Release Meeting Minutes/2014-05-13

From K5Wiki
Jump to: navigation, search

Will Fiveash, Thomas Hardjono, Nathaniel McCallum, Zhanna Tsitkov, Tom Yu

Will has a few Solaris Kerberos bugs for us to look at. One involves PKINIT and Solaris PKCS11 soft tokens. He'll send mail.

Do you still have a need for kdb5_util dump options "-rev" and "-recurse"? [krbdev.mit.edu #6481] i.e., have you heard any recent reports of database corruption that those options are supposed to work around since that btree bug got fixed?

Will doesn't think so.

Shawn Emery worked with Tom a few years ago on chasing down and fixing a bug [krbdev.mit.edu #5880] in Berkeley DB btree page split code that could make parts of the KDB disappear from either random access lookup or sequential lookup.

We'll be in the KDB iteration code anyway to deal with the long locks during dumps issue, so we might be able to restore the rev/recurse dump capabilities while we're there.

Will says a new hire is working on some memory leak type patches. (Some are in pull requests already.)

Is there a max number of entries in the KDB?
Not as such. If you build with largefiles support (64-bit OSes, many 32-bit ones), there's probably no practical limit due to file size. Performance might become a limiting factor before file size does.