[
https://issues.apache.org/jira/browse/SOLR-2193?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13017461#comment-13017461
]
Mark Miller commented on SOLR-2193:
-----------------------------------
{quote} I wonder how this should work with autocommit?
Someone may want a soft/reopen autocommit once every x seconds, but still may
want a hard "flush to stable storage in case I crash" commit at some other
duration. {quote}
Right - I agree. How about another simple start? Simply add another
commitTracker that does soft commits - then you can schedule a mix of soft and
hard commits.
{quote}
The other thing that might be cool is a client-specified freshness per request.
For example, when they pass in a query, they specify that they need data that's
no more than 1 second old... and if it's too old that will trigger a reopen
(and block that specific request until the new searcher can be used). The
benefit here is that big bulk uploads won't be interrupted if there is no time
sensitive query traffic. The downside is that a high latency may be exposed to
those requests if they depend on stuff that can take a lot of time the first
time (like faceting).
{quote}
Yeah - I remember you mentioning this before - I def think this would be cool -
perhaps as a follow on issue - though hopefully the affect on bulk updates will
be minimized when Lucene takes care of the 'flush blocks the world' issue.
> Re-architect Update Handler
> ---------------------------
>
> Key: SOLR-2193
> URL: https://issues.apache.org/jira/browse/SOLR-2193
> Project: Solr
> Issue Type: Improvement
> Reporter: Mark Miller
> Assignee: Mark Miller
> Fix For: 4.0
>
> Attachments: SOLR-2193.patch
>
>
> The update handler needs an overhaul.
> A few goals I think we might want to look at:
> 1. Cleanup - drop DirectUpdateHandler(2) line - move to something like
> UpdateHandler, DefaultUpdateHandler
> 2. Expose the SolrIndexWriter in the api or add the proper abstractions to
> get done what we now do with special casing:
> if (directupdatehandler2)
> success
> else
> failish
> 3. Stop closing the IndexWriter and start using commit (still lazy IW init
> though).
> 4. Drop iwAccess, iwCommit locks and sync mostly at the Lucene level.
> 5. Keep NRT support in mind.
> 6. Keep microsharding in mind (maintain logical index as multiple physical
> indexes)
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
