openstack/swift - swift - OpenDev: Free Software Needs Free Tools

Two improvements: first, document that the container-sync process
connects to the remote cluster's proxy server, so outbound
connectivity is required.
Second, rewrite the behind-the-scenes container-sync example and add
some ASCII-art diagrams.
Fixes bug 1068430.
Bonus fix of docstring in wsgi.py to squelch a sphinx warning.
Change-Id: I85bd56c2bd14431e13f7c57a43852777f14014fb

If the CONTRIBUTING[.md] file exists, Github will show a link to it to
anyone who files an issue on Github or opens a pull request on
Github. We don't want people to do that, so this file points people at
the OpenStack wiki page with instructions on how to contribute
properly. This should cut down on the number of pull requests and
Github issues that we then have to spend our valuable time ignoring.
See also <https://github.com/blog/1184-contributing-guidelines>.
Change-Id: Icd23b65c642c5ae748ca1f7f397e2c8d63173492

Change-Id: Ieef65f05e66b12347752c0e02648858e3242d8e7

Change-Id: I2ae277b6f435f56549560684697486b56d24d46d

Change-Id: I1ed6afe4336c238044174c93ae1e0078fbcef971

Change-Id: I0466fd2a462e94e5a1e0f88adf901de9f5ac131d

Bug 1079439
Change-Id: I1985a7176f34931ecb0f24c0289f18df5e934598

The determination of the client IP looked at the X-Cluster-Client-Ip
and X-Forwarded-For headers in the incoming HTTP request. This is
trivially spoofable by a malicious client, so there's no security
gained by having the check there.
Worse, having the check there provides a false sense of security to
cluster operators. It sounds like it's based on the client IP, so an
attacker would have to do IP spoofing to defeat it. However, it's
really just a shared secret, and there's already a secret key set
up. Basically, it looks like 2-factor auth (IP+key), but it's really
1-factor (key).
Now, the one case where this might provide some security is where the
Swift cluster is behind an external load balancer that strips off the
X-Cluster-Client-Ip and X-Forwarded-For headers and substitutes its
own. I don't think it's worth the tradeoff, hence this commit.
Fixes bug 1068420 for very small values of "fixes".
DocImpact
Change-Id: I2bef64c2e1e4df8a612a5531a35721202deb6964

When responding to a GET request for a manifest, it was intended that
the proxy server lazily fetch the pieces of the container
listing. That way, a single client request doesn't immediately turn
into a bunch of requests to backends. The additional requests should
only get made if the client is putting in the work of receiving the
object body.
However, commit 156f27c accidentally changed this so that all the
pieces of the container listing are eagerly fetched up-front. Better
yet, if an object has more than CONTAINER_LISTING_LIMIT (default
10,000) segments, the container listing is then fetched a second time,
albeit lazily, while streaming out the response.
This commit restores the laziness and adds tests for it.
Change-Id: I49840a7059e6f999ce19199ecb10cdb77358526b

We use a delta timeout value for timeouts under 30 days (in seconds)
since that is the limit which the memcached protocols will recognize a
timeout as a delta. Greater than 30 days and it interprets it as an
absolute time in seconds since the epoch.
This helps to address an often difficult-to-debug problem of time
drift between memcache clients and the memcache servers. Prior to this
change, if a client's time drifts behind the servers, short timeouts
run the danger of not being cached at all. If a client's time drifts
ahead of the servers, short timeouts run the danger of persisting too
long. Using delta's avoids this affect. For absolute timeouts 30 days
or more in the future small time drifts between clients and servers
are inconsequential.
See also bug 1076148 (https://bugs.launchpad.net/swift/+bug/1076148).
This also fixes incr and decr to handle timeout values in the same way
timeouts are handled for set operations.
Change-Id: Ie36dbcedfe9b4db9f77ed4ea9b70ff86c5773310
Signed-off-by: Peter Portante <peter.portante@redhat.com>

