[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [jgit-dev] [egit-dev] Counting Objects taking long in current master
- From: Markus Duft <markus.duft@xxxxxxxxxx>
- Date: Mon, 12 Nov 2012 16:43:01 +0100
- Delivered-to: email@example.com
- Organization: Salomon Automation GmbH
- User-agent: Mozilla/5.0 (X11; Linux i686; rv:16.0) Gecko/20121106 Thunderbird/16.0.1
On 11/06/2012 02:37 PM, Matthias Sohn wrote:
> 2012/11/6 Markus Duft <markus.duft@xxxxxxxxxx <mailto:markus.duft@xxxxxxxxxx>>
> On 11/05/2012 11:28 PM, Matthias Sohn wrote:
> > 2012/11/5 Markus Duft <markus.duft@xxxxxxxxxx <mailto:markus.duft@xxxxxxxxxx> <mailto:markus.duft@xxxxxxxxxx <mailto:markus.duft@xxxxxxxxxx>>>
> > I had a look into the stack trace, the only thread related to jgit/egit is the "main" thread and it's in runnable state so nothing unusual on the client side. Could you capture a couple of thread dumps on both egit and gerrit side while the slow push is underway ?
> Here you go. 9 dumps on each side, one every two seconds while pushing. The actual push lasted for ~50 seconds this time, counting 50.000 objects, approx 1000/s. Looking at the dumps i think the problem is entirely on the client side, and not on the server. Moreover it seems to me that maybe the updating of the progress monitor may cause the problem? It causes full round-trips over GTK every object it seems...
> could you try to replace the offending ProgressMonitor by a NullProgressMonitor
> in order to check if this is the culprit ?
I tried now, and it still took some time, but was a /little/ faster, but not really notably. It now seems that the eclipse is rather idle all in all...