Hooks uninstall themselves on Tomcat restart

[Lee Chamberlain, modificado hace 7 años.]

Whenever I restart Tomcat a couple of hooks I've added to Liferay uninstall themselves.

This hasn't been a problem, as I've just been able to recopy the hooks into the webapps folder manually and everything was fine.

It's now getting increasingly difficult to do this. I've tried copying them back across 5 times this morning, and they've not been picked up by Liferay.

Has anyone else had this problem and know a way around it?

Thanks
L

[Olaf Kock, modificado hace 7 años.]

Lee Chamberlain:

Whenever I restart Tomcat a couple of hooks I've added to Liferay uninstall themselves.

What's the method you use to stop/start Tomcat? This shouldn't happen: While the logs might state that hooks are "uninstalled" from Liferay (not from tomcat) on shutdown, they should be reinstalled on startup. If they physically disappear from the webapps folder (assuming you're using 6.2), something else is wrong - maybe some system policy to always start with a known state, e.g. to fight erroneously installed webapps, assuming that they've been a security threat?

If you're on Linux, you might want to try to add a harden/soften option to the daemon's startup script like I've documented here (Ctrl-F "harden") so that tomcat can't write to the webapps folder at all, unless you explicitly "soften" the permissions (e.g. to install/upgrade a hook)

[Lee Chamberlain, modificado hace 7 años.]

We use Apache Tomcat server on prod and Eclipse on dev. It happens both ways (though obviously not as big a problem on dev).

The hook folders stay in webapps folder, but for some reason they need to be redeployed either manually (in prod) or via eclipse (dev).

[Matthieu Levesque, modificado hace 6 años.]

I know it's an old thread, but we're having the same issue with 6.2. The only hook that has that symptom is one that overwrites a delivered jsp.

The hook is present in the webapp folder but the jsp that render the output is the vanilla one from Liferay 6.2. A simple redeploy fix this but it's not the best option.

I'll try to post anything I find.

[Olaf Kock, modificado hace 6 años.]

I lost track of this post back then - if you find steps to reproduce, they'd be interesting. I've never observed this behavior.