fix for issue #48 and logging fix #72
Conversation
| log = milterContext.uuid+" Sending "+ str(milterContext.qid)+" to "+ SERVER_ENDPOINT | ||
| self.logger.writeLog(syslog.LOG_DEBUG, "%s"%(str(log))) | ||
| myhostname = socket.gethostname() | ||
| externalObject = ExternalObject( | ||
| buffer=milterContext.fileBuffer, | ||
| externalVars=ExternalVars( | ||
| filename=milterContext.archiveFileName, | ||
| source=milterContext.milterConfig.milterName+"-"+str(myhostname[:myhostname.index(".")]), | ||
| source=milterContext.milterConfig.milterName+"-"+ \ | ||
| (str(myhostname[:myhostname.index(".")]) if "." in myhostname else str(myhostname)), |
There was a problem hiding this comment.
how about using myhostname.split(".")[0] instead? seems cleaner.
There was a problem hiding this comment.
Your right, I didn't realize splitting on something that might not exist would just throw it into the first entry of the list. That would be a better fix.
… error to the module - non-string children can crash the worker
|
@marnao A new commit has been added that fixes the non-string child object bug (it caused the worker to crash, an earlier pull was a partial fix for this). I see that this request hasn't been active - are there any other changes we need to make to JMPercival's work? |
|
@marnao Checking in again - these fixes have worked long-term for us, this should be ready to integrate. Let me know if it requires more changes. |
|
I am rolling these into our baseline for testing. I will merge as soon as I verify. |
|
@marnao Just checking in again. We have other stuff we want to submit but we don't want more stuff in this pull request probably. |
| if not type(child_buffer) is str: | ||
| raise Exception("Buffer of %s found, not creating child scanObject" % str(type(child_buffer))) | ||
|
|
||
| child_buffer = ensureNotUnicode(child_buffer) |
There was a problem hiding this comment.
I think you have a logic error here. This function will never get called for a unicode object because you're raising an exception above. Therefore instead of a unicode object getting encoded to a utf-8 string, you'd get an exception instead. Can you address this?
There was a problem hiding this comment.
I think you're right - we've run initial tests and put the code in our environment, we'll push the fix once it's run for awhile/been monitored for errors.
Moving a line around so that unicode is dealt with before throwing an error.
|
@marnao Pushed a fix - just moved that one line around. We have the same fix running in our environment and haven't seen it cause any errors (though we don't get unicode often). |
…es 3.support max memory threshold in redis queue 4. better handle not using ldap directory 5. better handle docker issues around mounting files vs directories 6. fix issues parsing base64 with invalid padding 7. fix lnk parser and pull out more metadata, 8. better logging, 9. support ARM process for mac binary metadata, 10. better documenation in readme, 11. change default container name ---WARNING BREAKING change
No description provided.