| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|\ |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
We accept old 'send_name' handshake messages used by OTP-22 and older.
OTP-23 or OTP-24 nodes, not using epmd, may send the old message
to us if they do not know our OTP version.
Hence, we accept the old 'send_name' and the accompanying
'send_complement', but we do not send them ourself.
This was removed prematurely in OTP-25.0 by
commit 722219f211092e0c4b919f3ba4f91df42191c0e7.
Therefore versions 25.0 to 25.2.* are broken in this regard and may
reject connection setup from OTP-23 and 24 not using epmd.
|
|/ |
|
|
|
|
| |
Introduced the 'dist_listen' and 'hidden' options.
|
|\
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
rickard/global-fixes/25.0/OTP-17934
* rickard/global-fixes/24.3.3/OTP-17934:
[kernel] Introduce connect_all kernel parameter
[kernel] global fixes
[kernel] Monitor nodeup/nodedown directly from global
[kernel] Fix global group configuration
[erts,kernel] Connection ID information
kernel: Fix test case monitor_nodess_down_up
Guarantee nodedown before nodeup messages
|
| |\
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
rickard/global-fixes/24.3.3/OTP-17934
* rickard/global-fixes/23.3.4/OTP-17934:
[kernel] Introduce connect_all kernel parameter
[kernel] global fixes
[kernel] Monitor nodeup/nodedown directly from global
[kernel] Fix global group configuration
[erts,kernel] Connection ID information
kernel: Fix test case monitor_nodess_down_up
Guarantee nodedown before nodeup messages
|
| | |\
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
rickard/global-fixes/23.3.4/OTP-17934
* rickard/global-fixes/22.3.4/OTP-17934:
[kernel] Introduce connect_all kernel parameter
[kernel] global fixes
[kernel] Monitor nodeup/nodedown directly from global
[kernel] Fix global group configuration
[erts,kernel] Connection ID information
kernel: Fix test case monitor_nodess_down_up
Guarantee nodedown before nodeup messages
|
| | | |
| | | |
| | | |
| | | | |
as DFLAG_HANDSHAKE_23 is mandatory in OTP 25.
|
|\ \ \ \
| |/ / / |
|
| |\ \ \
| | | | |
| | | | |
| | | | | |
into sverker/24/distr-kill-lingering-connection
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
Prevent someone to sabotage existing connection,
by trying to setup a new connection, without knowing the cookie.
|
| | |/ / |
|
| | | | |
|
|\ \ \ \
| |/ / /
| | | |
| | | |
| | | | |
* maint:
Net tick intensity
|
| |\ \ \
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
rickard/net-tick-intensity/maint/ERIERL-732/OTP-17905
* rickard/net-tick-intensity/23.3.4/ERIERL-732/OTP-17905:
Net tick intensity
|
| | |\ \ \
| | | |/ /
| | |/| |
| | | | |
| | | | |
| | | | |
| | | | | |
rickard/net-tick-intensity/23.3.4/ERIERL-732/OTP-17905
* rickard/net-tick-intensity/22.3.4/ERIERL-732/OTP-17905:
Net tick intensity
|
| | | |/ |
|
| | | | |
|
| | | | |
|
| | | |
| | | |
| | | |
| | | | |
Co-authored-by: Björn Gustavsson <bgustavsson@gmail.com>
|
|\ \ \ \
| |/ / / |
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
The initial check for ipv4 vs ipv6 did not use the
erl_epmd interface to lookup the address which caused
invalid addresses to not be resolvable by erl_epmd callback
modules.
Closes #5334
|
| | | |
| | | |
| | | |
| | | | |
in handshake_we_started(), the connecting node.
|
|\ \ \ \
| |/ / /
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
* maint:
Update preloaded
Test differing cookies with dynamic node names
Implement set_cookie/1 and get_cookie/1
Clarify documentation of cookie handling
Clean up handshake Cookie handling
Test command line -setcookie Node Cookie
Read other nodes' cookies from command line
Read other nodes' cookies from command line
Test differing cookies
|
| | | | |
|
|/ / /
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
During distribution connect setup between two nodes, the nodes exchange
flags words encoding their capabilities. The connection setup is aborted
if one of the node lacks a capability that the other node requires.
In version 5 of the protocol (introduced in OTP R6), there are 32
possible capabilities (each encoded in a single bit). Since it seemed
that we would soon run out of capabilitiy bits, protocol version 6
(introduced in OTP 23) expanded the number of possible capability bits
to 64.
To avoid having to extend the number of capabilites yet again from 64
to 128 in some future release, this commit introduces a scheme to
allow reusing the bit numbers for capabilities that are now
mandatory. The scheme is described in lib/kernel/include/dist.hrl.
There are 10 mandatory capability bits as of OTP 25. This scheme will
allow reusing 9 of those capability bits as early as in OTP 27.
|
| | | |
|
| | | |
|
| | | |
|
|/ /
| |
| |
| |
| | |
We previously only allowed references with 3 number words. This has
now been increased to 5.
|
| | |
|
| | |
|
| | |
|
| | |
|
|\ \
| |/
| |
| |
| | |
* maint:
Update copyright year
|
| | |
|
|\ \
| |/
| |
| |
| | |
* maint:
net_kernel: improve performance of debugging functions, and fix race condition
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
net_kernel contains several function intended only for debugging. For example,
node_info/1,2 and nodes_info(). While undocumented, these functions provide
extremely valuable monitoring information.
This patch fixes severl issues. First, get_status call was monitoring node,
and not distribution process. It was a subject to race condition, when
'nodedown' message was never delivered. With erlang:monitor/2, race does
not happen. Also it allowed to remove unneeded flush_down from dist_util.
Second, nodes_info/0 was implemented sequentially, being quite slow. Now
it is concurrent enough to complete within a second even with thousands
nodes in a cluster.
Third, nodes_info/0 was traversing the same sys_dist table two times,
where one was enough.
Fourth, dialyzer specs are somewhat helpful too.
|
| | |
|
| | |
|
|/ |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The dist messages EXIT, EXIT2 and MONITOR_DOWN have been
updated with new versions that send the reason term as
part of the payload of the message instead of as part
of the control message.
This allows the decode of the reason to be done by the
receiving process instead of the dist entry which in turn
makes it possible for multiple decodes to be done in
parallel.
This change is done in order to make it easier to fragment
the potentially large payload of EXIT, EXIT2 and MONITOR_DOWN
into multiple distribution messages.
OTP-15611
|
| |
|
| |
|
|
|
|
|
|
|
| |
* Move the node name check earlier
* Check while still a list
* Export allowed check function
* Make allowed check function handle atom and list mixes
|
|
|
|
| |
OTP-14899
|
|\
| |
| |
| |
| | |
* sverker/send-tick/OTP-14681:
kernel: Fix and improve send_tick logic
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Problem 1: Tick was not sent to hidden when pending sends. Don't hidden
need the tick in order to reply?
Problem 2: Written packets (#tick.write) was increased even when tick
was suppressed due to pending sends.
Code rewritten to handle outgoing and incoming separate.
|