I believe this issue was raised before, but we noticed that a publish may freeze or timeout on occasion. The only similarity we found when encountering these issues is that a user was logged in and had a page open in edit mode while the publish was running. Has anyone else had this issue? Is there a fix for it?
When you say a publish will freeze or timeout, what exactly happens to the publish operation in question? In the reports tab in the Publish section of CM1, does the publish’s status get marked as Failed or Complete with Failures? If so, can you click on the View Details link to see if any specific error message was outputted? Thanks!
Thank you for the quick reply.
While a publish usually takes 45-60 mins, when this issue occurs we will let the publish go for up to three hours and the progress will never leave 0%. When we check the publish details, it will show some of the files were publish, but the list never updated over the 2-3 additional hours. Also, if we check the file system, we see that some of the files were published, but not all. We never receive an error or failure message.
I hope this helps.
Thanks Chris. To further clarify, when these unusually long publish operations are encountered, do you let them run to competition, or do you hit the Stop button under the Status of Current Jobs after a certain time? If you do let the publishes run to completion, do they ever get marked as Completed in the publishing log, or is the Status field left blank?
The publish operations really shouldn’t be affected by users working in your system. A more likely issue might be network interference of some type. For reference, CM1 writes to your production environment using standard FTP (port 21, or 22 for SFTP) and over port 8443 for your DTS server. Typically, after these unusually long publish operations, can you simply re-run the publish operation successfully immediately afterwards? Or is there usually a period of time before a successful second attempt?
In the interest of time, we have never let the publish go beyond three hours.
When we cancel a stalled job, I again ask all employees to save their work, and I am able to rerun the publish successfully.
Perhaps there is something else causing the issue. That was the only connection I was able to make. Is there anything else that could cause the the publish to take a lot longer/not finish?
Because the publish will error out pretty fast if there is an FTP error, the next logical place to look would be the connection being made to your DTS Tomcat server. If a connection cannot be made to the DTS Tomcat server, the connection will time-out for every page with meta-data information being written to your DTS server. This process can take many hours before it gets to the point where it uses FTP to transfer the files and complete the operation.
However, it’s odd that simply restarting the publish operation resolves the issue. If it is an issue writing to the DTS server, I would expect some action like a restart of the DTS server to be required. At any rate, next time you encounter this issue, I would be very curious if you could make a copy and send me your server.log file located at \AppServer\server\rx\log so I can get a better idea of what’s happening here.
Thank you, Nathaniel. I will be sure to send you the log the next time it happens.