ColdFusion MX Mail Spooler Behavior Might Cause SpoolLockTimeoutException

Viewing By Entry

ColdFusion MX Mail Spooler Behavior Might Cause SpoolLockTimeoutException

The mail spooler for ColdFusion MX 6.x/7 locks the mail spool directory when writing mail to disk. If a mail thread is trying to write a mail file to disk and is waiting to obtain the lock, it will wait up to 30 seconds and then throw a coldfusion.mail.MailSpooler$SpoolLockTimeoutException or coldfusion.mail.MailSessionException. I'm still exploring what might require the lock for so long and prevent other mail from being spooled to disk, but so far it seems well correlated with having message bodies of 12-15 MB in one case. The 30 second wait for the lock is hard coded and immutable, but writing a large message to disk shouldn't take 30 seconds in most situations. I'll update this later if I find more.

StackTrace coldfusion.mail.MailSessionException: An exception occurred when setting up mail server parameters
at coldfusion.mail.MailSpooler.storeMail(MailSpooler.java:670)
at coldfusion.tagext.net.MailTag.doAfterBody(MailTag.java:618)

"Error","scheduler-1","06/08/05","09:03:24",,"A problem occurred when attempting to deliver mail. This exception was caused by: coldfusion.mail.MailSpooler$SpoolLockTimeoutException: A timeout occurred while waiting for the lock on the mail spool directory.."

posted on 10 June, 2005 at 4:33 PM.
ColdFusion | Comments (18)

Comments

Steven,

I've seen this error occur recently for me as well. In my case, it appears that the error was being generated by an e-mail that was trying to attach a file that no longer existed.

Are you trying to attach a large document to your cfmail files? Maybe the the attachments aren't being created before being sent...

-Dan

comment by Dan G. Switzer, II, posted at 6/10/05 4:28 PM

Thanks for the tip. I'll check out that angle.

comment by Steven Erat, posted at 6/10/05 4:51 PM

Hi Steven,

I am seeing this error to. In our case we are generating an html email report to a customer and it is over 12 MB in size.

Once the email spool and undelivr folder is cleared out, the lock on the Mail Spool still exists and is causing the error to appear in the logs every two minutes.

Any idea on how to release this lock w/o restarting ColdFusion?

comment by Jon Jasperson, posted at 7/1/05 10:58 AM

Steven,

Any update on this problem? This is becoming a regular headache for us.

-Dan

comment by Dan G. Switzer, II, posted at 8/31/05 12:36 PM

we recived this same error when sending mass emails (6k). It seems that if a few of the emails are bad then coldfusion will hang and the mail spool locks up. The only way to fix this that I know of is by restarting coldfusion.

I will try a simple "try and catch" to see if that willl help prevent the problem.
Any other thoughts?

comment by Joe, posted at 2/18/06 4:40 PM

Im suddenly getting this error on a mail server that is an exact copy version-wise of a server that works without a problem. If anyone has a fix for this, I would really appreciate it.

comment by Sean Walker, posted at 10/10/06 10:37 AM

I'm getting this error when sending large dumps trapped by a cferror. The worse part about it, is that CF never releases the lock and no further email goes out until I restart the service.

comment by Rob Gonda, posted at 11/14/06 8:58 AM

Rob,

while it is still not recommended to use the ServiceFactory (unless done carefully), perhaps this might be one exception to the rule where ServiceFactory methods might be helpful.

The ColdFusion MX 7 AdminAPI offers a support mechanism to programmatically adjust the Mail Service, but it does not expose the restart() method. You can get to the MailSpoolService.restart() method via the ServiceFactory.

What do you think of this code and does it have an effect on the problem:

<cflock type="EXCLUSIVE" name="MailSpoolServiceLock" timeout="10" throwontimeout="No">
<cfobject action="CREATE"
   type="JAVA"
   class="coldfusion.server.ServiceFactory"
   name="factory">

<cfset MailSpoolService = factory.getMailSpoolService()>
<cfset MailSpoolService.restart()>
</cflock>

comment by Steven Erat, posted at 11/14/06 9:49 AM

I got that error without having an attachment to the email. I guess something before this caused the problem. Can the a non-email issue cause this problem? That for the solution.

cfmx 7.0.2 enterprise

comment by Patrick Whittingham, posted at 1/19/07 6:35 AM

I get this problem the same way that Rob Gonda does. cfdumps in an email from cferror.
I think its usually when a search engine hits and hits and hits on an error and basically floods the spooler.

I tried the MailSpoolerService code - I even added a quick link to the hotfix extension do run the restart from within cfadmin, but, unfortunately, it looks like the service is well and truely dead, so the restart() doesn't fail, but doesn't work either. Even a stop()/start() doesn't do anything.

comment by Stephen Moretti, posted at 2/12/07 6:48 AM

I think I also got this problem the same way that Rob Gonda does. cfdumps in an email from cferror. I also tried the MailSpoolerService Code Like Seephen Moretti..with the same result..so I changed the Mail Spool Settings from disk to memory, increased Spool Interval (seconds) and Mail Delivery Threads. I restarted the CF server.. and all the mail is functioning correctly again ..changed the Mail Spool Settings back to disk.. so far so good. I will be monitoring the mail more carefully and see if this happens again.

comment by Mark Holm, posted at 2/27/07 1:08 PM

Steven,

Was there any permanent solution to this issue. We are now encountering it with CF 7.02 (latest).

Sami

comment by Sami Hoda, posted at 6/24/07 11:11 PM

ColdFusion 8 has increased the timeout from 30 to 60 seconds. In addition we added a System property to control the timeout: coldfusion.spooltimeout. To use it, add the following to the JVM arguments (runtime/bin/jvm.config for standalone): -Dcoldfusion.spooltimeout=120 to increase to a 120 seconds timeout on the spool directory.

This lock is an internal lock and generally should only be locked for milliseconds. There is no actual locking on the directory or spool files. Switching to "spool to memory" will avoid this lock entirely (and increase performance). This would be a very good option if you are mailing 10-12mb worth of data.

comment by Tom Jordahl, posted at 7/10/07 4:50 PM

There is a hotfix for CF 7.0.2 available here: http://kb.adobe.com/selfservice/viewContent.do?ext...

comment by Duncan Loxton, posted at 8/20/07 7:55 PM

This also happens when you run out of disk space where the mailspool is at.

comment by jonese, posted at 11/9/07 4:19 PM

A new version of the MailSpooler technote should publish in the next few days. It is "Controlling the Mail Spool Timeout - Hot Fix for ColdFusion MX 7.0.2" - http://www.adobe.com/go/kb402001. It will clearly indicate that is was updated for April 2008 and the new hot fix is hf702-71489. It makes several updates to the MailSpool locking and should be installed by all mail users with 7.02.

comment by Ken Smith, posted at 4/7/08 12:23 PM

Hi Steve,

I've had this happen in ColdFusion 8 too.

- Cass

comment by Cassius Seeley, posted at 3/18/09 10:33 PM

Hi Cass,

Yes, I suppose its possible to see this issue in CF8. The default timeout for the spool lock was upped from 30 to 60 seconds in CF8, but you may need more. See TomJ's comment above about using the custom argument to set it higher with: -Dcoldfusion.spooltimeout

comment by Steven Erat, posted at 3/18/09 10:38 PM

Calendar

Search This Site

This is an exact search only

About This Site

Adobe Alumni & Community Professional. Expert in ColdFusion, Flex, LCDS, Photoshop, Lightroom. Linux RHCE. Follow Me!. For my photography check out Boston Portrait Photographer.