DESCRIPTION

MemcachedConnection: handling previously unhandled exceptions in the main run loop of the IO thread. Fixes SPY-191

Handles previously unhandled exceptions in the run loop of the IO thread. Also adds tests to prove. Fixes SPY-191

Change-Id: I43fa77cbe3706ce97b2ca66b1c69609855005eb1


COMMENTS

author: Natasha Sullivan
date: 2012-07-23 15:34:03.840000000

Patch Set 1:

Here are some remarks on that:

- To me, while this fix seems to work out pretty well, the current test cases don't prove the issue. This is because the mocked node raises the error and it doesn't get caught in the catchall block, but rather bubbles up to the actual test case. If someone can quickly spot why this happens, it would be great if you can point me in the right direction to fix it.

- Interestingly, the AsciiClienTest triggers the catchall block (according to some code coverage analysis). This maybe needs also more investigation but is not directly related here. This also shows that the reconnect seems to work correctly and doesn't raise an error.

So, to sum up: seems to work, but the test cases need some tweaking.

-------------------------------------
author: Andreas Short
date: 2012-07-30 21:28:32.193000000

Patch Set 1: Do not submit

(1 inline comment)

Generally looks good, but we need more review and the summary line needs editing.

Line:7, /COMMIT_MSG -> This line is way too long.  Try to stay below 65 chars in the summary line.

Good job though on getting the bug number in there, but, for clarity there are two ways we refer to things:
issue-191 would refer to issues on the google code site
SPY-191 would refer to the Couchbase tracked issues.

In some cases, it may be in both.

Can you try to creatively shorten the summary line and reference issue 191?

-------------------------------------
author: Andreas Short
date: 2013-11-26 07:43:04.369000000

Patch Set 1:

Would be good to finally fix this one up, don't you think?

-------------------------------------
author: Natasha Sullivan
date: 2013-11-28 12:27:28.166000000

Patch Set 1:

jeah, I try to remeber what we did here ... when I started ;)

From this point, I think we should check which node is in a good state and put the others in the reconnect queue, what do you think?

-------------------------------------
author: Natasha Sullivan
date: 2015-01-29 15:23:43.636000000

Abandoned

-------------------------------------
