I think the restart bug is finally fixed
Drew Csillag <[email protected]>
| Newsgroups | gmane.comp.web.skunkweb |
|---|---|
| Message-ID | <[email protected]> |
I finally think the long standing restart bug is now fixed in CVS! The problem was that if the stderr and stdout from the original start were dead (because the starting user logged out and the terminal no longer existed), SkunkWeb would die on swmgr restart. The problem was twofold. First, skunkweb.py, on exception was writing something to the now-bogotified stderr which would itself raise an exception (so you wouldn't see the real problem). Now, it logs the existing exception first, then writes to stderr -- if that fails, that's ok. Second, there were cases where it was trying to redirect stderr/out to logfiles more than once, whereby the Redirector instance has no isatty attribute, and it would fail. Now it checks to see that sys.__stderr__ == sys.stderr (and similarly for stdout) before redirecting them. I also saw something with a bind() call that failed, though that may have been something else, but I put a retry in the bind so that if you do swmgr stop; swmgr start, the starting skunkweb will try binding for about 5 seconds or so before giving up. Please beat up on this code to see if the bug is indeed squashed. Thanks, Drew __________________________________ Do you Yahoo!? Yahoo! Mail - More reliable, more storage, less spam http://mail.yahoo.com ------------------------------------------------------- This SF.Net email is sponsored by: IBM Linux Tutorials Free Linux tutorial presented by Daniel Robbins, President and CEO of GenToo technologies. Learn everything from fundamentals to system administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click