The question I'd like to raise to the entire list regards Zope's stability. We are running Zope integrated with Apache off FreeBSD3.2 . As far as I know, this is not such an exotic setup (probably rather common?). As Chas explained previously, Zope just suddenly died. Teodore also had the same error occur and solved it by rebooting his server.
How often has this occured with other people? Are there any good solutions regarding providing a backup in case Zope does die? (and keep the site running) Are there any serious Zope instability issues which need to be discussed and alerted?
Obviously, any stability problems get the highest priority from us here at DC - if you don't look good, *we* don't look good :^) We are _extremely_ interested in these "hanging" reports and would like to nail them down as quickly as possible so that we can address any issues internal to Zope and get a 2.1.1 release out there if necessary. For folks that are seeing these symptoms, you can help us speed the diagnosis by: o providing details on your environment (OS, version, zope version in use). Have there been any recent changes in the environment? o symptom description - does CPU go to 100%? If so, does there seem to be any pattern to the requests preceding the hang that you can determine (perhaps by looking at the server log)? Can you regularly reproduce it using a contrived test (like using ab or another benchmarking tool on a particular url or sequence of urls)? Does the problem happen only when accessing the system via pcgi or ZServer (or both)? o run in debug mode (using -D) and send any funny looking or unusual messages that you see o provide any unexplained tracebacks that you see from your site o Look at the "debug" screen at Control_Panel/manage_debug. This can be used to spot memory leaks and stuck database connections. If you think you may see something suspicious, save the document src and send it to us. o You can turn on more detailed logging by adding the following to your z2.py script: # turn on debug logging from ZServer import DebugLogger logfile=os.path.join(INSTANCE_HOME,'var/debug.log') DebugLogger.log=DebugLogger.DebugLogger(logfile).log It should be insertted after Zope is imported, e.g. after this line: exec "import "+MODULE in {} In particular, we want to look for requests that don't complete or for apparent leaking requests. Thanks! Brian Lloyd brian@digicool.com Software Engineer 540.371.6909 Digital Creations http://www.digicool.com