Bugzilla – Bug 4033
d_scan output in log.txt
Last modified: 2008-09-15 14:39:24 UTC
The output of d_scan that appears in log.txt is much, much more sparse than we had in 6.3. In fact it appears to be just one line: Setup::rescan - initiating scan of type: [wipecache] Full log output is of course still available by running scanner.exe with the --d_scan option from the command line.
I thought it was to be used on combination with other. --d_scan --d_import shows more, --d_info, d_musicmagic and d_moodlogic should all be passed to the sacanner process. admittedly, I'm usually doing this from command line. maybe the logfile option isn't being passed to scanner?
--logfile needs to be passed. It's not right now.
Fixed in change 9388 Note that since http://localhost:9000/log.txt is kept in memory, it is inherently broken for logging with a multi-process application. I've added the passing of --logfile to the scanner process, but that data won't get back to the main server. Logging is being completely rewritten for 7.0, so it will be addressed properly there.
Does that mean that in order for me or support to get a scanner log from a customer, we'll have to have them run it from the command line?
Unfortunately yes. The only alternative I can think of is to always log to a file, and then display that log when requesting /log.txt But there's no easy way to truncate / rotate that now, so it could grow incredibly large, both on disk, and when loading that file in a web browser.
FYI Kevin. Might be a good idea to have your team familiarize themselves with running scanner.exe from the command line.
My workaround is as follows: Open up a command prompt window, and type "cd \program files\slimserver\server" to get to the right directory. Then type "scanner --wipe --d_import --d_scan > log.txt 2>&1" to save the output of the scanner process into a file called log.txt. You can of course use a different filename if you wish. The strange "2>&1" bit on the end is necessary to also grab the output that is sent to stderr, if you're familiar with programming.