. We havedatabase tables that contain our HTML and database tables as well assome standard tables for each database. We have an in house app thatlets us tweak both of these things and creates a new web site anddatabase tailored to each project.Each of these sites has a table that stores a schedule are clientsuse.The records in this schedule table change when information in othercustom generated tables change.My company currently uses a legacy foxpro app to update the scheduletable.The foxpro app contacts sqlserver, reads a table with a list of tablesand scheduling information to check, checks each of those items andupdates the schedule table.I would like to lose the foxpro app.At first thought.........as a database newbie.......putting triggersin each of the tables to update the schedule when something changesseems the way to go.However, since we change a part of the schema ( we have an app thatgenerates the database tables unique to each client ) for each clientI would like a scheme that would not involve having to create adifferent trigger for each new table.I would also like something that updates in real time. Right now thefoxpro app is executed once a day.I was thinking of making a large stored procedure and putting anidentical call to that procedure in each table.Each table would have the same trigger in it that would get fired whenthe record was altered. It would call the stored procedure withrelevent arguments to update the schedule.Does this sound like a smart way to solve this problem or am I notthinking "database enough"?Any thoughts are welcome.I would like to build a better solutionSteve
I noticed that the current SLQCe driver does not offer support for the APM(Asynchronous Programming Model). Are there any plans to do this in the future? In light of the lack of APM functionality doe anyone have any ideas or thoughts on how async operations could be done, or if they are even needed in the context of applications that use SQL Ce
I was hoping to elicit some feedback on a trend I am seeing in the Portal market, and specifically with SharePoint development.
If you are not familiar with SharePoint, there is a data table abstraction within SharePoint called a "List". Lists are used for storing data (duh!). However, they are built using the SharePoint front end, and the data entered into all lists is stored in a few tables in the SharePoint content database.
What I am seeing happening is SharePoint gurus reccomending AGAINST storing your relational data within database tables, and within SharePoint lists instead. I am not sold on this approach, and it actually makes me think we are taking a step backwards with regards to persistent data storage and best practices.
- Lists cannot be natively related to one another, however they support "lookups" - Anyone can create a list...and repeat the same data all over the enterprise. - Lists are maintained in two tables within the SharePoint content database using meta-data patterns. - Portals contain a multitude of sites. Users and portal admins can create lists all over the place, thus spreading related data over a wide swath of the enterprise.
Is it just me, or are SharePoint pundits absolutely CRAZY to be recommending persistent data storage using lists? I see nothing but problems arising from this approach.
I apologize beforehand if you have not worked with SharePoint and Lists, as this post may not make much sense to you. ;)
Thought I should post in the newbie forum for a while, instead. :-)
I have a couple of scripts that I've generated that drop a couple of system stored procedures and recreate them. I'm not sure why I did it in the first place, but I think it was that it wouldn't let me run an ALTER statement on them. Specifically, I'm now looking at sp_add_operator. I changed it to a 500 character email field instead of whatever it was (100, I think.)
/* Explanation: Why did I do that? SQL Mail is prohibited here, so I'm using CDO_Sysmail to email myself and the developers if a job fails. The list of people to email is determined by the owner of the database, who is also an operator in SQL. I get the list of emails from the email field of the operator properties. Hence, I need a bigger email field. Yes, I now know it would most likely be better to create an ADMIN database on each server for this kind of stuff. (Thanks to Tara for that blogged suggestion.) */
While I will probably go back to the default stored procedure, this got me to thinking: when would it be better to use an ALTER statement on a SProc rather than to do a DROP and CREATE?
Folks, I have a quick question that I would very much appreciate somefeedback on. We are a not for profit charity organization that has decidedto develop a software in-house to manage our volunteers. We have SQL andthat makes the most sense from a database solution but we have some issuessurrounding the choice of the development language. Some have suggested100% java while others say Visual Basic. The head of our team has suggestedwe do it in Cold Fusion since this will be an internet based application andI guess I would very much like some feedback on that choice. We have about 5organizations that we will tie into this system with about 5000 userslogging in once per month.Any suggestions or comments would be greatly appreciated.CheersWade
I have a database with a dozen or so tables. No table constraints. Logic is all in stored procedures.
I have several Excel spreadsheets of data to import into the database, one speadsheet to a table. Each spreadsheet has additional data(columns) that each table has no interest in and should be ignored.
I would appreciate your thoughts on methods and best practices for loading this data to the database.
I am about to investigate SQL Server 2005 Express handling of XML. I am familiar with XML and XSL conversions and it seems to me that XSL conversion of Excel data to XML gives me a lot of flexibility prior to database import for shaping the data.
In short, importing data to the database from an XML source.
I am not famliar with SQL Server's XML capability and would appreciate thoughts on this while I look into it.
And of course alternate ways that I am overlooking.
I like the new gig a lot. Real busy, smart folks and I have been in high demand since 5 minutes after my butt hit the chair. I already have code in production.
Anyhow, we have a security situation on the sql servers I pointed out on my first day. So they want me to roll everything over to Windows Authentication and give the developers and report writers more restricted rights inside SQL Server. So they have NT Groups for different kinds of users and all of that jazz and I layed on the typical stuff about using NT groups vs individual accounts and ease of admin vs granularity of control. Well the boss came back and said he wants ease of admin and granularity of control over security. So, does anyone have any fresh thinking on turning my eitheror into an AND.
