Commit: 9112fb367b
Introduced the idea of setting the socket buffer
send/receive sizes. BSD's in general have the fun
issue of not allowing nearly as large as a size as
linux. Since the above commit was developed on linux
and not run on bsd it was never tested. Modify the
codebase to use the backoff setsockopt that we have
in the code base and use the returned values to allow
us to notice what was set and respond appropriately.
Signed-off-by: Donald Sharp <sharpd@nvidia.com>
The output buffer vty->obuf is a linked list where
each element is of 4KB.
Currently, when a huge sh command like <show ip route json>
is executed on a large scale, all the vty_outs are
processed and the entire data is accumulated.
After the entire vty execution, vtysh_flush proceeses
and puts this data in the socket (131KB at a time).
Problem here is the memory spike for such heavy duty
show commands.
The fix here is to chunkify the output on VTY shell by
flushing it intermediately for every 128 KB of output
accumulated and free the memory allocated for the buffer data.
This way, we achieve ~25-30% reduction in the memory spike.
Fixes: #16498
Note: This is a continuation of MR #16498
Signed-off-by: Srujana <skanchisamud@nvidia.com>
Signed-off-by: Rajasekar Raja <rajasekarr@nvidia.com>
Correct FRR startup counts on a daemon's vty socket to be open when the
parent process exits. The parent process waits for `frr_check_detach()`
to be called by the child before exiting. The problem is when the
`FRR_MANUAL_VTY_START` flag is set the vty socket was not opened but
`frr_check_detach()` was called anyway.
Instead add a bool option for `frr_check_detach()` to be called when the
socket is opened with `frr_vty_serv_start()`, and do so when "manually"
calling said function (i.e., when FRR_MANUAL_VTY_START is set).
The `FRR_MANUAL_VTY_START` flag is only set by mgmtd. The reason we
wait to open the vty socket is so that mgmtd can parse the various
daemon specific config files it has taken over, after the event loop has
started, but before we receive any possible new config from `vtysh`.
fixes#16362
Signed-off-by: Christian Hopps <chopps@labn.net>
This operation basically implements support for RESTCONF operations. It
receives an xpath and a data tree in JSON/XML format, instead of a list
of (xpath, value) tuples as required by the current protobuf interface.
Signed-off-by: Igor Ryzhov <iryzhov@nfware.com>
Make errors look the same way as in regular non-mgmtd vty. We don't need
to show information about some internal request names.
Before:
```
ERROR: SET_CONFIG request failed, Error: YANG error(s):
Path: Data location "/frr-affinity-map:lib/affinity-maps/affinity-map[name='a']".
Error: Unique data leaf(s) "value" not satisfied in "/frr-affinity-map:lib/affinity-maps/affinity-map[name='b']" and "/frr-affinity-map:lib/affinity-maps/affinity-map[name='a']".
```
After:
```
% Configuration failed.
YANG error(s):
Path: Data location "/frr-affinity-map:lib/affinity-maps/affinity-map[name='b']".
Error: Unique data leaf(s) "value" not satisfied in "/frr-affinity-map:lib/affinity-maps/affinity-map[name='a']" and "/frr-affinity-map:lib/affinity-maps/affinity-map[name='b']".
```
Signed-off-by: Igor Ryzhov <iryzhov@nfware.com>
Use consistent `e_somepath` names for expanded versions of `somepath`.
Also remove all paths from `config.h` and put them into
`lib/config_paths.h` - this is to make more obvious when someone is
doing something probably not quite properly structured.
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
Like in RESTCONF GET request and NETCONF get-data request, make it
possible to request state-only, config-only, or all data.
Signed-off-by: Igor Ryzhov <iryzhov@nfware.com>
Currently it's the same as get-tree request for the backend, but it is
going to be expanded in the following commits.
Signed-off-by: Igor Ryzhov <iryzhov@nfware.com>
Replace operation removes the current data node configuration and sets
the provided value. As current northbound code works only with one
xpath at a time, the operation only makes sense to clear the config of
a container without deleting it itself. However, the next step is to
allow passing JSON-encoded complex values to northbound operations which
will make replace operation much more useful.
