Why does the RTC client in Eclipse show intermittent repository disconnects?
Our users run Eclipse 3.7 with the RTC 4.0.1 plugin. A couple of them suffer from a strange intermittent error. From time to time, in their Team Artifacts view, their repository connection will turn red, indicating a disconnect. Related information (source control, etc.) becomes unavailable. This lasts a few minutes. Then the red mark disappears and normal use can resume. We though it could be a symptom of Eclipse exhausting its heap. We turned on the heap status display (Windows -> Preferences -> General -> Show heap status). During the last disconnect episode, the heap status was showing 330 MB free, so this is not a heap problem. In Windows -> Preferences -> General -> Network Connections, the Active Provider is set to Native. However, while the plugin shows a repository disconnect, the user can still browse the Jazz server using the same URL as in the Jazz repository connection definition, so the native OS is definitely not barring access to this server. There is nothing in the $WORKSPACE/.metadata/.log file. How do we debug this? Is there a log somewhere else? Is there an option to turn on more logging in the RTC plugin? |
4 answers
the connections are configured to timeout by default.. did your users change that setting, for each connection defined?
|
|
Hi Fredrick,
Hi Frederic, 1. Does this problem persist when the network provider is set to 'Direct". 2. At the time the problem occurrs, are there any errors in the ccm.log or reverse proxy (if one exists) 3. In the workspace folder, are there any additional files being created...if there are problems, eclipse may create other log files which can capture failiar information. 4. In the eclipse home directory, are there any cores / snaps being generated 5. Has an attempt to increase the time out been done 7. SInce this only occurs for a sub set of users, is there any difference between the setups of the working and problematic environments. |
Hello Abraham,
> 1. Does this problem persist when the network provider is set to 'Direct". Yes, it does. > 2. At the time the problem occurrs, are there any errors in the ccm.log or reverse proxy (if one exists) We have no reverse proxy. Tailing the logs shows no correlation between the disconnect incidents and the logs. > 3. In the workspace folder, are there any additional files being created...if there are problems, eclipse may create other log files which can capture failiar information. The only log updated within a few hours is /.metadata/.log, but it is not updated at the time of the problem. Its most recent error shows:
!MESSAGE
CRJAZ1887I The list of certificates available in the Windows-MY
KeyStore could not be read. If you'd like to access the certificates
from that key store, make sure that you are running with a JVM that
supports reading from that key store.
which is unrelated.
> 4. In the eclipse home directory, are there any cores / snaps being generated No, no Java cores or snaps. > 5. Has an attempt to increase the time out been done No, still 480 seconds. > 7. SInce this only occurs for a sub set of users, is there any difference between the setups of the working and problematic environments. One of the problem users is running Ubuntu 12.04 with a KDE desktop. Not sure about the others. I need to gather data. Comments
Jamie Berry
commented Mar 18 '15, 3:21 p.m.
Did you ever get a resolution on this? We have a group of users with Eclipse 4.3.2 and RTC 4.0.2 that are reporting the same issue you described.
|
Your answer
Dashboards and work items are no longer publicly available, so some links may be invalid. We now provide similar information through other means. Learn more here.