Part way to a fix, I think the problem lies in the ThresholdClass in this function:
def getDataPointNamesString(self):
"""
Return a string that lists the datapoints used in this threshold.
Indicate missing datapoints with (missing) after the name.
"""
names = []
availableDPNames = self.rrdTemplate.getRRDDataPointNames()
for dsName in self.dsnames:
if dsName in availableDPNames:
names.append(dsName)
else:
names.append('%s(<span style="color: red">missing</span>)' % dsName)
return ', '.join(names)
Which returns the datapoints seperated by both a comma and a space, other functions seem to be splitting on the comma but treating the space as part of the dp name, thus breaking it when the threshold is updated. The situation gets worse because the html tag is then appended to the dp name as well, and again each time the form is saved, so there needs to be some 'intelligence' put in to the threshold update function in ZenRRD/Thresholds.py here:
def add(self, threshold):
self.byKey[threshold.key()] = threshold
d = self.byDevice.setdefault(threshold.context().deviceName, {})
d[threshold.key()] = threshold
ctx = threshold.context()
for dp in threshold.dataPoints():
## Need to clean up dp name - remove any HTML tags etc left from web gui ###
self.byFilename.setdefault(ctx.fileKey(dp), []).append((threshold, dp))
Not sure if this is the best place to do it - yes because the data is validated at the point it is stored, or no it should be validated/cleaned by the originating function (which I haven't tracked down yet!). Also not sure of the impacts / best way of doing it...
for now I have altered the return code from getdatapoints as follows:
# return ', '.join(names)
return ','.join(names)
So I shall see if I have broken anything else!
David
Moved to zen-dev as a bug....
