OVS events plugin #1971

vmytnykx · 2016-09-30T16:39:04Z

The OVS events plugin monitors the link status of OVS connected interfaces, dispatches the values to collectd and send the notification whenever the link state change occurs in OVS DB.

The plugin consists of two parts:

OVS events plugin itself (ovs_events.c, uses OVS utils);
OVS utils (utils_ovs.c, utils_ovs.h). It provides API for the OVS events plugin (and future OVS plugin(s)) to subscribe to specific OVS DB table update event(s) and API to poll table data directly from OVS DB. The module implements part of the OVS DB communication routine specified by RFC7047 (https://tools.ietf.org/html/rfc7047). It includes:

Connecting/disconnecting to/from OVS DB (via TCP/UNIX);
Mechanism to subscribe to OVS DB table events like init/insert/modify/delete table rows;
API to send custom request (e.g. poll table data) and receive result;
Recovery connection mechanism with OVS DB;
Handling of ECHO request to verify the liveness of a database connection;
The OVS DB connection lost/establish notification;
Helpers functions.

vmytnykx · 2016-10-03T08:36:48Z

Hi @octo ,

The new OVS plugin depends on YAJL library which is already integrated into collectd. The plugin uses yajl_tree.h API to easily parse JSON string. Unfortunately, some of CI builds are failed because of the following error:

utils_ovs.h:69:28: fatal error: yajl/yajl_tree.h: No such file or directory

Looks like the problem occurs because of old YAJL lib version pre-installed on some of CI builds. The "yajl_tree.h" API are missing in YAJL v1.x but have been added in YAJL 2.x version (~5 years ago).

Failing CI builds:

pull-requests-build-on-precise-amd64-with-default-toolchain
pull-requests-build-on-precise-i386-with-default-toolchain
pull-requests-build-on-epel6-i386-with-default-toolchain
pull-requests-build-on-epel6-x86_64-with-default-toolchain

Is the YAJL lib version 1.x expected for builds above? Should it be changed to use more recent version of YAJL?

Thanks and Regards,
Volodymyr

octo · 2016-10-03T13:02:01Z

Hi @vmytnykx,

thank you very much for your patch! Precise and Wheezy often come with quite old library versions and should still work. In this case, the plugin should be disabled by the configure script. You can add something along these lines to configure.ac:

with_libyajl2="no"
# …
AC_CHECK_HEADERS(yajl/yajl_tree.h,
                 [with_libyajl2="yes"],
                 [with_libyajl2="no (yajl/yajl_tree.h not found)"])
# …
AC_CHECK_LIB(yajl, yajl_tree_parse,
             [with_libyajl2="yes"],
             [with_libyajl2="no (Symbol 'yajl_tree_parse' not found)"])

Then use the with_libyajl2 variable to control whether to build your new plugin.

Best regards,
—octo

octo

Hi @vmytnykx,

thanks for fixing the configure script! I started a review, but this is quite complex and will take a while to wrap my head around, sorry. I've left some comments that I'll publish right away, should you be blocked on me.

Best regards,
—octo

octo · 2016-10-04T19:17:34Z

src/utils_ovs.c

+  union {
+    struct sockaddr_in s_inet;
+    struct sockaddr_un s_unix;
+  } addr;


This is not portable, because struct sockaddr_in may be an "incomplete type", as is the case on FreeBSD.

I recommend to replace the union with:

struct sockaddr_storage addr;

That struct is guaranteed to be "large enough to accommodate all supported protocol-specific address structures".

octo · 2016-10-04T19:18:11Z

src/utils_ovs.c

+  int sock;
+  int domain;
+  int type;
+  int addr_size;


Please use a socklen_t for this.

octo · 2016-10-04T19:19:38Z

src/utils_ovs.c

+/* clang-format on */
+
+/* collectd headers */
+#include "common.h"


Please always include "collectd.h" as the first header.

octo · 2016-10-04T19:33:15Z

src/utils_ovs.c

+
+/* Check if POLL thread is still running. Returns
+ * 1 if running otherwise 0 is returned */
+static inline int ovs_db_poll_is_running(ovs_db_t *pdb) {


Here and below: please remove the inline – it's premature optimization.

