memcached crashes on setting first key #64
Comments
|
What's the commit hash you were running? A bug was fixed recently in this area. |
|
On Sun, Oct 18, 2015 at 09:10:04AM -0700, Avi Kivity wrote:
|
|
@avikivity It's commit 46fd389. Thanks! |
|
@gleb-cloudius when I close the output stream it is not jolly at all: diff --git a/apps/memcached/memcache.cc b/apps/memcached/memcache.cc
index 659c66b..f487c94 100644
--- a/apps/memcached/memcache.cc
+++ b/apps/memcached/memcache.cc
@@ -1303,7 +1303,9 @@ public:
auto conn = make_lw_shared<connection>(std::move(fd), addr, _cache, _system_stats);
do_until([conn] { return conn->_in.eof(); }, [this, conn] {
return conn->_proto.handle(conn->_in, conn->_out).then([conn] {
- return conn->_out.flush();
+ return conn->_out.flush().then([conn] {
+ return conn->_out.close();
+ });
});
});
}); |
|
On Sun, Oct 18, 2015 at 09:52:44AM -0700, 0xmjk wrote:
|
|
Hmm, I've tried it and tried also: diff --git a/apps/memcached/memcache.cc b/apps/memcached/memcache.cc
index 659c66b..ceca3e8 100644
--- a/apps/memcached/memcache.cc
+++ b/apps/memcached/memcache.cc
@@ -1303,7 +1303,7 @@ public:
auto conn = make_lw_shared<connection>(std::move(fd), addr, _cache, _system_stats);
do_until([conn] { return conn->_in.eof(); }, [this, conn] {
return conn->_proto.handle(conn->_in, conn->_out).then([conn] {
- return conn->_out.flush();
+ return conn->_out.close().finally([conn] {});
});
});
});Which should call flush() according to whats in the iostream-impl: template <typename CharType>
future<>
output_stream<CharType>::close() {
return flush().finally([this] {
if (_in_batch) {
return _in_batch.value().get_future();
} else {
return make_ready_future();
}
}).then([this] {
// report final exception as close error
if (_ex) {
std::rethrow_exception(_ex);
}
}).finally([this] {
return _fd.close();
});
}But it is still giving me a stacktrace that suggests that conn->_in has been cleared up already. |
|
On Sun, Oct 18, 2015 at 11:49:32AM -0700, 0xmjk wrote:
|
|
@tgrabiec ^^ |
|
I think I resolved it, but I don't understand why it works. This works @@ -1303,7 +1307,7 @@ public:
auto conn = make_lw_shared<connection>(std::move(fd), addr, _cache, _system_stats);
do_until([conn] { return conn->_in.eof(); }, [this, conn] {
return conn->_proto.handle(conn->_in, conn->_out).then([conn] {
- return conn->_out.flush();
+ return conn->_out.flush().then([conn] { conn->_out.close().finally([conn] { });});
});
});
});But this doesn't: @@ -1303,7 +1307,7 @@ public:
auto conn = make_lw_shared<connection>(std::move(fd), addr, _cache, _system_stats);
do_until([conn] { return conn->_in.eof(); }, [this, conn] {
return conn->_proto.handle(conn->_in, conn->_out).then([conn] {
- return conn->_out.flush();
+ return conn->_out.flush().then([conn] { return conn->_out.close().finally([conn] { });});
});
});
}); |
|
On Mon, Oct 19, 2015 at 08:32:18AM -0700, 0xmjk wrote:
Closing connection here is incorrect since this is processing loop and diff --git a/apps/memcached/memcache.cc b/apps/memcached/memcache.cc
index 659c66b..115aa0e 100644
--- a/apps/memcached/memcache.cc
+++ b/apps/memcached/memcache.cc
@@ -1305,6 +1305,8 @@ public:
return conn->_proto.handle(conn->_in, conn->_out).then([conn] {
return conn->_out.flush();
});
+ }).finally([conn] {
+ return conn->_out.close().finally([conn]{});
});
});
}).or_terminate(); |
|
It works and the tests are much happier http://pastebin.com/nwwRzfdx |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Hi,
I get the following error
when running a simple test:
I get quite a lot of errors when running ./test.py. (http://pastebin.com/d2dkePc0)
The above error seems related to a commit from couple days ago. However, I'm not sure if it is this or my setup - I'm running under VirtualBox and docker; my lscpu output:
And Dockerfile:
Thanks!
The text was updated successfully, but these errors were encountered: