Bugzilla – Full Text Bug Listing
| Summary: | .snap file of a second workspace grows up to 44 GB | ||
|---|---|---|---|
| Product: | [Eclipse Project] Platform | Reporter: | info |
| Component: | Resources | Assignee: | Platform-Resources-Inbox <platform-resources-inbox> |
| Status: | RESOLVED WORKSFORME | QA Contact: | |
| Severity: | normal | ||
| Priority: | P3 | CC: | remy.suen, sptaszkiewicz, yevshif |
| Version: | 3.6 | ||
| Target Milestone: | --- | ||
| Hardware: | PC | ||
| OS: | Linux | ||
| Whiteboard: | |||
|
Description
info
What's a .snap file? It's not something the CDT produces. The plugin name under the metadata should give you a hint of where this really belongs. Thank you for the quick response.
Does the path
workspace_1/.metadata/.plugins/org.eclipse.core.resources/.snap
tell to which plugin the .snap file belongs?
I googled for .snap file and found "*.snap files represent the changes in workspace state of the IDE during the runtime. This is mostly for eclipse crash recovery plan. When a crash happens these files are used to recover the state fo eclipse workspace." at http://stackoverflow.com/questions/2026502/what-are-snap-files-in-eclipse It would be somewhere in the Eclipse Platform. org.eclipse.core.resources is the plug-in name. What do you mean by "I created a few C++ projects from existing sources"? Did you attach the same projects from the same location to another workspace or you copied the contents of the projects to the new workspace? 44GB in 20 hours means that your .snap was growing with average speed 640 KB/s. I tried to prepare a test to check if it is possible but I could not get even half of that speed. Since .snap is usually much smaller than 1 MB in total, there must have been some very extensive job running over night that was performing workspace operations. Do you know what it could be (build, download, repository synchronization, etc)? What was your CPU and network usage during that night? Thanks for your response. It was the same source trees than in other workspace. One tree was on a server and checked out from another server. A second tree with the whole framework probably has 10,000 files or more. It was copied to a local disk for performance reasons. I wanted to have two Eclipse open on two screens both operating on the same sources but one was used read-only. The CPU was a two-years old Intel Dual Core - probably 3.0 GHz - running with a Redhat linux. I don't know about network traffic but I created the second workspace in the afternoon and next day shortly before noon I was informed that my PC had produced a 44 GB logfile on a common network drive - shared by about 20 people - what made the drive full (and produced serious performance issues for the whole crew). The file was already deleted, so I could not look into but I got a report of the directory listing where I could verify the huge size of the file. In the night no one was working and the performance problems already were reported from colleagues which worked at 8 a.m. So I assume that the disk nearly was full already in the morning. I myself came at 10 and had problems to access the server but not to access my local disk. Also I didn't remind of CPU issues though that could be hidden by the two cores of the CPU. No, I don't know the reasons for the 44 GB but as I had two Eclipse programs open both trying to analyze huge project trees, I assume those automatical activities were the only one what happens in the night on the computer. I didn't start a compile job or other things. The only reason why I didn't close the Eclipse programs was the poor performance on program start which could last up to 10 minutes. After that I never tried (dared) to start a second instance of Eclipse again. The case seriously damaged my reputation as a professional and I had no motivation to run a risk again. We were never able to reproduce this problem. Please reopen if you can reproduce the problem with Luna. |