The inline statement lead to some confusing effects, such as function pointers no longer comparing equal. On the other hand, compilers are very good at deciding which functions to inline. So we get likely no benefit but the potential for hard to track down bugs.

octo · 2016-10-04T19:33:36Z

src/utils_ovs.c

+
+/* Check if POLL thread is still running. Returns
+ * 1 if running otherwise 0 is returned */
+static inline int ovs_db_poll_is_running(ovs_db_t *pdb) {


Please consider changing the return value of this function to _Bool.

octo · 2016-10-04T19:43:22Z

src/utils_ovs.c

+  pthread_mutex_lock(&pdb->mutex);
+  for (ovs_callback_t *del_cb = pdb->remote_cb; pdb->remote_cb;
+       del_cb = pdb->remote_cb) {
+    pdb->remote_cb = pdb->remote_cb->next;


At your option: I find the following more intuitive:

pdb->remote_cb = del_cb->next;

octo · 2016-10-04T19:46:17Z

src/utils_ovs.c

+  pthread_mutex_lock(&pdb->mutex);
+  for (ovs_callback_t *cb = pdb->remote_cb; cb != NULL; cb = cb->next)
+    if (cb->uid == uid) {
+      pthread_mutex_unlock(&pdb->mutex);


I think cb needs to be protected by pdb->mutex, too, otherwise another thread could call ovs_db_callback_remove_all(), freeing the memory cb points to while the caller still performs work with cb. In other words, I think the caller needs to hold pdb->mutex when calling this function.

octo · 2016-10-04T19:54:31Z

src/utils_ovs.c

+    goto failure;
+
+  /* parse url depending on domain */
+  if ((nexttok = strtok_r(NULL, ":", &saveptr)) != NULL) {


Calling strtok_r() when configuring an AF_UNIX socket will break paths that contain a colon. If tmp_conn.domain == AF_UNIX, don't call strtok_r() again.

octo · 2016-10-04T20:07:47Z

src/utils_ovs.c

+      tmp_conn.addr.s_inet.sin_family = AF_UNIX;
+      sstrncpy(tmp_conn.addr.s_unix.sun_path, nexttok, strlen(nexttok) + 1);
+    } else {
+      /* <IP:PORT> */


This doesn't handle IPv6.

I'd recommend to implement a config option that takes either one or two arguments. The first argument is the address, the second argument an optional port, for example:

Address "127.0.0.1" "6640" Address "2001:DB8::c011:ec7d" "6640" Address "2001:DB8::c011:ec7d" "service-name" Address "unix:/path/to/socket"

You can then easily detect UNIX sockets: strncmp (arg1, "unix:", strlen("unix:")) == 0

Otherwise, call getaddrinfo() to get a struct sockaddr: getaddrinfo(arg1, arg2, hints, &res)

octo · 2016-10-04T20:20:59Z

src/collectd.conf.pod

+
+=over 4
+
+=item B<OvsDbServerUrl> I<server>


Please consider turning this into an option with two values, "node" and "service" (aka. "host" and "port"). That makes it much easier to support IPv6. See also my comments below.

On second thought, it might make sense to copy what some other plugins are doing. They use four config options:

Host

Controls what goes into the value_list_t.host field.

Address

IP address or hostname to connect to.

Port

Port number or service to use when connecting.

Socket

Path for the UNIX socket. If this option is set, Address and Port are ignored.

vmytnykx · 2016-10-07T08:43:35Z

Hi @octo ,

Thank you for reviewing the plugin implementation and your comments.

I've fixed all your comments and re-implement plugin configuration based on your suggestion. Also, IPv6 support has been added (automatically) and verified.

