Comments for domas mituzas http://dom.as Sat, 17 Dec 2011 20:14:59 +0000 hourly 1 http://wordpress.com/ Comment on Blowing up in memory by Mark Callaghan http://dom.as/2011/09/25/blowing-up-in-memory/#comment-2313 Sat, 17 Dec 2011 20:14:59 +0000 http://dom.as/?p=1509#comment-2313 Wasted a few hours on this today.

]]> Comment on on MySQL replication prefetching by Mark Callaghan http://dom.as/2011/12/03/replication-prefetching/#comment-2303 Fri, 16 Dec 2011 00:28:57 +0000 http://dom.as/?p=1536#comment-2303 Robert – only our recent RPMs have IO statistics for the replication threads in information_schema.user_statistics. That is a good metric but it will take a while to deploy the RPM.

]]> Comment on Blowing up in memory by Marko http://dom.as/2011/09/25/blowing-up-in-memory/#comment-2297 Tue, 13 Dec 2011 14:24:09 +0000 http://dom.as/?p=1509#comment-2297 Can you, please, describe how are you debugging and profiling UNIX programs such as MySQL in XCode?

]]> Comment on Blowing up in memory by Marko http://dom.as/2011/09/25/blowing-up-in-memory/#comment-2292 Sun, 11 Dec 2011 13:12:51 +0000 http://dom.as/?p=1509#comment-2292 Is there any howto for creating Xcode projects and using Xcode for not Mac/iOS software?

]]> Comment on Google :-( by hashar http://dom.as/2011/12/05/google/#comment-2290 Sat, 10 Dec 2011 20:11:33 +0000 http://dom.as/?p=1539#comment-2290 Get your Facebook friends to implement a XMPP gateway with voice/video support compatible with iChat and Google. Then the next thing missing would be Docs wich Microsoft can probably help with :-)

Et voila. You wil be free from Google!

]]> Comment on Google :-( by Giuseppe Maxia http://dom.as/2011/12/05/google/#comment-2282 Tue, 06 Dec 2011 22:35:37 +0000 http://dom.as/?p=1539#comment-2282 But of course! You work for the competition. What did you expect?

]]> Comment on on MySQL replication prefetching by Robert Hodges http://dom.as/2011/12/03/replication-prefetching/#comment-2280 Tue, 06 Dec 2011 06:20:35 +0000 http://dom.as/?p=1536#comment-2280 Dumb question to follow last comment: how do you see I/O on the slave thread? That’s a great prefetch metric.

]]> Comment on Google :-( by Matt Lord http://dom.as/2011/12/05/google/#comment-2279 Mon, 05 Dec 2011 21:33:05 +0000 http://dom.as/?p=1539#comment-2279 I’ve heard of similar things happening to G+ users whom administrators suspected to be in violation of their terms of service…

]]> Comment on on MySQL replication prefetching by Robert Hodges http://dom.as/2011/12/03/replication-prefetching/#comment-2272 Sun, 04 Dec 2011 01:45:48 +0000 http://dom.as/?p=1536#comment-2272 Hi Domas! Tungsten Replicator will soon have replication prefetch as well. We are using the same technique as Yoshinori but let Tungsten take care of parallelization using parallel apply, which can work on things other than actual updates. We’ll post a blog article as soon as we get some plausible test results.

]]> Comment on Blowing up in memory by James Day http://dom.as/2011/09/25/blowing-up-in-memory/#comment-2250 Wed, 23 Nov 2011 23:40:39 +0000 http://dom.as/?p=1509#comment-2250 Anyone interested in this should subscribe to or otherwise watch http://bugs.mysql.com/bug.php?id=57480 which is where work on this is being done. Very extensive work has been done. Among other things the work has included:

1. Moving many allocations to the heap instead of malloc.
2. Merging multiple allocations into one in several cases.
3. Better estimates of sizes needed for some allocations.
4. Defer allocation of upd_buff, of the size of the data row, that’s only needed if a row is updated.

Those who have an Oracle MySQL support subscription can also watch its mirror bug 11764622 in that system.

I asked for escalation of your patch for innodb_max_dirty_pages_pct and that’s been assigned to a developer. No ETA, just means it’s likely to get done sometime reasonably soon. Not guaranteed, of course.

This isn’t the official position of Oracle corp; for that consult a PR person.

James Day, Oracle

]]>