up until now there has been a very annoying bug with get, we would
get halting behavior. I'm not 100% sure this commit fixes it,
but it should. It certainly fixes others found in the process of
digging into the get / tar extractor code. (wish we could repro
the bug reliably enough to make a test case).
This is a much cleaner tar writer. the ad-hoc, error-prone synch
for the tar reader is gone (with i believe was incorrect). it is
replaced with a simple pipe and bufio. The tar logic is now in
tar.Writer, which writes unixfs dag nodes into a tar archive (no
need for synch here). And get's reader is constructed with DagArchive
which sets up the pipe + bufio.
NOTE: this commit also changes this behavior of `get`:
When retrieving a single file, if the file exists, get would fail.
this emulated the behavior of wget by default, which (without opts)
does not overwrite if the file is there. This change makes get
fail if the file is available locally. This seems more intuitive to
me as expected from a unix tool-- though perhaps it should be
discussed more before adopting.
Everything seems to work fine, and i have not been able to reproduce
the get halt bug.
License: MIT
Signed-off-by: Juan Batiz-Benet <juan@benet.ai>
this PR greatly speeds up providing and add.
(1) Instead of idling workers, we move to a ratelimiter-based worker.
We put this max at 512, so that means _up to_ 512 goroutines. This
is very small load on the node, as each worker is providing to the
dht, which means mostly waiting. It DOES put a large load on the DHT.
but i want to try this out for a while and see if it's a problem.
We can decide later if it is a problem for the network (nothing
stops anyone from re-compiling, but the defaults of course matter).
(2) We add a buffer size for provideKeys, which means that we block
the add process much less. this is a very cheap buffer, as it only
stores keys (it may be even cheaper with a lock + ring buffer
instead of a channel...). This makes add blazing fast-- it was being
rate limited by providing. Add should not be ratelimited by providing
(much, if any) as the user wants to just store the stuff in the local
node's repo. This buffer is initially set to 4096, which means:
4096 * keysize (~258 bytes + go overhead) ~ 1-1.5MB
this buffer only last a few sec to mins, and is an ok thing to do
for the sake of very fast adds. (this could be a configurable
paramter, certainly for low-mem footprint use cases). At the moment
this is not much, compared to block sizes.
(3) We make the providing EventBegin() + Done(), so that we can
track how long a provide takes, and we can remove workers as they
finish in bsdash and similar tools.
License: MIT
Signed-off-by: Juan Batiz-Benet <juan@benet.ai>
Fedora/RedHat distros comply with US patent law and remove this curve,
which makes it impossible to run ipfs with distro provided Golang.
License: MIT
Signed-off-by: Pavol Rusnak <stick@gk2.sk>
This patch update fixes a problem we introduced in 0.3.6 and did not
catch: the webui failed to work with out-of-the-box CORS configs.
License: MIT
Signed-off-by: Juan Batiz-Benet <juan@benet.ai>
need to do it this way to avoid VERY confusing situations where
the user would change the API port (to another port, or maybe even
to :0). this way things dont break on the user, and by default,
users only need to change the API address and things should still
"just work"
License: MIT
Signed-off-by: Juan Batiz-Benet <juan@benet.ai>
ServeOptions take the node and muxer, they should get the listener
too as sometimes they need to operate on the listener address.
License: MIT
Signed-off-by: Juan Batiz-Benet <juan@benet.ai>
> ipfs add a b c
added Qmbvkmk9LFsGneteXk3G7YLqtLVME566ho6ibaQZZVHaC9 a
added QmR9pC5uCF3UExca8RSrCVL8eKv7nHMpATzbEQkAHpXmVM b
added QmetGxZTgo8tYAKQH1KLsY13MxqeVHbxYVmvzBzJAKU6Z7 c
added QmXg3WHLcjnz4ejeYF6FKVBkb4m1oKjQmF5fEWL9M1uQF3
> ipfs ls QmXg3WHLcjnz4ejeYF6FKVBkb4m1oKjQmF5fEWL9M1uQF3
Qmbvkmk9LFsGneteXk3G7YLqtLVME566ho6ibaQZZVHaC9 10 a
QmR9pC5uCF3UExca8RSrCVL8eKv7nHMpATzbEQkAHpXmVM 10 b
QmetGxZTgo8tYAKQH1KLsY13MxqeVHbxYVmvzBzJAKU6Z7 10 c
License: MIT
Signed-off-by: Juan Batiz-Benet <juan@benet.ai>
this commit changes the behavior of ipfs add -w:
- it makes it able to work with ipfs add -r <dir>
- instead of hacking around the add, we simply just add a wrapper
directory around the whole result of the add. this means that
ipfs add -w calls will output _two_ lines, but this is actually
more correct than outputting one line, as two objects were added.
this _may_ break scripts out there which expect the output to
look a certain way. we should consider whether the old output is
more _useful_ (even if less in-line with the model.)
License: MIT
Signed-off-by: Juan Batiz-Benet <juan@benet.ai>
unmounting wasn't happening, mostly because of a recent bug in
goprocess.SetTeardown. This commit bumps up some messages to
log.Warnings, as users may want to see them, and makes sure to
Unmount when a node shuts down.
License: MIT
Signed-off-by: Juan Batiz-Benet <juan@benet.ai>
this commit introduces more serious CORS tests that check
status response codes, and run real HTTP requests.
License: MIT
Signed-off-by: Juan Batiz-Benet <juan@benet.ai>
this commit makes the API handler short circuit the request if the
CORS headers say its not allowed. (the CORS handler only sets the
headers, but does not short-circuit)
It also makes the handler respect the referer again. See security
discussion at https://github.com/ipfs/go-ipfs/issues/1532
License: MIT
Signed-off-by: Juan Batiz-Benet <juan@benet.ai>
this commit adds the ability to specify arbitrary HTTP headers
for either the Gateway or the API. simply set the desired headers
on the config:
ipfs config --json API.HTTPHeaders.X-MyHdr '["meow :)"]'
ipfs config --json Gateway.HTTPHeaders.X-MyHdr '["meow :)"]'
License: MIT
Signed-off-by: Juan Batiz-Benet <juan@benet.ai>
