Comment 355 for bug 417757

Revision history for this message
In , drepper (drepper-redhat-bugs) wrote :

Opening two sockets is slow and eats up precious resources (yes, I consider 64k sockets a precious resource). I already said I'll implement this as a fallback solution in case everything else fails and the fallback mode will be automatically reached (after various timeouts). That'll likely happen next week. But it won't be the default. If workarounds to broken behavior is the default the underlying bugs will never be fixed. If people are inconvenienced because of the bugs but it still works they hopefully continue to harass the responsible parties.

The good news is that at least some of the vendors are responsive. Cisco apparently already fixed some of their code. Let's hope this continues to be the case for others as well.