Change-Id: I889c5eecf860cc947763c5d79148419f9741a2c2
Signed-off-by: Peter Portante <peter.portante@redhat.com>

The auth_token middleware has moved, updating the doc.
Change-Id: I8133b876a53d61bf94169cf08c8b1fa6bbf9681a

roundrobin_datadirs was returning any .db file at any depth in the
accounts/containers structure. Since xfs corruption can cause such
files to appear in odd places at times (only happened on one drive of
ours so far, but still...), I've refactored this function to only
return .db files at the proper depth.
Change-Id: Id06ef6584941f8a572e286f69dfa3d96fe451355

The current catch_errors (ie without this patch) relinquishes control
before the underlying middleware/app has been evaluated. This results
in not catching errors in the stack when they occur in either the
start_response or in generating the first chunk sent to the client of
the underlying stack.
Change-Id: Iecd21e4fc7e30fa20239d011f69216354b50baf1

This set of changes reworks the DiskFile class to remove the "extension"
parameter from the put() method, offering the new put_metadata() method with
an optional tombstone keyword boolean, and changes the mkstemp method to only
return the file descriptor.
Reviewing the code it was found that the temporary file name created as a
result of calling DiskFile.mkstemp() was never used by the caller, but the
caller was responsible for passing it back to the DiskFile.put() method. That
seems like too much information is exposed to the caller, when all the caller
requires is the file descriptor to write data into it.
Upon further review, the mkstemp() method was used in three places: PUT, POST
and DELETE method handling. Of those three cases, only PUT requires the file
descriptor, since it is responsible for writing the object contents. For POST
and DELETE, DiskFile only needs to associate metadata with the correct file
name. We abstract the pattern that those two use (once we also refactor the
code to move the fetch of the delete-at metadata, and subsequent
delete-at-update initiation, from under the mkstemp context) by adding the new
put_metadata() method.
As a result, the DiskFile class is then free to do whatever file system
operations it must to meet the API, without the caller having to know more
than just how to write data to a file descriptor. Note that DiskFile itself
key'd off of the '.ts' and '.meta' extensions for its operations, and for that
to work properly, the caller had to know to use those correctly. With this
change, the caller has no knowledge of how the file system is being used to
accomplish data and metadata storage.
See also Question 213796 at:
 https://answers.launchpad.net/swift/+question/213796
Change-Id: I267f68e64391ba627b2a13682393bec62600159d
Signed-off-by: Peter Portante <peter.portante@redhat.com>

Only re-chain response iter if start_response hasn't been called yet.
Change-Id: I9670f49419104fb8f949e6edc12d9047d1289cde

Every request of container and object will invoke account_info() or
container_info() to query the meta of account and container. The meta
will be cached in memcache with the key 'account/{$account}' or
'container/{$container}', So, if any request to update account and
container, we should delete the cache. But in the cache deletion of
account, it use the wrong key 'account/v1/{$account}'. This could lead
to inconsistency of account meta.
Change-Id: Ied116a58a2d5866ac76d75ae50f21277d66e5755

There are 3 sections in there, all useless.
Section 1 tells you how to install Swift packages from the swift-core
PPA. However, the latest version there is ancient.
Section 2 tells you how to build your own Swift packages. However, it
talks about getting the source code from the "debian" branch in bzr,
which is obviously really old.
Section 3 tells you how to take the packages from section 2 and
install them. This isn't too out-of-date, but since section 2 doesn't
work any more, section 3 is useless.
Since stale docs are worse than no docs, there's no current
information in this document, and bringing it up-to-date requires a
whole pile of work, I've chosen to delete it entirely.
Also pulled out a couple references to the PPA elsewhere.
Fixes bug 917385.
Fixes bug 1026145.
Change-Id: I510bd8619531fe110419e5488bd20d3602868d66

Most of the test files set the HASH_PATH_SUFFIX so you can run the test
file stand alone. This change made it easier for me to run specific
proxy tests separately.
Change-Id: I87d70367dac7f240a2b6779649f8a02cf324ae0f