Thanks and Regards,
Volodymyr

maryamtahhan · 2016-10-07T14:21:57Z

src/ovs_events.c

+      }
+      DEBUG("name=%s, uuid=%s, ext_iface_id=%s, ext_vm_uuid=%s", ifinfo.name,
+            ifinfo.uuid, ifinfo.ext_iface_id, ifinfo.ext_vm_uuid);
+      ovs_events_link_status_submit(&ifinfo);


should you not check if send nofication is enabled at this point? then either dispatch a notification or submit a value?

NVM - this is for values from tables and not events...

maryamtahhan · 2016-10-09T12:15:03Z

src/ovs_events.c

+
+/* OVD DB reply callback. It parses reply, receives
+ * interface information and dispatches the info to
+ * collecd


octo

Hi @vmytnykx,

I've made some more headway. I'm not yet 100% through, but I wanted to share this because I don't know how much bandwidth I'll have this week.

Best regards,
—octo

octo · 2016-10-05T15:04:07Z

src/collectd.conf.pod

+
+=over 4
+
+=item B<OvsDbServerUrl> I<server>


On second thought, it might make sense to copy what some other plugins are doing. They use four config options:

Host

Controls what goes into the value_list_t.host field.

Address

IP address or hostname to connect to.

Port

Port number or service to use when connecting.

Socket

Path for the UNIX socket. If this option is set, Address and Port are ignored.

octo · 2016-10-08T06:18:20Z

src/utils_ovs.c

+
+  /* node cannot be unset */
+  if (node == NULL || strlen(node) == 0)
+    return (NULL);


This leaks pdb. You could move this check to before the calloc().

octo · 2016-10-08T06:43:37Z

src/utils_ovs.c

+  return pdb;
+
+failure:
+  if (pdb->sock)


Can't you call ovs_db_destroy(pdb) here?

octo · 2016-10-08T06:47:19Z

src/utils_ovs.c

+
+  if (cb) {
+    /* register result callback */
+    if ((new_cb = malloc(sizeof(ovs_callback_t))) == NULL)


Please use:

new_cb = calloc(1, sizeof(new_cb))

octo · 2016-10-08T06:59:12Z

src/utils_ovs.c

+  if (new_cb) {
+    /* destroy callback */
+    sem_destroy(&new_cb->result.sync);
+    ovs_db_callback_remove(pdb, new_cb);


That order seems dangerous; I'd remove the callback first before starting to freeing it.

Not sure about this as ovs_db_callback_remove() function actually does this. It takes mutex, remove the callback, frees it and release the mutex.

octo · 2016-10-10T07:19:42Z

src/ovs_events.c

+    if (new_buff == NULL)
+      goto failure;
+    buff = new_buff;
+    ret = ssnprintf(buff + offset, buff_size, option_fmt, iface->name);


The available buffer size is only buff_size - offset.

Thank for that :)

octo · 2016-10-10T07:22:55Z

src/ovs_events.c

+  }
+
+  /* allocate memory for OVS DB select params */
+  buff_size = offset + sizeof(params_fmt);


It would be more consistent with the above code to do:

buff_size = sizeof(params_fmt) + strlen(buff);

octo · 2016-10-10T07:23:32Z

src/ovs_events.c

+  }
+
+  /* allocate memory for OVS DB select params */
+  buff_size = offset + sizeof(params_fmt);


I'm not a huge fan of re-using variables like this. Please consider using char *params and size_t params_size instead.

octo · 2016-10-10T07:32:03Z

src/ovs_events.c

+#define OVS_EVENTS_EXT_IFACE_ID_SIZE 64
+#define OVS_EVENTS_EXT_VM_UUID_SIZE 64
+#define OVS_EVENTS_PLUGIN "ovs_events"
+#define OVS_EVENTS_CTX_LOCK                                                    \


Clever, I love it :)

octo · 2016-10-10T07:34:00Z

src/ovs_events.c

+          0)
+        OVS_EVENTS_CONFIG_ERROR(child->key);
+    } else if (strcasecmp("OvsDbAddress", child->key) == 0) {
+      if (child->values_num < 1) {


Please use cf_util_get_string_buffer() here, which will do all necessary checks and error reporting:

cf_util_get_string_buffer(child, ovs_events_ctx.config.ovs_db_node, sizeof(ovs_events_ctx.config.ovs_db_node));

vmytnykx · 2016-10-28T15:06:37Z

Hi @octo,

Thank you for the addition review comments. I've tried to address all of them, including

configuration update (UNIX socket and IP address configuration are separated now, which looks event better ) )
memory leaks, critical sections, init/destroy API improvement
others comments.

