Feature: tools: List common fencing resource instance attributes via crm_resource #3399

nrwahl2 · 2024-03-25T07:44:45Z

Next step on T620. This allows us to use pcmk__daemon_metadata() in the fencer and deprecate the daemon metadata command.

The plan is to use a similar approach in crm_resource for the various kinds of resource meta-attributes. Their descriptions will have to be pulled from Pacemaker Explained since we don't have a nice template sitting in another C file :)

It might also be nice for crm_resource --list-options to accept the name of a resource and output all the meta-attributes that apply to it (for example, clone and primitive meta-attributes for a clone resource). But that's not strictly required, and it might get messy/counter-intuitive if we don't include instance attributes... or if we do.

Edit: went ahead and added primitive meta-attributes. I can break this up at any point, preferably somewhere after --list-options=fencing.

There are a lot of commits but most of them are trivial.

nrwahl2 · 2024-03-25T23:38:09Z

rebased on main

nrwahl2 · 2024-03-26T10:15:51Z

Went ahead and added primitive meta-attributes. I can break this up at any point, preferably somewhere after --list-options=fencing.

There are a lot of commits but most of them are trivial.

nrwahl2 · 2024-03-26T19:53:13Z

Added missing { NULL, } to the end of primitive_meta (thanks @clumens)

kgaillot

awesome

kgaillot · 2024-04-04T16:07:26Z

tools/crm_resource.c

+        g_set_error(&error, PCMK__EXITC_ERROR, CRM_EX_USAGE,
+                    "Invalid --list-options value '%s'. Allowed values: "
+                    PCMK_VALUE_FENCING,
+                    pcmk__s(options.opt_list, "(BUG: none)"));


It's better to do validation in an option callback when practical, so invalid arguments are handled consistently (by glib). In this case we could save the filter rather than the text in options

IIRC this was another case where I was trying to allow invalid values as long as the LAST value was valid. I've got no problem changing it as you described. Will do.

Yeah I think validation should take precedence over last-wins, since otherwise it could obscure an error (consider a script that incorrectly adds a value intended to be used as a default, then sometimes correctly adds a different value after it)

xml/api/crm_attribute-2.37.rng

lib/common/options.c

kgaillot · 2024-04-04T16:30:42Z

include/crm/common/roles.h

@@ -20,6 +20,10 @@ extern "C" {
 * \ingroup core
 */

+// String equivalents of enum rsc_role_e
+
+#define PCMK_ROLE_STOPPED       "Stopped"


I left these internal since external code can use pcmk_role_text(v) to get them. However they can be used in the CIB, so I suppose this makes sense.

That was my thinking. It's not required -- we could rely on pcmk_role_text(), and use literal strings in the option metadata -- but this has been our usual approach for strings used in the CIB.

What I really dislike is having these string constants split across public and internal header files. But I don't want to expose unknown and the legacy ones.

kgaillot · 2024-04-04T16:40:21Z

lib/common/options.c

+        N_("Number of seconds after a failed action for this resource before "
+            "acting as if the failure had not occurred, and potentially "
+            "allowing the resource back to the node on which it failed. "
+            "A value of 0 indicates that this feature is disabled."),


We really need to update all these descriptions ... someday

Agreed. For the most part, I copied from Pacemaker Explained, sometimes updating if I noticed obvious errors.

kgaillot · 2024-04-04T16:44:02Z

tools/crm_resource.c

@@ -472,6 +472,7 @@ static GOptionEntry query_entries[] = {
    { "list-options", 0, G_OPTION_FLAG_NONE, G_OPTION_ARG_CALLBACK, command_cb,
      "List all available options of the given type\n"
      INDENT "Allowed values:\n"
+      INDENT PCMK__VALUE_PRIMITIVE "(primitive resource meta-attributes), "
      INDENT PCMK_VALUE_FENCING " (parameters common to all fencing resources)",


any reason just primitive is internal?

Yep! PCMK_VALUE_FENCING already existed as a public constant for the "requires" meta-attribute. We don't need PCMK__VALUE_PRIMITIVE for anything public yet.

Though we could use it for resource variant. I've got a personal TODO note for resource variant strings... whether they should have their own constants (PCMK_VALUE or a new PCMK_<something_else>), or just use the existing PCMK_XE constants in all contexts.

