[memcached] bradfitz, r309: notes
commits at code.sixapart.com
commits at code.sixapart.com
Mon Aug 21 05:14:26 UTC 2006
notes
U branches/facebook/TODO
A branches/facebook/test/TODO.txt
Modified: branches/facebook/TODO
===================================================================
--- branches/facebook/TODO 2006-08-21 03:45:03 UTC (rev 308)
+++ branches/facebook/TODO 2006-08-21 05:14:25 UTC (rev 309)
@@ -1,3 +1,20 @@
+* bug as shown with netcat (w/ small 16 byte object reproduces)
+
+>>I've done the following script to check that memcached has a key in it's
+>>inside, and thus know that it's working correctly:
+>>echo -e "get is_ok\r\nquit\r\n" | netcat $host $ip
+>>
+>>and I find that sometimes it returns the VALUE in it's inside, but other
+>>not.
+
+* namespaces
+
+* binary get protocol
+
+* refresh/touch command.
+
+* finer granularity of time for flush_all/delete, or generation number.
+
* slab class reassignment still buggy and can crash. once that's
stable, server should re-assign pages every 60 seconds or so
to keep all classes roughly equal. [Update: fixed now?, but
Added: branches/facebook/test/TODO.txt
===================================================================
--- branches/facebook/test/TODO.txt 2006-08-21 03:45:03 UTC (rev 308)
+++ branches/facebook/test/TODO.txt 2006-08-21 05:14:25 UTC (rev 309)
@@ -0,0 +1,5 @@
+# set / delete / set / get all in a row -- subsecond resolution needed?
+ -- same with flush_all followed by get?
+
+
+
More information about the memcached-commits
mailing list