haufe.requestmonitoring 0.3.0 (beta)

Zope 2 request monitoring

Fine resolution request logging

Used as base for ztop and zanalyse, i.e. helps to determine the Zope load, detect long running requests and to analyse the causes of restarts.

The implementation in this module registers subscribers for IPubStart and IPubSuccess/IPubFailure. For each of these events, a log entry of the form:

timestamp status request_time type request_id request_info

is written.

<product-config timelogging>
filebase /path/to/request-logs/instance-foo
</product-config>

Success request logging

This logging writes two files <base>_good.<date> and <base>_bad.<date>. For each request, a character is written to either the good or the bad logfile, depending on whether the request was successful or unsuccessful. This means, that only the file size matters for these logfiles.

Usually, response codes >= 500 are considered as unsuccessful requests. You can register an ISuccessFull adapter, when you need a different classification.

To activate this logging, both successlogging.zcml must be activated (on by default) and a product-config section with name successlogging must be defined containing the key filebase. It specifies the basename of the logfiles (represented as <base> above).

Example:

<product-config successlogging>
filebase /path/to/request-logs/successful-foo
</product-config>

Monitoring long running requests

haufe.requestmonitoring allows you to monitor long-running request. The following configuration within your zope.conf configuration file will install the DumpTracer and check after the period time passed for requests running longer than time:

%import haufe.requestmonitoring
<requestmonitor requestmonitor>
    # default is 1m
    period 10s
    # default is 1
    verbosity 2
    <monitorhandler dumper>
        factory Haufe.RequestMonitoring.DumpTraceback.factory
        # 0 --> no repetition
        repeat -1
        time 10s
    </monitorhandler>
</requestmonitor>

A typical dump trace looks like this (it shows you the URL and the current stacktrace):

2009-08-11 14:29:09 INFO Zope Ready to handle requests
2009-08-11 14:29:09 INFO RequestMonitor started
2009-08-11 14:29:14 INFO RequestMonitor monitoring 1 requests
2009-08-11 14:29:19 INFO RequestMonitor monitoring 1 requests
2009-08-11 14:29:24 INFO RequestMonitor monitoring 1 requests
2009-08-11 14:29:24 WARNING RequestMonitor.DumpTrace Long running request
Request 1 "/foo" running in thread -497947728 since 14.9961140156s
Python call stack (innermost first)
Module /home/junga/sandboxes/review/parts/instance/Extensions/foo.py, line 4, in foo
Module Products.ExternalMethod.ExternalMethod, line 231, in __call__
- __traceback_info__: ((), {}, None)
Module ZPublisher.Publish, line 46, in call_object
Module ZPublisher.mapply, line 88, in mapply
Module ZPublisher.Publish, line 126, in publish
Module ZPublisher.Publish, line 225, in publish_module_standard
Module ZPublisher.Publish, line 424, in publish_module
Module Products.ZopeProfiler.ZopeProfiler, line 353, in _profilePublishModule
Module Products.ZopeProfiler.MonkeyPatcher, line 35, in __call__
Module ZServer.PubCore.ZServerPublisher, line 28, in __init__

The log line "RequestMonitor monitoring X requests" simply says that a request is under monitor and sometimes you get useless noise in the log file.

You can play with the verbosity option: put the value to 0 for disable the log line. Default value (1) will display the log line every time one or more requests are under monitor. A value of 2 is more verbose, displaying also info about requests URLs.

Installation

In addition you must haufe.requestmonitoring to the zcml option of your buildout.cfg file or include it within the site.zcml file using:

<include package="haufe.requestmonitoring" />

0.3.0 (2012-10-16)

• do not use deprecated threadframe dependency anymore on recent Python versions [keul]
• fixed egg dependencies for Zope 2.13 [keul]
• added the verbosity configuration option for the logger [keul]

0.2.3 - (2009/08/11)

• updated documentation

0.2.2 - (2009/07/20)

• minor cleanup
• minor documentation cleanup

0.2.1 - (2009/05/28)

• configure 'successlogging' by default
• slightly updated documentation

0.2.0 - (2009/05/12)

• Initial release