Suggestions for optimizing EGit on Eclipse

Question

I use EGit for a fairly large and complex set of Java projects (over a million lines of code) and a decade of history.
Here I am facing serious performance issues with EGit, since even a small change in one line in the Java file causes EGit to re-index a couple of minutes, which slows down the whole system. Indeed, even the git command line is a bit slow, since "git status" takes about a minute from the command line, but I can live with this performance problem, and the problem is with the delay of the EGit () dialog box. Since I can use the git command line to commit and update, but I do not want to compromise the performance of Eclipse, as this affects performance.

The following is what I tried to do googling and ask people around:

• Added folder of all classes in the exclude file. In fact, I also tried adding the folderin .gitignore folder.
• Give enough time to complete the indexing by holding the machine for a day.
• Git staging, history, and all other header views are closed at the Eclipse workstation during development.
• Did "git gc" - it affected the command line performance, but hardly any difference for EGit.
• Unverified shortcut for git. Settings → General → Appearance → Labels.
• Removed cygwin from the path, as read somewhere in the forum, that JGit can use cygwin to convert the path.
• Increased window cache from 10 to 70 m in Eclipse (Preferences → Team → git → window cache).

PS: git repository points to the remote svn repository. Also, I am a newbie to git, so I might have made a mistake in setting it up, so please feel free to point out something.

Here is my system information, I do not have very fancy hardware specifications, but with some RAM in reserve (8 GB).

• git -gui version 0.16 GITGUID
• Git version: 1.7.10.mysysgit.1
• JDK 1.6_025
• Eclipse version: 3.7.2 Java EE version with parameters -Xms1536m -Xmx1536m
• EGit: 1.3.0.201202151440
• Windows 7 processor: Core 2 Duo 2.6GHZ

Answer 1

This is the problem between CVCS (Centralized VCS) and DVCS (Distributed) VCS:

• One SVN repo can contain data in GB format.
• a Git repo should be kept small and submodules should be used to represent across multiple Git repos .

I suspect that many repositories might work better than one giant Git repository. Otherwise, synchronization problems will begin, for example, in error 323839 .

But this means managing (simplified) synchronization between Git repositories and one SVN repo manually, through the SVN workspace from which you copy to your Git repositories, or into which you copy Git repos new designs, return to the SVN workspace for commit.

Answer 2

This is probably not exactly your problem, but this page appears on google regarding egit performance. When is the source of performance problems unverified (indexed?) Files. Make sure that there are not a lot of unprocessed files in the directory tree, as this will seriously affect performance. I removed the director with the 10K + files and fixed the efficiency from 1 + minutes to open the commit dialog for a couple of seconds.