The connection is locked and the client needs to be restarted

Hello everybody.
I have problems with one of the scadas that are in the factory, in the gateway there are 3 projects, luckily it only happens in one of them.

When the error occurs, the scada hangs or disconnects, having to close the client and start it without the need to restart the PC.

Error.txt (35.6 KB)

It is also not specified when it started to happen since it was working for a long time without this problem.
Regards.

Looks like something is hitting the project list gateway API and that is tying up the internal database. What version of Ignition is this?

The version is 7.8.5

Thanks for the reply.

I suspect this is a bug, as that API endpoint should be dropping the internal connection before streaming to the consumer.

Unfortunately, that version is end of life.

Which means you probably need to find out what in your deployment is hitting that endpoint and consuming it too slowly (or dropping the connection before complete). And stopping it.

Or upgrading.

In case we decide to update the ignition.
Do you think it can be solved?
And what about the functions that change like system.tag.write?

I have legacy code in a 8.0.xx version that still uses system.tag.write. They still exist but aren’t documented in those versions as they want you to move to system.tag.writeBlocking - Ignition User Manual 8.0 - Ignition Documentation as I think it’s more performative. However system.tag.write will still work - I don’t believe any of the underlying mechanism will change.

Of course, with upgrading, you should definitely test it on a test server first if possible.

Your best bet might be to call Ignition at this point though. This sort of error I think will require some knowledge of the code base.

This. Use the two-hour trial mode as much as you need to prove that the upgrade works.

Consider upgrading to 7.9.17 first, as that will be a smaller change. After getting that working (should be easy), test upgrade from 7.9.17 to 8.1.7. (The vast majority of user testing of upgrades was from various 7.9.x to various 8.y.z. You want to use the best-tested upgrade path.)

1 Like

This is good news for me, one of the projects is big enough to migrate all the functions.

Unfortunately the testing time would not be enough with 2 hours, the error is not periodic, it can happen once a day or not.

We will analyze the possibility of migrating to 7.9.

Thank you both very much.
If you have any other ideas, I am all ears. :grinning:

You can reset the two hours as many times as you need fwiw.

Sure, I meant to leave it in continuous.