Tuesday, June 12
An extraneous nohup in a startup script.
That was it.
That's what was behind the bug I've been tracing for seven hours.
I need a drink.
Okay, so while I was crawling around in the logfiles I was able to find and fix a couple of intermittent bugs that have been plaguing us for a while. Including a nasty one where the session management code would, under certain circumstances, delete the database cursor for a thread and so ensure that the subsequent page request would drop dead. (The error management code would reinstate the cursor on the next request, but that particular page would fail. It happened to me just now, editing this post.)
But still. Seven hours for a nohup.
Send me an email will you, I made a quick pass looking for a contact email for you didn't find one.
Posted by: blackflag at Tuesday, June 12 2007 01:02 AM (Mq5jS)
with the advent of job control in 1980s. It was 20 years ago. Not five, not ten.
Posted by: Pete Zaitcev at Wednesday, June 13 2007 07:25 AM (9imyF)
I haven't completely switched the application from a console-based app (which makes for easier debugging) to a daemon, so I whacked a nohup into the startup script. That worked, until I ended up putting two nohups into cascading scripts and it all went flooie.
Posted by: Pixy Misa at Wednesday, June 13 2007 03:58 PM (PiXy!)
Posted by: Pixy Misa at Wednesday, June 13 2007 05:25 PM (PiXy!)
Posted by: Pixy Misa at Thursday, June 14 2007 03:19 AM (PiXy!)
56 queries taking 0.1693 seconds, 335 records returned.
Powered by Minx 1.1.6c-pink.