Also, I've replied to some of your comments which seems don't require any changes...

Best Regards,
Volodymyr

rubenk · 2016-11-19T13:17:32Z

src/collectd.conf.pod

+=head2 Plugin C<ovs_events>
+
+The I<ovs_events> plugin monitors the link status of OVS connected interfaces,
+dispatches the values to collectd and send the notification whenever the link


s/send/sends/

rubenk · 2016-11-19T13:18:54Z

src/collectd.conf.pod

+
+=item B<Address> I<node>
+
+The address of OVS DB server JSON-RPC interface used by the plugin. To enable


The address of the OVS DB server JSON-RPC interface

rubenk · 2016-11-19T13:20:10Z

src/collectd.conf.pod

+
+=item B<Interfaces> [I<ifname> ...]
+
+List of interface names to be monitored by this plugin. If this option is missed


is not specified or is empty

rubenk · 2016-11-19T13:22:55Z

src/ovs_events.c

+               .ovs_db_serv = "6640",      /* use default OVS DB service */
+               .ovs_db_unix = "",          /* UNIX path empty by default */
+               .ifaces = NULL},
+    .ovs_db_select_params = NULL,


there's no need to initialize static variables to their zero value.

rubenk · 2016-11-19T13:28:44Z

src/ovs_events.c

+                            "\"where\":[[\"name\",\"==\",\"%s\"]],"
+                            "\"columns\":[\"link_state\",\"external_ids\","
+                            "\"name\",\"_uuid\"]}";
+  const char default_opt[] = ",{\"op\":\"select\",\"table\":\"Interface\","


can you make these static?

rubenk · 2016-11-19T13:31:23Z

src/ovs_events.c

+    opt_buff = new_buff;
+    ret = ssnprintf(opt_buff + buff_off, buff_size - buff_off, option_fmt,
+                    iface->name);
+    if (ret < 0) {


rubenk · 2016-11-19T13:31:49Z

src/ovs_events.c

+  }
+  /* if no interfaces are configured, use default params */
+  if (opt_buff == NULL)
+    opt_buff = strdup(default_opt);


please check for failure.

rubenk · 2016-11-19T13:35:27Z

src/ovs_events.c

+  for (ovs_events_iface_list_t *iface = ovs_events_ctx.config.ifaces; iface;
+       iface = iface->next, buff_off += ret) {
+    /* allocate new buffer (format size + ifname len is good enough) */
+    buff_size += (sizeof(option_fmt) + strlen(iface->name));


I don't believe you need the outer braces.

rubenk · 2016-11-19T13:37:22Z

src/ovs_events.c

+
+/* Dispatch OVS interface link status event to collectd */
+static void
+ovs_events_dispatch_notification(const ovs_events_iface_info_t *ifinfo) {


please combine these two lines.

This is done by clang-format. I've addressed it anyway but I suppose it will be changed back during next clan-format running :)

rubenk · 2016-11-19T13:37:46Z

src/ovs_events.c

+
+/* Dispatch OVS interface link status value to collectd */
+static void
+ovs_events_link_status_submit(const ovs_events_iface_info_t *ifinfo) {


please combine these two lines.

rubenk · 2016-11-19T13:39:23Z

src/ovs_events.c

+
+  /* register update link status event if needed */
+  if (ovs_events_ctx.config.send_notification) {
+    ret = ovs_db_table_cb_register(pdb, tb_name, columns,


vmytnykx · 2016-11-21T17:47:12Z

Hi @rubenk, last commit addresses your comments. Thank you.

rubenk · 2016-11-21T17:57:53Z

README

+      installed.
+      Detailed instructions for installing and setting up Open vSwitch, see
+      OVS documentation.
+      <http://openvswitch.org/support/dist-docs/INSTALL.md.html>


Heh, that link gives a 404. Should this point to https://github.com/openvswitch/ovs/blob/master/INSTALL.rst instead?

