Forums

Looped consoles freezing near ~ 2am GMT

Math fail. Make that more like 10m - Noon GMT... :) ... was early and I had not had my coffee yet.

Since the last major update, all of my consoles have stopped/frozen around 2am GMT. I don't have an exact time on it, seems to somewhat vary. Nothing in my code has been changed and prior to the major update with some downtime, I was able to go weeks without a single script stopping.

I am aware this is not best practice/suggested here to run looped consoles, so do with the report as you please. Just seems weird that this is happening so frequently, basically on schedule, all of a sudden.

To get my scripts back running, I can either do 1 of 2 things:

  1. Simply go to Dashboard, consoles, and literally just click on the running console. This will get it to start running again.

  2. Restart the script.

The ability to just go to dashboard and click on the already running script to get it running again is the weird part. I reported this issue a long time ago when I first starting running here. I don't know that it was ever figured out directly, but it wound up being resolved for many months until now again.

I have two accounts here, it's happening on both... but they are both running looped consoles. I am entering tarpit, though it seems that the freezing occurs after that point. Also, tarpit still moves, just slowly. This is 100% frozen until I do 1 of those 2 things above.

We have an idea about why the daily reset is happening, it's to do with a housekeeping script, we've made some tweaks to it to try and stop it happening, so watch this space on that front.

Re: the fact that clicking on the consoles to restart them works, that's a consequence of the way we start consoles in "normal" circumstances, we store information about what kind of console it is when you first ask for it, and that lets us re-start it if the server it was running on ever restarts for whatever reason. So, consoles you start from the "save & run" button will always restart the same script after a reboot...

Harry,

Thanks for looking into this and issuing such a quick fix, and thanks for the information about why clicking on it from the running consoles is comparable to save and run when a server had restarted.

It looks like whatever you thought it was, was correct so far! I'll update if I notice any more trouble.

One thing that might have been making this worse -- when we released our most recent version last week, we started up the cluster configured so that any kind of console could go to any server in the cluster. Normally we set things up so that consoles for people who are logged in to PythonAnywhere go to one set of servers, and people who are using consoles while they're not logged in (for example, from the interactive shell on Python.org) go to a different set.

The second configuration is obviously preferable, as the traffic to the open-access sites can be unpredictable, so we've switched back to that now.