Repository navigation
[DP-100] Implement monitoring for registry values #212
Description
[Imported from JIRA. Reported by Alexander Vershilov @qnikst) as DP-100 on 2015-01-30 12:04:08]
Currently it's possible to store remote processes in registry, but such processes are not guaranteed to be monitored. So when remote process dies current node will never receive an event about process death so will never free value from registry and this could potentialy lead to memory leaks.
We need to elaborate a good solution for this issue.
Activity
[-]Implement monitoring for registry values[/-][+][DP-100] Implement monitoring for registry values[/+]qnikst commentedon Jun 19, 2015
Addressed in PR #184.
hyperthunk commentedon Oct 23, 2026
@LaurentRDC how does the removal of the remote table resolve this PR? I'm slowly reintroducing myself and not actively contributing at the moment, so I'm mostly just curious.
LaurentRDC commentedon Oct 23, 2026
My understanding of the issue is that if the remote table is gone, the underlying issue of this ticket (storing a remote process in the remote table leaks) doesn't exist anymore.
Maybe I misunderstood and this issue is worth solving separately?
hyperthunk commentedon Oct 26, 2026
I think there issue here was less the leaking process handle (thought that is clearly bad) and more than the notification wasn't arriving. I don't think the bug described here has been confirmed to be directly related to the use of the remote table, but it's quite possible I'm missing some nuance here about the changes your static pointers patch brings with it?
hyperthunk commentedon Oct 26, 2026
Hmmm, all of the linked / related issues are merged resolved. I wonder if this is a dangling record, and actually no fix is needed?
I'll try and get a build environment set up over the weekend and see if I can raise a failing test case, otherwise this could be a non-issue!
hyperthunk commentedon Oct 26, 2026
Okay, I am pretty certain the original scenario was fixed in #322, but we have an already-dead case remaining here. Personally I think this should be closed, but I'm not a maintainer these days so I leave that to your judgement @LaurentRDC. I will submit a PR for the remaining bug shortly.
Minimal fix to handle registration of names against dead pids
LaurentRDC commentedon Oct 27, 2026
Thanks @hyperthunk ! I'll close this for now
Minimal fix to handle registration of names against dead pids (#499)