Yeah, you are right! OvS has changed it (*.md -> *.rst, ) after the PR has been created. Thank you for discovering this :) I will fix it to the right one: http://openvswitch.org/support/dist-docs/INSTALL.rst.html.

rubenk · 2016-11-21T21:04:57Z

Thanks @vmytnykx. Personally I think this is good to be merged. @octo did you have any further comments?

vmytnykx · 2016-11-23T23:47:22Z

Hi @octo,

Do you have any farther comments regarding the changes? Sorry to ping you again but we have already implemented new plugin that rely on OVS utils API (utils_ovs.[ch]). Even, if we create new PR with the new changes based on that branch, it will also include #1971 changes, which will confuse the review process (also, rebasing will be required, etc., once current changes are merged). If you have some idea how to share the new code in this case, that would be appreciated too.

FYI, new changes: https://github.com/maryamtahhan/collectd/compare/feat_ovs_link...maryamtahhan:feat_ovs_stats?expand=1

Thanks and Regards,
Volodymyr

maryamtahhan · 2016-12-13T13:32:37Z

Hi @octo @rubenk
If there are no more review comments, can this be merged into master? we would like to enable this plugin in collectd in OPNFV and need it on the master branch
Thanks in advance
BR
Maryam

This plugin consists of two parts: - OVS link The implementation of the plugin itself, which uses OVS utils API to be able to monitor a link status of OVS connected interfaces and dispatch the values through collectd notification mechanism whenever the link state change occurs. - OVS utils This module implements the OVS DB communication routine specified by RFC7047. It includes: - Connecting/disconnecting to/from OVS DB (via TCP/UNIX); - Mechanism to subscribe to OVS DB table events like init/insert/modify/delete table rows; - API to send custom request and receive result; - Recovery connection mechanism with OVS DB; - Handling of ECHO request to verify the liveness of a database connection; - Helpers functions. Change-Id: Icac392bd1bd40f7dd156bfd2fc4ff08d9725a22f Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

- Implement read callback; - Add SendNotification option; - Update plugin documentation; - Clean-up. Change-Id: Ie645192498d234d47e1fe0272ec30e7c3d9c1774 Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

Extend OVS utils to notify OVS plugin about OVS DB connection lost. If the connection is lost, OVS link plugin will dispatch notification and print error message. OVS plugin treats status of all interfaces as UNKNOWN. Change-Id: I22fe3cb0740e0f4779a5c4f6b92e78f1ad9777a3 Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

Change-Id: Ic50891722074aec017b1329601a3bcbafb030ce0 Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

- Rename files Change-Id: Ic662d5d673c1c66b2057e4b35fa3cf664e92825e Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

- Interface meta data are added to dispatched values & notification. interface metadata: - UUID (always); - iface-id (if exists) - vm-uuid (if exists) - Fix read callback to poll actual data instead of reading cache; - Code clean up; Change-Id: I291cb190d31ae091c1d47ce0f5e9d439f6958d61 Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

- Fix Ubuntu compilation warnings. Change-Id: I7bb72ec0ca732b00c4169e0174d64407d99a2751 Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

Disable clang-format for diagram in utils_ovs.c. Change-Id: Idfb1510c996f5ebbdadc96663eb032537837502a Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

Change-Id: I731b633b5f78ed63b643574a336c59717e408a78 Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

If old YAJL version is installed on a system (ver < 2.0), the OVS plugin compilation will fail as it requires tree API to be supported in YAJL library. For this reason, it was decided to change 'configure' script to detect YAJL tree API also. Clean-up according updated collectd style: collectd#1931 collectd#1951 Change-Id: I90c82cdc9780ee8c0c9b794986662a39a5ab0011 Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

