Closed arjan closed 9 months ago
@arjan thanks for creating this pr, we're running into a similar issue.
@derekkraan is there something I can do to help get this pr to a point where you feel comfortable merging it?
@kinson I need to find time to look at this PR, nothing you can do at the moment to speed things up.
Fwiw I have been running this in production for quite some time now without noticable issues...
Thanks @arjan for this and for the other PR. As usual, apologies for taking so long to get around to merging them. I suppose a 0.9 release is in order?
Fwiw I have been running this in production for quite some time now without noticable issues...
Not sure, technically these are bugfixes i guess?
These are, but there are breaking changes in master
that necessitate a 0.9.
As usual, apologies for taking so long to get around to merging them.
@derekkraan no problem, thanks for merging and releasing this 🚀
I noticed that when a cluster is restarting, Horde tends to crash because it does some administration that is incorrect. The child lookup for the linked ETS table can fail but this was not handled on all places in the Dynamic Supervisor implementation.
This fix addresses three places that resulted in crashes because nil was not handled from the get_item / pop_item calls.
Note that there were several other places where this was already addressed.