What is the appropriate way to purge nuxeo.tmp.dir?
We recently had nuxeo.tmp.dir fill up, and from looking at the contents, it seems that removing old files is not something done by Nuxeo itself. I selectively removed large files to resolve our outage, but this seemed to result in very unstable operation. After removing the rest of nuxeo.tmp.dir and reloading some software, the instability seems to have been resolved.
What is the correct way to remove unused data from nuxeo.tmp.dir? If there is not a way for users to do this, is there a recommended minimum available space – ours was greater than 12GB – or a way to set the maximum size of the tmpdir?
Thanks, Steve
 |
Steven Huwig
Member (2.3K) 3 5 6 |
01/10/2014 ( History) |
The nxbincache directory is used to store a local cache of files for the S3BinaryManager. Its default maximum size should be 100 MB although this can be increased by the nuxeo.s3storage.cachesize property in nuxeo.conf (see Amazon S3 Online Storage).
There may be a bug which does not clear this cache when the server is shut down. I'll check. In the meantime this cache can be safely purged when the server is stopped.
 |
Florent Guillaume
Member (36K) 6 8 8 |
01/30/2014 ( History) |
I just don't want you to waste too much time in case it was the latter. :)
The space is consumed by files in nxbincache directories, and I assume they are local copies of uploaded files through the drag-and-drop "main file" interface.
We are using S3BinaryManager if that is helpful to know. The only other possibly relevant information I can think of is that the PDFBox indexing was broken (Java 7 bug) when these files were uploaded.
In the absence of knowing why the consumption is so large, is there a generally accepted way to purge just those nxbincache directories?
I expect to reinstall the base binaries from the 5.9.x fast track releases so hopefully that will help eliminate the versioning issues.