Not the greatest solution, but makes the 3 bro processes more likely to
run sequentially so that the controller2 process doesn't happen to be
scheduled before the controller process. In that case, the controllee
gets the shutdown request before the configuration update. FreeBSD
especially seemed to schedule them the unintended way frequently.
* remotes/origin/topic/bernhard/input-warn-on-invalid-numbers:
...and another small change to error handling -> now errors in single lines do not kill processing, but simply ignore the line, log it, and continue.
Ok, this one was a little bit sneaky.
ok, this one might really be a bit too big for 2.1
* origin/fastpath:
Ok, this one is not really necessary for 2.1 and more of a nice-to-have
another small bug found while searching for something else...
Fix two little bugs:
sorry. the patch for the set_separator.
make set_separators different from , work for input framework.
Bug found bei Keith & Seth: input framework was not handling counts and ints out of 32-bit-range correctly.
Before this patch, empty values were not hashed at all. Which had the unfortunate side-effect
that e.g. the lines
TEST -
and
- TEST
have the same hash values. On re-reads that means that the change will
be ignored.
This is probably pretty academic, but this patch changes it and adds a testcase.
Output of the reread test changes due to re-ordering of the output (probably
due to the fact that the internal hash values are changed and thus transferred
in a different order)
Escaped ,'s in sets and vectors were unescaped before tokenization
Handling of zero-length-strings as last element in a set was broken (sets ending with a ,).
Hashing of lines just containing zero-length-strings was broken (now a \0 is appended to each
string before it is hashed - giving us a hash of something for a line just consisting of \0s.
This also allows to differentiate between vectors with varying numbers of zero-length-strings).
* origin/topic/bernhard/input-allow_invalid_types:
to be sure - add a small assertion
add an option to the input framework that allows the user to chose to not die upon encountering files/functions.
That's the last feature for 2.1!
If the metrics break interval happened to occur between first and second
worker starting up and getting connected to the cluster, the test
would fail because the second worker didn't get a chance to connect and
send data. The test now waits for the cluster setup to complete
before workers send metrics data.
This one would fail intermittently in the cases where log files were
opened or closed on a different second of the time of day from each
other since the "out" baseline contains only a single "#open" and
"#close" tag (indicating all logs opened/closed on same second of time
of day). Piping aggregated log output through the timestamp canonifier
before `uniq` makes it so "#open" and "#close" tags for different
seconds of the time of day are reduced to a single one.
- Some baselines for tests in "leaks" group were outdated.
- Changed a few of the cluster/communication tests to terminate
more explicitly instead of relying on btest-bg-wait to kill
processes. This makes the tests finish faster in the success case
and makes the reason for failing clearer in the that case.
* origin/topic/seth/reporter-to-stderr:
A couple of tests for printing reporter messages to STDERR.
Small improvements for printing reporter messages to STDERR.
Reporter warnings and error now print to stderr by default.
Closes#836.
An assertion would trigger in the case when a predicate refuses
a new entry and another entry with the same index elements was
already in the table. (I thought that code block was unreachable
... did not think of this case).
* origin/fastpath:
Small (potential performance) improvement for logging framework.
Script-level rotation postprocessor fix.
update input framework documentation to reflect want_record change.
Fix crash when encountering an InterpreterException in a predicate in logging or input Framework.
make want_record=T the default for events
Small tweak: I added the "same writer" constraint to the loop
condition as well. Makes sense?
* origin/fastpath:
Change path conflicts between log filters to be auto-corrected.
There seems to be a race condition in capturing the external shell's
stdout output reliably. As far as I can tell, Bro's doing everything
correctly though, the log postprocessors gets executed as expected. So
I rewrote the test to capture the output in a separate file first, and
that seems to solve the test failures.
This change makes it so when differing logging filters on the same
stream attempt to write to the same writer/path combination, the path
of the filter doing the later write will be automatically adjusted so
that it does not conflict with the other. The path is adjusted by
appending "-N", where N is the smallest integer greater or equal to 2
required to resolve the path name conflict.
Addresses #842.
add missing binary testcase (Baseline is in master, testcase is missing for some reason)
make error output for nonmatching event types much more verbose
Since WriterFrontend objects are looked up internally by writer type and
path, and they also expect to write consistent field arguments, it could
be the case that more than one filter of a given stream attempts to
write to the same path (derived either from $path or $path_func fields
of the filter) with the same writer type. This won't work, so now
WriterFrontend objects are bound to the filter that instantiated them so
that we can warn about other filters attempting to write to the
conflicting writer/path and the write can be skipped. Remote logs don't
appear to suffer the same issue due to pre-filtering.
Addresses #842.
Turns out the finish methods weren't called correctly, caused by a
mess up with method names which all sounded too similar and the wrong
one ended up being called. I've reworked this by changing the
thread/writer/reader interfaces, which actually also simplifies them
by getting rid of the requirement for writer backends to call their
parent methods (i.e., less opportunity for errors).
This commit also includes the following (because I noticed the problem
above when working on some of these):
- The ASCII log writer now includes "#start <timestamp>" and
"#end <timestamp> lines in the each file. The latter supersedes
Bernhard's "EOF" patch.
This required a number of tests updates. The standard canonifier
removes the timestamps, but some tests compare files directly,
which doesn't work if they aren't printing out the same
timestamps (like the comm tests).
- The above required yet another change to the writer API to
network_time to methods.
- Renamed ASCII logger "header" options to "meta".
- Fixes#763 "Escape # when first character in log file line".
All btests pass for me on Linux FC15. Will try MacOS next.
The output on stderr for this test is the results of many backgrounded
"echo" commands, one for each rotation, so the order in which they
occur may be subject to OS process scheduling and can't be relied upon
* robin/topic/writer-info:
Extending the log writer DoInit() API.
Reworking log writer API to make it easier to pass additional information to a writer's initialization method.
Conflicts:
src/logging/WriterBackend.cc
src/logging/WriterBackend.h
src/logging/WriterFrontend.cc
Generally tried to make them more reliable and execute quicker.
They all now load the listen script as a trick to make sure input
sources are fully read, but also terminate() at appropriate times
so that they don't take more time than needed. They're also all
serialized with the 'comm' group so listening on a port doesn't
interfere with the communication tests.
Using the default scripts, the events from RemoteSerializer::LogStats()
were attempting to use the logging framework after logging/threading
had been terminated which never worked right and sometimes caused
crashes with "fatal error: cannot lock mutex".
Also made communication log baseline test pass more reliably.
We now pass in a Info struct that contains:
- the path name (as before)
- the rotation interval
- the log_rotate_base_time in seconds
- a table of key/value pairs with further configuration options.
To fill the table, log filters have a new field "config: table[string]
of strings". This gives a way to pass arbitrary values from
script-land to writers. Interpretation is left up to the writer.
Also splits calc_next_rotate() into two functions, one of which is
thread-safe and can be used with the log_rotate_base_time value from
DoInit().
Includes also updates to the None writer:
- It gets its own script writers/none.bro.
- New bool option LogNone::debug to enable debug output. It then
prints out all the values passed to DoInit(). That's used by a
btest test to ensure the new DoInit() values are right.
- Fixed a bug that prevented Bro from terminating..
(scripts.base.frameworks.logging.rotate-custom currently fails.
Haven't yet investigated why.)
This lead to hanging bro's because pclose apparently can wait for eternity
if things go wrong. And there probably are a couple of other problems with this approach.
Adjust twotables testcase - now it is faster. Shorten the output
-- because of threading, the results did not always come out in
the same order (it depends on which thread manages to sneak in
the results into the queue earlier).