Signed-off-by: Igor Ryzhov <iryzhov@nfware.com>
Currently, there's a single operation type which doesn't return error
if the object doesn't exists. To be compatible with NETCONF/RESTCONF,
we should support differentiate between DELETE (fails when object
doesn't exist) and REMOVE (doesn't fail if the object doesn't exist).
Signed-off-by: Igor Ryzhov <iryzhov@nfware.com>
Currently, there's no difference between CREATE and MODIFY operations.
To be compatible with NETCONF/RESTCONF, add new CREATE_EXCL operation
that throws an error if the configuration data already exists.
Signed-off-by: Igor Ryzhov <iryzhov@nfware.com>
Currently, nb_operation enum means two different things - edit operation
type (frontend part), and callback type (backend part). These types
overlap, but they are not identical. We need to add more operation
types to support NETCONF/RESTCONF integration, so it's better to have
separate enums to identify different entities.
Signed-off-by: Igor Ryzhov <iryzhov@nfware.com>
Add configure.ac tests for libyang functions, if not present supply the
functionality ourselves in yang.[ch]
Signed-off-by: Christian Hopps <chopps@labn.net>
`nb_cli_apply_changes` can be called with base xpath which should be
prepended to xpaths of every change in a transaction. This base xpath is
respected by regular northbound CLI but not by mgmtd. This commit fixes
the problem.
Signed-off-by: Igor Ryzhov <iryzhov@nfware.com>
Modify empty json object to take input obj
instead of allocating always one.
There are situation where in error condition or no data
case print empty json (`{}`) with already allocated
Signed-off-by: Chirag Shah <chirag@nvidia.com>
Also:
- replace all /* fallthrough */ comments with portable fallthrough;
pseudo keyword to accomodate both gcc and clang
- add missing break; statements as required by older versions of gcc
- cleanup some code to remove unnecessary fallthrough
Signed-off-by: Igor Ryzhov <iryzhov@nfware.com>
When the configuration node is entered in file-lock mode, candidate
and running datastores are locked. Any configuration change is followed
by an implicit commit which leads to a crash of mgmtd, because double
lock is prohibited by an assert. When working in file-lock mode, we
shouldn't do implicit commits which is disabled by allowing pending
configuration changes.
Signed-off-by: Igor Ryzhov <iryzhov@nfware.com>
Coverity is complaining that vty->state could be VTY_PASSFD here. It
can't, it really shouldn't, and if it actually is then something went
seriously wrong somewhere earlier so assert()ing out is the best thing
to do.
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
The lock/unlocks are being done short-circuit so they are never pending;
however, the handling of the unlock notification was always resuming the command
if pending was set. In all cases pending is set for another command. For example
implicit commit locks then when notified its done unlocks which was clearing the
set-config pending flag and resuming that command incorrectly.
Signed-off-by: Christian Hopps <chopps@labn.net>
Move away from things like "lock if not locked" type code, require the
user has locked prior to geting to that point.
For now we warn if we are taking a lock we already had; however, this
should really be a failure point.
New requirements:
SETCFG -
not implicit commit - requires user has locked candidate DS and they
must unlock after
implicit commit - requires user has locked candidate and running DS
both locks will be unlocked on reply to the SETCFG
COMMITCFG -
requires user has locked candidate and running DS and they must unlock
after
rollback - this code now get both locks and then does an unlock and
early return thing on the adapter side. It needs to be un-special
cased in follow up work that would also include tests for this
functionality.
Signed-off-by: Christian Hopps <chopps@labn.net>
- log names of datastores not numbers
- improve logging for mgmt_msg_read
- Rather than use a bool, instead store the pending const string name of
the command being run that has postponed the CLI. This adds some nice
information to the logging when enabled.
Signed-off-by: Christian Hopps <chopps@labn.net>
Avoid recursion into vty_close() when being notified of a session closure that
happened inside vty_close().
If a vty is closed with outstanding config yet to be commited
issue a warning that it is being lost.
Signed-off-by: Christian Hopps <chopps@labn.net>
Remove an obfuscating abstraction layer whose existence was entirely
based on using a uintptr_t rather than a pointer to an declared-only struct.
As the code is no longer using a global FE "client context", and instead
create client objects, rename the structure and it's uses to reflect this.
Change init "params" structure into a single use callbacks one.
Signed-off-by: Christian Hopps <chopps@labn.net>
