Community
Participate
Working Groups
Summary was a little lame, here goes: (Eclipse newbie here, btw, wasn't certain which component of Eclipse the CVS subsystem belonged to, so marked it 'core') I have a CVS repository on a linux box. There are projects there which I would like to work with in Eclipse. I'm using the pserver connection method and login procedure works fine, no problems. 1) Checking out modules work fine 2) Updating files in projects work fine 3) Probably many other CVS operations work fine too 4) BUT: Whenever I do a Team->'Compare to HEAD' or Team->'Synchronize with Repository' on the _whole_project_tree_(not individual files), Eclipse always stops in the middle of the operation(it's starts off okay), reporting a simple network IO error/timeout. Eclipse doesn't crash after I close the error dialog, but in fact, problems are indeed present(5). There is _nothing_ wrong with my network connection _before_ this. (It's my own 100mbit home LAN) 5) After Eclipse has failed, the system is shaky: ALL Windows networking is down and totally dead. A restart of windows _IS_ needed (I've tried many actions to recover without doing a restart, including ethernet interface restart, but no luck, something has gone terribly wrong in the networking code somewhere). This makes me think the bug is severe, perhaps it's even in the JVM networking code. Note1: The same thing happens on all the projects I have in my repository (fresh checkouts into Eclipse workspace). Note2: All other CVS client software I've tried (including WinCVS, netbeans builtin, official win32 CVS commandline client 1.11.5, jCVS) all work flawlessly with my repository, and have never reported any such errors. Note3: Only Eclipse is running (at least, the only running foreground task) Note4: Excluding this problem, Eclipse seems fine. Next part of this bug report is version details(mostly): Windows XP SP1 (fully patched as of 05.03.03) Eclipse 2.1, downloaded 05.03.2003, fresh install w/no additional plugins. Java(TM) 2 Runtime Environment, Standard Edition (build 1.4.1_02-b06) (automatically discovered by Eclipse install procedure) CVS server is: official 1.11.5 (compiled and running on an upgraded redhat 7.3 box) System memory on laptop running Eclipse is 512MB. Very stable XP installation. Let me know of methods which might provide a little more details, when this happens, and I'll be happy to try them out.(Though I might get bored of seeing the XP logo, after all the restarts that will probably be necessary). I haven't marked the bug as severe, because I have a feeling it will be hard to track down, perhaps it's a very obscure case. I don't know. But still, it certainly is severe enough to render Eclipse useless for handling CVS projects for me.
I have an idea of what may be causing this. Is this repeatable for you? If so, would you be willing to try out a patch?
It is very repeatable indeed, as stated in the report. I could try a patch, sure.
Before I attach a patch, I need to know the exact build. Did you download the official 2.1. build or the latest integration build. If your not sure, please perform the following. From the help menu of Eclipse please choose About Eclipse and copy the version and copy the version and build id into the bug report.
Eclipse Platform Version: 2.1.0 Build id: 200303272130
I have released code to HEAD which may help in this situation. It will be in tomorrow's integration build (which should be available sometime tomorrow morning EST).
Please download build I20030506 and try it out. Let me know if this fixes the problem.
This build seems unavailable for the windows platform at the moment, because of errors in the win32_refactoring module. Checked at around 1030PM CET.
Correction: not unavailable, just had errors (in build, I guess). I'll try it, since refactoring isn't partof the problem anyway.
If you are using pserver or ext, then go ahead and try it. If you use extssh, you may want to hold off on it. A bug was just found that is not critical but is anoying (see bug 37204). Hopefully the fix will go into a rebuild.
Oops, I meant "see bug 37292".
I'm using the plain pserver connection method. I can confirm that I was unable to reproduce this bug with build I20030506 of Eclipse ("Integration Build"). In other words: It was fixed, as far is I know. (Not thorough testing, but the Eclipse release 2.1 crashed so heavily all the time, that I'm pretty sure it's OK now)
I am using Windows XP and build I20030507 of Eclipse and I am experiencing exactly the same -> I have to restart windows because it freezes totaly for twenty seconds, works one second and then freezes again I am using a Netgear F311 network card.
The original problem has been fixed as indicated by the reporter of the bug. The issue mentioned in comment #12 is most likely a separate issue, I would ask the reporter of that problem to open a separate bug report.
*** Bug 21276 has been marked as a duplicate of this bug. ***
This bug is still present in Eclipse 3.0, build 200402122000. I have exactly the same symptoms reported in the original description for this bug. I am running WinXP and have a Realtek RTL8139C+ NIC. The only "solution" is to reboot. Was the fix applied to the 3.0 stream as well? (I was unable to change the version number to 3.0 for this report, and could not change the severity although I feel it is major.)
Hi, I have the problem as described at the head of this bug report. Win XP, Eclipse 2.1.3 build ID 200403101828. If I synchronise with repository it kills my network connection completely. I then need to reboot. The interesting thing is for about 10 days on my new laptop the connection to CVS was fine!!! so I wonder whether XP or network card itself has a problem? BTW I am using and always was using the default settings for CVS (under Windows > Preferences > Team) - other guys on the project have same version of Eclipse and same CVS settings same settings and none of them have the problem Do you know how I can get round this?
These problems are usually related to the network card. I suspect that the pserver connection method is sending data in chunks that are too small for some cards to handle. I plan to investigate before 3.0 ships.
*** This bug has been marked as a duplicate of 38430 ***
I observed also network freezes with Eclipse, but could track it down to the firewall, F-Secure in my case. After deinstalling the firewall, the problem went away. I wonder if the firewall mistakes Eclipse for a worm or so, when Eclipse does a lot of network connections in quick succession.