The proxy_logging middleware was asserting that the response contained
either a Content-Length header or a Transfer-Encoding header. If not,
it would either add one (if app_iter was a list) or blow up
(otherwise). This blowing up is observable on a GET request to a
manifest object that references more than
swift.common.constraints.CONTAINER_LISTING_LIMIT segments.
If a response makes it up to eventlet.wsgi without a Content-Length
header, then a "Transfer-Encoding: chunked" header is automatically
stuffed into the response by eventlet. Therefore, it's not an error
for a response to not have a Content-Length header, and proxy_logging
should just let it happen.
Fixes bug 1078113.
Change-Id: I3751a8ae14dc68bab546f2746b61267a5115e252

Change-Id: Iab43ac342faf3722ccab1ed034f08752dbbfaae1

I know it's just TempAuth, but bug #959953 just caught my eye as
something interesting to solve.
This does a best guess on the storage URL to return for a given
request. It allows $HOST to be used in the storage URL configuration,
where $HOST will resolve to scheme://host:port. It bases the scheme
on how the server is running or on storage_url_scheme if set. The
host:port comes from the request's Host header if it exists, and
falls back to the WSGI SERVER_NAME:SERVER_PORT otherwise.
Fixes: bug #959953
DocImpact
Change-Id: Ia494bcb99a04490911ee8d2cb8b12a94e77820c5

Change-Id: I3f0db3b70ea5df323597b556682c2d1fb28afe53

Before, swift-bench would create a string of the right size and pass
that to [direct_]put_object. Uploading 5 GB objects w/concurrency of
4? Hope you've got a spare 20 GB of RAM in that machine.
Now it lazily generates chunks (of size 64 KiB) to send. In my limited
testing, this keeps RAM usage low while not impacting speed.
There's backwards-compatibility code in
direct_client.direct_put_object for any callers who are still passing
in strings. There aren't any such in Swift's codebase, though. All one
of those callers now pass an iterable.
Fixes bug 911925.
Change-Id: I40669fc00c7fb6699d8fb514e1be9b69436eca42

This should fix up the few remaining API regressions from swob, by giving all
responses a default content-type of "text/html; charset=UTF-8".
Change-Id: Ib3b859e63c5e8c4af6f7fe535947402b5df57399

It is already the case that a DELETE of a deleted object results in a
404, and GET/HEAD/POSTs to both expired and deleted objects result in
404s. However, a DELETE of an expired object resulted in a 202. This
change makes it consistent with the other verbs.
Fixes bug 1076245.
Change-Id: I793e62d72461a4fb9fb3404e10658ddcc4c3a7a6

Set a utf-8 charset on default text/html responses.
Change-Id: Ic5f661bd475dca83763d4b55419ad031279e3ba1

If the Host header is available, use it for making relative URLs absolute.
Otherwise, continue using SERVER_NAME.
Change-Id: Ifc028264ad1b122a2d5dff9d5528cb369090429f

Change-Id: I72448cd8578b53a658f151354a26066772e4b9ec

These bug fixes are lumped together because they all caused problems
with the object expirer doing its job.
There was a bug with the internal client doing listings that happened
to run across a Unicode object name for use as a marker.
There was a bug with the object expirer not utf8 encoding object
names it got from json listings, causing deletes to fail.
There was a bug with the object expirer url quoting object names when
calling the internal client's make_request, when make_request already
handles that.
Change-Id: I29fdd351fd60c8e63874b44d604c5fdff35169d4

The object server knows the size of the file on disk already since
it's counted the bytes as they were written out. There's no need to go
ask the filesystem.
This also lets object PUTs work when the filesystem is some crazy,
non-POSIX thing that doesn't always give the right answers from
fstat(). See bug 904723 for a case when this happened.
Fixes bug 904723.
Change-Id: Ibdf44ea991e0cc4f9e0997cc9ff566cb0eac5332