Home Dashboard Directory Help
Search

Hung connection wait type PREEMPTIVE_OS_PIPEOPS by john_s50


Status: 

Closed
 as External Help for as External


6
0
Sign in
to vote
Type: Bug
ID: 556545
Opened: 5/3/2010 7:26:21 AM
Access Restriction: Public
1
Workaround(s)
view
5
User(s) can reproduce this bug

Description

A user ran a stored procedure (I'm not sure what it does).
It wasn't working, so he killed the connection.
Now his connection is hung on the server with a command type of killed / rollback and a wait type of preemptive_os_pipeops.
I can't get the process off the system.
Details
Sign in to post a comment.
Posted by jakkampudi on 1/21/2013 at 6:41 AM
We have same problem when am exporting the some table using BCP command with XP_CMDSHELL.. in a SP .. but after use NOLOCK in my query I can able to execute with out any issues , I mean I don't see PREEMPTIVE_OS_* wait type :)
Posted by Westergaard on 6/13/2012 at 3:19 AM
I have seen this problem also, any news about this issus
Posted by mstreetman on 10/21/2010 at 5:42 AM
ive seen this a few times, all with cmd_shell involved. if you can find the os process and kill it using task manager the spid will go away. in my cases its been with bcp.exe and ftp.exe i think not returning inside a cmdshell process.
Posted by Microsoft on 7/7/2010 at 3:44 PM
Clarification to my previous post -- the entire SQL instance doesn't hang but rather the failed operation.

We've looked a bit more into this and there's not much more that we can do from the SQL side. The underlying issue is outside of SQL and you may required help from the Windows side to help identify and resolve the problem. If contacting CSS is an option available to you, then you may want to get their assistance.

Il-Sung.
Posted by Microsoft on 7/4/2010 at 4:42 PM
Hi,

Based on the problem description, it appears that xp_cmdshell is being used to spin up a cmd shell and the called process hung or for whatever reason. The lead to SQL hanging trying to read the processes pipes which is the expected behaviour. If you're able to generate a dump file, we can confirm this.

Also, if this is an issue affecting a production server, then I highly recommend that you contant the MS Customer Service and Support for assistance as MS Connect is not the appropriate channel.

Thanks,
Il-Sung.
Posted by Don Vilen on 6/2/2010 at 1:11 PM
A similar hang has been experienced by my customer, 3 times now, one of them on a production system. The instance had to be restarted or, when on a cluster, failed over to another node. Investigation is on-going, but no strong similarities in the connections' operations have been identified.
Sign in to post a workaround.
Posted by jakkampudi on 1/21/2013 at 6:44 AM
Use Hints like (NOLOCK , READPAST ) in the query that you are using in BCP . it might help . for me it worked