* Correctly retry and back off if we get a HTTPerror response
* Refactor request sending to have better excpetions
MatrixFederationHttpClient blindly reraised exceptions to the caller
without differentiating "expected" failures (e.g. connection timeouts
etc) versus more severe problems (e.g. programming errors).
This commit adds a RequestSendFailed exception that is raised when
"expected" failures happen, allowing the TransactionQueue to log them as
warnings while allowing us to log other exceptions as actual exceptions.
Broadly three things here:
* disable W504 which seems a bit whacko
* remove a bunch of `as e` expressions from exception handlers that don't use
them
* use `r""` for strings which include backslashes
Also, we don't use pep8 any more, so we can get rid of the duplicate config
there.
Wrap calls to deferToThread() in a thing which uses a child logcontext to
attribute CPU usage to the right request.
While we're in the area, remove the logcontext_tracer stuff, which is never
used, and afaik doesn't work.
Fixes#4064
Synapse doesn’t allow for media resources to be played directly from
Chrome. It is a problem for users on other networks (e.g. IRC)
communicating with Matrix users through a gateway. The gateway sends
them the raw URL for the resource when a Matrix user uploads a video
and the video cannot be played directly in Chrome using that URL.
Chrome argues it is not authorized to play the video because of the
Content Security Policy. Chrome checks for the "media-src" policy which
is missing, and defauts to the "default-src" policy which is "none".
As Synapse already sends "object-src: 'self'" I thought it wouldn’t be
a problem to add "media-src: 'self'" to the CSP to fix this problem.
ExpiringCache required that `start()` be called before it would actually
start expiring entries. A number of places didn't do that.
This PR removes `start` from ExpiringCache, and automatically starts
backround reaping process on creation instead.
This commit replaces SynapseError.from_http_response_exception with
HttpResponseException.to_synapse_error.
The new method actually returns a ProxiedRequestError, which allows us to pass
through additional metadata from the API call.
It turns out that looping_call does check the deferred returned by its
callback, and (at least in the case of client_ips), we were relying on this,
and I broke it in #3604.
Update run_as_background_process to return the deferred, and make sure we
return it to clock.looping_call.
parse_integer and parse_string can take a request and raise errors
in case we have wrong or missing params.
This PR tries to use them more to deduplicate some code and make it
better readable
(instead of everywhere that writes a response. Or rather, the subset of places
which write responses where we haven't forgotten it).
This also means that we don't have to have the mysterious version_string
attribute in anything with a request handler.
Unfortunately it does mean that we have to pass the version string wherever we
instantiate a SynapseSite, which has been c&ped 150 times, but that is code
that ought to be cleaned up anyway really.
While I was going through uses of preserve_fn for other PRs, I converted places
which only use the wrapped function once to use run_in_background, to avoid
creating the function object.
There were a bunch of places where we fire off a process to happen in the
background, but don't have any exception handling on it - instead relying on
the unhandled error being logged when the relevent deferred gets
garbage-collected.
This is unsatisfactory for a number of reasons:
- logging on garbage collection is best-effort and may happen some time after
the error, if at all
- it can be hard to figure out where the error actually happened.
- it is logged as a scary CRITICAL error which (a) I always forget to grep for
and (b) it's not really CRITICAL if a background process we don't care about
fails.
So this is an attempt to add exception handling to everything we fire off into
the background.
The old style raise is invalid syntax in python3. As noted in the docs,
this adds one more frame in the traceback, but I think this is
acceptable:
<ipython-input-7-bcc5cba3de3f> in <module>()
16 except:
17 pass
---> 18 six.reraise(*x)
/usr/lib/python3.6/site-packages/six.py in reraise(tp, value, tb)
691 if value.__traceback__ is not tb:
692 raise value.with_traceback(tb)
--> 693 raise value
694 finally:
695 value = None
<ipython-input-7-bcc5cba3de3f> in <module>()
9
10 try:
---> 11 x()
12 except:
13 x = sys.exc_info()
Also note that this uses six, which is not formally a dependency yet,
but is included indirectly since most packages depend on it.
Signed-off-by: Adrian Tschira <nota@notafile.com>
Add federation_domain_whitelist
gives a way to restrict which domains your HS is allowed to federate with.
useful mainly for gracefully preventing a private but internet-connected HS from trying to federate to the wider public Matrix network
Create the url_cache index on local_media_repository as a background update, so
that we can detect whether we are on sqlite or not and create a partial or
complete index accordingly.
To avoid running the cleanup job before we have built the index, add a bailout
which will defer the cleanup if the bg updates are still running.
Fixes https://github.com/matrix-org/synapse/issues/2572.
in commit 0e28281a the code for thumbnailing got refactored and the
renaming of this variables was not done correctly.
Signed-Off-by: Matthias Kesler <krombel@krombel.de>
When we proxy a media request to a remote server, add a query-param, which will
tell the remote server to 404 if it doesn't recognise the server_name.
This should fix a routing loop where the server keeps forwarding back to
itself.
Also improves the error handling on remote media fetches, so that we don't
always return a rather obscure 502.
This includes:
- Splitting out methods of a class into stand alone functions, to make
them easier to test.
- Adding unit tests to split out functions, testing HTML -> preview.
- Handle the fact that elements in lxml may have tail text.
Using XPath is slow on some machines (for unknown reasons), so use a
different approach to get a list of text nodes.
Try to generate a summary that respect paragraph and then word
boundaries, adding ellipses when appropriate.
The existing content can still be downloaded. The last upload to the
matrix.org server was in January 2015, so it is probably safe to remove
the upload API.
Always set the config key with an empty list, even if a list isn't specified.
This means that the codepaths are the same for both the empty list and
for a missing key. Since the behaviour is the same for both cases this
makes the code somewhat easier to reason about.
This is so that a single MediaRepository can be shared across all
resources, rather than having a "copy" per resource.
In particular this allows us to guard against both the thumbnail and
download resource triggering a download of remote content at the same
time.
Add url_preview_ip_range_blacklist to let admins specify internal IP ranges that must not be spidered.
Add url_preview_url_blacklist to let admins specify URL patterns that must not be spidered.
Implement a custom SpiderEndpoint and associated support classes to implement url_preview_ip_range_blacklist
Add commentary and generally address PR feedback
Introduce a User object
I'm sick of passing around more and more things as tuple items around
the whole world, and needing to edit every call site every time there is
more information about a user. So pass them around together as an
object.
This object has incredibly poorly named fields because we have a
convention that `user` indicates a UserID object, and `user_id`
indicates a string. I tried to clean up the whole repo to fix this, but
gave up. So instead, I introduce a second convention. A user_object is a
User, and a user_id_object is a UserId. I may have cried a little bit.
This tracks data about the entity which made the request. This is
instead of passing around a tuple, which requires call-site
modifications every time a new piece of optional context is passed
around.
I tried to introduce a User object. I gave up.
This follows the same flows-based flow as regular registration, but as
the only implemented flow has no requirements, it auto-succeeds. In the
future, other flows (e.g. captcha) may be required, so clients should
treat this like the regular registration flow choices.
Removes device_id and ClientInfo
device_id is never actually written, and the matrix.org DB has no
non-null entries for it. Right now, it's just cluttering up code.
This doesn't remove the columns from the database, because that's
fiddly.
Before, we returned a thumbnail that was at least as big (if possible)
as the requested size. Now, if we don't have a thumbnail of the given
size we generate (and persist) one of that size.