- Change OVS implementation to use suggested configuration: OvsDbAddress "127.0.0.1" "6640" OvsDbAddress "2001:DB8::c011:ec7d" "6640" OvsDbAddress "2001:DB8::c011:ec7d" "service-name" OvsDbAddress "unix:/path/to/socket" - Update documentation; - Change OVS utils to use getaddrinfo(); - Clean-up compilation warnings on FreeBSD system. - Add IPv6 support; Change-Id: I60ca700c15406c783b62ee52135266d67b60393a Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

If multiple interfaces are configured, only last one will be monitored (notifications aren't sent for all others). Change-Id: Ica918ef0557747d84677db739a64698d3756a675 Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

- Change configuration format to suggested one; - Fix init/destroy API; - Fix memory leaks; - Code-clean-up. Change-Id: I1ff94271b777c69f3d07a66f43dc10d034e71101 Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

If some of interface data is not available, the garbage information may be returned by ovs_events_get_iface_info() function. Add clean-up interface information structure into the function. Change-Id: Ia3d2bdfe31d0b3db81ad7ad773eca1df5d5f6f6a Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

Clean-up OVS events plugin based on PR comments. Change-Id: Ibd18924dd2a6f936d0ea83ed4eeb3b34ff8416b5 Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

Change-Id: I5bd4d7aabaf0ff66edd9c1cd75caf7f37448e00d Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

- Fix OVS documentation - Code clean-up Change-Id: I84fb003aea19f73381192f31935c79b51eaba1c9 Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

- the prototype of plugin_thread_create () has been changed in latest master. Change-Id: I492942442717c43cd6e6e73478248164b5083894 Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

The location of OVS sources in Malefile.am is `src/*` folder now. Change-Id: Iffe8a439faf531e74e0a752c3fb544369928cfb6 Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

vmytnykx · 2016-12-26T13:54:43Z

Hi @rubenk ,

I've done re-base second time because of the recent changes into build system (src/Makefile.am -> Makefile.am). So, now conflicts are resolved.

Since, there are no more comments, can this be merged to avoid further re-basing ?

Thanks & Regards,
Volodymyr

vmytnykx · 2016-12-26T14:31:13Z

Hi, @rubenk , the epel5 builds are failed because of mcelog error:

src/mcelog.c: In function 'socket_reinit':
src/mcelog.c:163: error: 'SOCK_CLOEXEC' undeclared (first use in this function)
src/mcelog.c:163: error: (Each undeclared identifier is reported only once
src/mcelog.c:163: error: for each function it appears in.)
src/mcelog.c:163: error: 'SOCK_NONBLOCK' undeclared (first use in this function)
make[1]: *** [src/mcelog.lo] Error 1

This failure isn't related to OVS changes. Based on this comment looks like it is expected, right?

Thanks,
Volodymyr

rubenk · 2016-12-31T11:47:55Z

Dear Volodymyr, yes, we've dropped support for EPEL5 but still need to adjust the CI.

Now merged, thanks for all your work!

octo added the Pending contributor action label Sep 30, 2016

octo added the New plugin label Oct 3, 2016

octo reviewed Oct 4, 2016

View reviewed changes

octo added Pending feedback and removed Pending contributor action labels Oct 7, 2016

maryamtahhan approved these changes Oct 9, 2016

View reviewed changes

octo reviewed Oct 10, 2016

View reviewed changes

rubenk reviewed Nov 19, 2016

View reviewed changes

rubenk reviewed Nov 21, 2016

View reviewed changes

octo added Pending contributor action and removed Pending feedback labels Nov 21, 2016

octo added Pending feedback and removed Pending contributor action labels Nov 24, 2016

octo added Pending feedback and removed Pending contributor action labels Dec 6, 2016

vmytnykx added 18 commits December 26, 2016 13:26

OVS link plugin: Add read callback

ed2b2f6

- Implement read callback; - Add SendNotification option; - Update plugin documentation; - Clean-up. Change-Id: Ie645192498d234d47e1fe0272ec30e7c3d9c1774 Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

OVS link: rename plugin to "ovs_events"

38176b3

Change-Id: Ic50891722074aec017b1329601a3bcbafb030ce0 Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

OVS link: rename "ovs_link" -> "ovs_events"

96a95db

- Rename files Change-Id: Ic662d5d673c1c66b2057e4b35fa3cf664e92825e Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

ovs_events: Fix plugin collectd config file

cb59d85

- Fix Ubuntu compilation warnings. Change-Id: I7bb72ec0ca732b00c4169e0174d64407d99a2751 Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

ovs_events plugin: Filter through clang-format.

