Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Threaded View
HBase >> mail # dev >> HConnectionManager Docs

Copy link to this message
Re: HConnectionManager Docs
Hi Ted,

Sure thing, that part is clear. See my questions underneath for what I
was after. If it closes the RPCProxy globally then it would kill all
other connections too. And if it does not, then why offer this option
at all and not simply always close the RPCProxy - why would you keep
it open?


On Mon, Jun 20, 2011 at 8:19 PM, Ted Yu <[EMAIL PROTECTED]> wrote:
> For 1, stopProxy flag is for RPC proxies.
> On Mon, Jun 20, 2011 at 10:38 AM, Lars George <[EMAIL PROTECTED]> wrote:
>> Hi,
>> It seems that the HCM docs are now outdated, as we now keep a reference
>> count of each connection, and as long as you call HTable.close() you are OK
>> with shared resources not lingering on. What I was wondering is:
>> 1. Why do we offer a "stopProxy" flag?
>> Is that closing out all proxies for all connection, even if they are still
>> active? Or is that for only the RPC proxies used by that connection?
>> And why do we have a flag at all? Wouldn't you always want this to be
>> cleaned up? Especially for deleteAllConnection()?
>> 2. What is getConnection() for?
>> Who is using this outside of HBase and Ruby scripts?
>> Lars