#opendaylight-meeting: clustering hackers

Meeting summary

1. agenda bashing (colindixon, 15:02:16)
   1. TomP asks for his singleton cluster patches to be reviewed (colindixon, 15:14:56)
   2. https://lists.opendaylight.org/pipermail/controller-dev/2016-July/012301.html colindixon brings up somebody asking about cross-node notifications so people can fact-check what he's said (colindixon, 15:16:04)
   
   
2. TomP asks about rovarga's patches (colindixon, 15:16:14)
   1. rovarga says he hopes to have his patches done this week (at worst by the end of the month) (colindixon, 15:17:20)
   2. https://bugs.opendaylight.org/show_bug.cgi?id=5280 they pertain to this bug (colindixon, 15:17:30)
   3. rovarga says that, in his opinion, we should delay the release if this isn't fixed (colindixon, 15:17:45)
   4. rovarga says that this causes the system to become unusable under load (colindixon, 15:19:54)
   5. rovarga the plan is to make transactions idempotent and thus allow for more easily replayed transactions when they fail (colindixon, 15:20:25)
   6. TomP agrees this is important work, but wonders how much time we have to test this new code at this point given that RC0 is coming (colindixon, 15:21:47)
   7. TomP asks Muthu can do some testing, Muthu says sure, when we give him the sign to go for it, they can run some tests for HA and stuff (colindixon, 15:24:37)
   8. TomP, rovarga, and colindixon discuss how we can deliver this and when the deadlines are, colindixon notes that RC0 is 8/11 and it would be *really*, really good if we could have the full, mildly tests implementation in by then if we want it (colindixon, 15:33:20)
   9. then it needs to be tested heavily and then the intent is to release the build that happens on 9/1/2016 unless there's blocking bugs in it (colindixon, 15:33:57)
   
   
3. infrastructure issues (colindixon, 15:34:03)
   1. jenkins seems to be better behaved right now, on the other hand nexus timeouts are still happening (colindixon, 15:34:40)
   2. rovarga says that he's seen failure to deploy and failure to clone at least a few times this week (colindixon, 15:35:00)
   3. rovarga and TomP say that infrastructure has cost us a lot of time this release, there's significant time lost to having to do rechecks and investigating why failures are happening (colindixon, 15:39:04)

Action items

1. (none)

People present (lines said)

1. colindixon (24)
2. odl_meetbot (5)