I'm working on a script that creates a MySQL dump via <cfexecute> and then FTPs the SQL script to another server. I've resorted to checking once per second to see if the filesize has changed, and if it has not changed within the past five seconds I assume it has completed.
This is fine for the current application, but eventually I would like to be able to import the SQL script on the second server and provide some sort of notification that it has completed.
Is there some way to track the status of a running process?
If not, is there a way to accomplish a full DB export and import via ColdFusion alone?
Actually you may not realize it, but when you call <cfexecute> without passing a timeout attribute it defaults to '0' timeout. And if you read the docs on <cfexecute> you'd see:
If the value is 0:
ColdFusion starts a process and returns immediately. ColdFusion may
return control to the calling page
before any program output displays. To
ensure that program output displays,
set the value to 2 or higher.
So I would suggest passing a higher value for timeout which will cause ColdFusion to wait for mysqldump to complete before moving on.
Reference
Check out Event Gateways[1] for one way to deal with asynchronous operations. There's a Directory Watcher gateway that comes with CF as an example.[2]
Barring that, create some sort of batch processing facility using CF Scheduled Tasks. Add the job to a database table and have a scheduled task periodically pull jobs out of the table and execute them, reporting on the result. A second scheduled task can detect that the first completed and carry out the next step of the process.
[1] http://help.adobe.com/en_US/ColdFusion/9.0/CFMLRef/WSc3ff6d0ea77859461172e0811cbec214e3-7fa7.html
[2] http://help.adobe.com/en_US/ColdFusion/9.0/Developing/WSc3ff6d0ea77859461172e0811cbec22c24-77f7.html
Related
I want to create a CFschedule task to run every 30 seconds and was wondering what would happen if a Task tries to fire up while the previous task is still running - does it wait or die or ???
cfschedule doesnt care much about the content of your jobs, think of it like cron daemon on Linux. Its just the scheduling agent.
If you need to ensure only one job of a certain type is running its up to you to implement some kind of busy/locking behavior, probably using some kind of a datastore such as a relational database, plain files (ala the presence of a file indicates in-process, then delete it right before job completion, this would be a good use of the onComplete handler), redis, etc.
Coldfusion 2016
Microsoft Server 2012
Oracle 12
ODBC connection
I turned on profiling and monitoring and now I can see that there are requests that are stuck and cannot be terminated by the CF monitor; Some are over 200k seconds.
I know I can increase the number of simultaneous requests but I want to solve the underlying problem. As I read the stack traces of these “zombie requests” they are getting stuck on and some are in but some are not. I ran the query in my oracle client and they resolve instantly.
Is there a way to terminate these requests or prevent this from happening at all?
EDIT: The server monitor does not treat these requests as slow or hung, the alerts are not triggering for any of these. Honestly, they should have be going off constantly considering how many of these there are.
Also, the execution time is a mere .003 seconds so what happened? Why doesn't ColdFusion know this?
An example of a "zombie"
The active query that is stuck
We have a similar situation with a different database engine - redbrick, which runs on a unix server. We solved it as follows.
We set up a cron job on the database server to run every 5 minutes. This job uses a combination of unix and awk commands.
This job runs a query against the system table that looks for queries that have been running for more than 120 seconds, where the database account is the one used by ColdFusion. Records are outputted to a file. Something like this:
print "alter system cancel user command userName process " $1 ";"
$1 comes from the query and is the process Id we want to stop.
Then we run the file, which executes all those alter system commands.
With a different database engine, and possible different OS for the database server, the details would be different, but the approach should work.
Edit Starts Here
To prevent recurrence, look at the pages that call the ones with the long running queries. If impatient users are able to repeatedly click something because nothing is happening, do something about that. You can use javascript to make the link/button go away. Alternatively, you can go to an intermediate page with a display for the user and something that carries them through to the real page.
I have simple bpl process with 3 invokes in loop. One of instances hang without any visible reason. So process is in active state but it is not executing any longer. Last logged activity is call to invoke. I search database and find out that both request and response are present in table ode_message and they looks correctly. But output variable from invoke in table ode_xml_data is not filled. There is no logs in bps from time when message arrived. Is any way to find out what happen wrong?
I'm try to use Wso2 BPS 2.1.2
I use Django-Celery +rabbitmq to execute some asyn tasks,I define a queue 'sendmail' to execute send email task,send mail is triggered by a specific task(this task has own queue), but now I encounter a problem,after the specific task finish, the mail sometimes send at once, sometimes need 5-20minutes.I want to know what reason caused it.
Django-celery will package the taskname and param as message to rabbitmq when call task.delay().
I want to know when the message go to the rabbitmq, but use web management tool only can see total messages,can't see the every message's detail, especially the time the message reached. Django-celery log can only see the work got from broker time and execute task time.I want to know all related timepoint to sure which step the time main consumed.
Django-Celery does (I believe) report task data on a per-task basis. When you sync your database, it crates a bunch of monitoring tables which are accessible via the admin. However, in order for these tasks to be recorded in these tables, you need to run the celerycam program in the django context (python ./manage.py celerycam). The celerycam program will take "snapshots" of your tasks every second or so (by default) and record information about them. Another useful tool for monitoring is the celerymon program (which also has to run in the django context). This is a command line ncurses program that reports real-time information about tasks as they occur. Finally, rabbitmqctrl has a bunch of options that might help with monitoring.
This is a particularly useful page in the docs:
http://celery.github.com/celery/userguide/monitoring.html
Anyway, this is what I use to monitor my tasks when using celery.
I have a custom django-command that reads and RSS, looks for new feeds and, if any new feed is found, I push it (pusher.com) to my webapp hosted in Heroku (heroku.com). This checking needs to be done as much as possible to be able to get the new feeds as soon as possible, let's say, every second.
The two issues I have are:
As this app will only be used by a few people(2-3), the command must be run only if any of these people are inside the app so I don't overload server jobs.
Once the user left the app (may be they just closed it, or they have certain time of inactivity, i.e. not clicking anything), the command must stop checking RSS.
My questions are,
where should I run the command from? directly from a view, from a signal?
How could I interrupt such command once the user leaves the app?
Thanks in advance for any help :)
You could use request-finished signal. In signal handler you could run celery task, so user hasn't wait the rss server request end