193d73e

Disable clang-format for diagram in utils_ovs.c. Change-Id: Idfb1510c996f5ebbdadc96663eb032537837502a Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

OVS events: Fix 'make distcheck'

5563e62

Change-Id: I731b633b5f78ed63b643574a336c59717e408a78 Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

OVS events: Fix multiple interface config issue

02baaed

If multiple interfaces are configured, only last one will be monitored (notifications aren't sent for all others). Change-Id: Ica918ef0557747d84677db739a64698d3756a675 Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

ovs_events: Address PR comments

5a90d0e

- Change configuration format to suggested one; - Fix init/destroy API; - Fix memory leaks; - Code-clean-up. Change-Id: I1ff94271b777c69f3d07a66f43dc10d034e71101 Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

ovs_events: PR clean-up

3c2f6d2

Clean-up OVS events plugin based on PR comments. Change-Id: Ibd18924dd2a6f936d0ea83ed4eeb3b34ff8416b5 Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

ovs_events: Fix OvS documentation URL address

b56cc86

Change-Id: I5bd4d7aabaf0ff66edd9c1cd75caf7f37448e00d Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

ovs_events: Fix PR code clean-up comments

bc40bd5

- Fix OVS documentation - Code clean-up Change-Id: I84fb003aea19f73381192f31935c79b51eaba1c9 Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

ovs_events: fix using of plugin_thread_create ()

f5bbbcd

- the prototype of plugin_thread_create () has been changed in latest master. Change-Id: I492942442717c43cd6e6e73478248164b5083894 Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

vmytnykx force-pushed the feat_ovs_link branch from 8732da7 to f5bbbcd Compare December 26, 2016 13:37

ovs_events: Fix root Makefile.am

20d247b

The location of OVS sources in Malefile.am is `src/*` folder now. Change-Id: Iffe8a439faf531e74e0a752c3fb544369928cfb6 Signed-off-by: Mytnyk, VolodymyrX <volodymyrx.mytnyk@intel.com>

rubenk merged commit 3e891e5 into collectd:master Dec 31, 2016

vmytnykx deleted the feat_ovs_link branch January 3, 2017 15:15

octo added this to the 5.8 milestone Oct 11, 2017

vmytnykx mentioned this pull request Feb 22, 2018

intel_pmu: core groups feature #2681

Merged


		=item B<Address> I<node>

		The address of OVS DB server JSON-RPC interface used by the plugin. To enable


		=item B<Interfaces> [I<ifname> ...]

		List of interface names to be monitored by this plugin. If this option is missed

OVS events plugin #1971

OVS events plugin #1971

Conversation

vmytnykx commented Sep 30, 2016

vmytnykx commented Oct 3, 2016

octo commented Oct 3, 2016

octo left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

octo Oct 4, 2016 • edited

Choose a reason for hiding this comment

octo Oct 4, 2016 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

octo Oct 4, 2016 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

octo Oct 4, 2016 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

vmytnykx commented Oct 7, 2016

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

octo left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

octo Oct 10, 2016 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

vmytnykx commented Oct 28, 2016

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

vmytnykx commented Nov 21, 2016

Choose a reason for hiding this comment

Choose a reason for hiding this comment

rubenk commented Nov 21, 2016

vmytnykx commented Nov 23, 2016

maryamtahhan commented Dec 13, 2016

vmytnykx commented Dec 26, 2016

vmytnykx commented Dec 26, 2016

rubenk commented Dec 31, 2016

octo Oct 4, 2016 •

edited

octo Oct 4, 2016 •

edited

octo Oct 4, 2016 •

edited

octo Oct 4, 2016 •

edited

octo Oct 10, 2016 •

edited