#opendaylight-clustering: clustering_hackers

Meeting summary

1. agenda bashing and action items (tbachman, 16:01:08)
   1. https://meetings.opendaylight.org/opendaylight-clustering/2014/clustering_hackers/opendaylight-clustering-clustering_hackers.2014-11-25-16.02.html Minutes from last week’s meeting (tbachman, 16:01:14)
   2. topic: bug-review (tbachman, 16:06:06)
   3. topic: two-node clustering (tbachman, 16:06:11)
   4. topic: open issues (tbachman, 16:06:15)
   
   
2. bug review (tbachman, 16:06:22)
   1. The unit tests that were failing in jenkins have been addressed (tbachman, 16:06:43)
   2. There was one issue brought up on the mailing list where the controller runs out of memory when connected to mininet. This bug should be fixed (tbachman, 16:07:12)
   3. moizr says the only bug out for review is a fix for a test, which as been merged to master but not yet stable/helium (tbachman, 16:08:13)
   4. There is a bug where ODL startup occasionally fails — this may be related to jolokia, so we may not be able to worry about it (tbachman, 16:08:39)
   5. The bug where there’s a log exception hasn’t been reproduced (tbachman, 16:08:50)
   6. moizr says there’s a bug where a remote RPC isn’t being invoked. This needs to be investigated (bug 2449) (tbachman, 16:09:18)
   7. ACTION: harman to look at bug 2449 (tbachman, 16:09:37)
   8. The stable/helium release date is soon (1/12?), so there’s not much time to address additional things (tbachman, 16:10:42)
   9. moizr is going to move all the performance improvements to lithium release (i.e. not stable/helium) (tbachman, 16:11:12)
   10. pantelis asks if these are integration, and are available to all (tbachman, 16:12:11)
   11. jmedved says that everything he runs has been committed (tbachman, 16:12:20)
   
   
3. two-node clustering (tbachman, 16:12:30)
   1. mark says a new 2-node strategy was committed last week (tbachman, 16:12:47)
   2. half of the commit is comments explaining rationale, and a test case file laying out all the situations that will likely need testing (tbachman, 16:13:06)
   3. mark is working on a wiki page for 2-node based on the code, and will hold a meeting this thursday for those interested in the two-node design; interested parties should contact mark directly (tbachman, 16:13:48)
   
   
4. open issues (tbachman, 16:14:01)
   1. jmedved says there’s still a memory leak somewhere. When he runs a lot of transactions, the oldgen still keeps growing and doesn’t get freed up (tbachman, 16:14:30)
   2. ACTION: jmedved to collect more information on the memory leak issue (tbachman, 16:14:48)
   3. moizr says there was a similar issue discovered by mirantis on the mailing list4 (tbachman, 16:15:01)
   4. moizr says it’s not just a memory leak, but there’s also just a lot of data in memory (tbachman, 16:15:28)
   5. jmedved says he sees the memory issues even after disconnecting mininet (tbachman, 16:15:44)

Action items

1. harman to look at bug 2449
2. jmedved to collect more information on the memory leak issue

People present (lines said)

1. tbachman (32)
2. odl_meetbot (4)
3. moizr (1)