Recent monitoring changes report stats per table, however they provide only
the table_id but not the table names. This is somewhat pointless since end
users have no obvious way to figure out the table names from the ids.

Just using the Hyperspace 'readdir' API to track the mapping is somewhat
inefficient or buggy. To do it correctly with the current APIs (esp for the
case where a table is dropped and immediately recreated) would require a
readdir followed by opening all the files and reading the id attribute. This
could be very expensive when the number of tables is large.

Theres a couple of ways to do it more efficiently which require a Hyperspace
API change:

1. Have Hyperspace store the id -> name mapping similar to the way it stores
name-> id mapping. It should be fairly straightforward to implement right
now. However  this might make the upcoming table namespace support a little
more challenging. Also this will be a significant format change and probably
require a re-import or at least an upgrade tool to patch in the mapping for
an existing Hypertable instance.

2. Augment readdir/create readdir_attr_i32 which given a dir and an
attribute, returns a set of filenames which have that attribute and the
value of the attribute. This shd be a quicker and more painless if less
comprehensive change for now.

I'm going to go ahead and implement 2 for now unless concerns are raised or
a better solution is proposed.

-Sanjit

-- 
You received this message because you are subscribed to the Google Groups 
"Hypertable Development" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/hypertable-dev?hl=en.

Reply via email to