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

nrwahl2 · 2024-04-05T09:23:25Z

rebased on main

Nothing uses this yet Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

Nothing uses this yet. It's not necessary for filtering the options in the fencing_params array. However, it will be useful for telling pcmk__daemon_metadata() which option list to use and for storing the option list as an enum in crm_resource. Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

This lists all the parameters that are common to all fencing resources as an OCF-like metadata XML string. Bump the CRM_FEATURE_SET. Ref T620 Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

Not used yet Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

This lists all the meta-attributes applicable to a primitive resource as an OCF-like metadata XML string. Bump the CRM_FEATURE_SET. Ref T620 Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

Ref T620 Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

Fix formatting of stop_start and note that promote/demote are affected by is-managed/maintenance. Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

Unused since 73a5d63. Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

nrwahl2 · 2024-04-09T02:45:38Z

retest this please

nrwahl2 · 2024-04-09T04:08:14Z

@kgaillot There's a race condition in CI right now. Fabio would like Chrissie to look at the logs before we merge this PR (which clears the logs). She's out today, so we need to hold off.

… 1168817 https://build.opensuse.org/request/show/1168817 by user yan_gao + anag+factory - Update to version 2.1.7+20240411.81041cf0b: - libcrmcommon: Avoid use-after-free in mark_xml_changes() - libcrmcommon: Mark parents dirty in pcmk__mark_xml_created() - Update to version 2.1.7+20240410.ae4b38ab5: - scheduler: deprecate Nagios and Upstart resources even if built with --enable-compat-2.0 (gh#ClusterLabs/pacemaker#3417) - Update to version 2.1.7+20240410.74b7a09c5: - tools: New crm_resource --list-options=primitive option (gh#ClusterLabs/pacemaker#3399) - libcrmcommon: Use PCMK_VALUE_VERSION as option type (gh#ClusterLabs/pacemaker#3399) - libcrmcommon: Use PCMK_VALUE_TIMEOUT as option type (gh#ClusterLabs/pacemaker#3399) - libcrmcommon: Use PCMK_VALUE_SCORE as option type (gh#ClusterLabs/pacemaker#3399) - libcrmcommon: Use PCMK_VALUE_NONNEGATIVE_INTEGER as opt type (gh#Clu

nrwahl2 force-pushed the nrwahl2-T620_new branch 5 times, most recently from 6ba3a1f to f9d25b6 Compare March 25, 2024 20:49

nrwahl2 marked this pull request as draft March 25, 2024 23:33

nrwahl2 force-pushed the nrwahl2-T620_new branch from f9d25b6 to 01c437d Compare March 25, 2024 23:36

nrwahl2 force-pushed the nrwahl2-T620_new branch 2 times, most recently from aa7813c to c1f267a Compare March 26, 2024 10:13

nrwahl2 marked this pull request as ready for review March 26, 2024 10:13

nrwahl2 force-pushed the nrwahl2-T620_new branch from c1f267a to b1a5381 Compare March 26, 2024 10:15

nrwahl2 force-pushed the nrwahl2-T620_new branch from b1a5381 to 1fc5e3b Compare March 26, 2024 19:52

nrwahl2 mentioned this pull request Apr 2, 2024

Bring rule schemas, documentation, and code in alignment #3409

Merged

kgaillot reviewed Apr 4, 2024

View reviewed changes

nrwahl2 marked this pull request as draft April 5, 2024 09:22

Doc: libcrmcommon: Drop duplicate log level doxygen from header file

e8cfc51

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

nrwahl2 force-pushed the nrwahl2-T620_new branch from 1fc5e3b to 8057c95 Compare April 5, 2024 09:23

nrwahl2 added 9 commits April 8, 2024 15:58

Refactor: libcrmcommon: New pcmk__output_fencing_params()

9833e93

Nothing uses this yet Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

Refactor: libcrmcommon, fencer: Use pcmk__daemon_metadata() for fencer

0389fc4

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

Doc: libcrmcommon: Mark pcmk__opt_DAEMON flags as deprecated

7e95742

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

Refactor: libpacemaker: New pcmk__list_fencing_params()

351d5f8

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

API: libpacemaker: New pcmk_list_fencing_params()

d4774a9

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

Low: schemas: Copy schema in preparation for changes

5c8409a

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

API: schemas: Allow options in crm_resource output

31b3c36

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

Feature: crm_resource: New --list-options=fencing option

6a354ba

This lists all the parameters that are common to all fencing resources as an OCF-like metadata XML string. Bump the CRM_FEATURE_SET. Ref T620 Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

nrwahl2 added 23 commits April 8, 2024 16:53

Test: cts-cli: Update for PCMK_VALUE_TIMEOUT as option type

b15d8d2

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

Refactor: libcrmcommon: Use PCMK_VALUE_VERSION as option type

b5e291d

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

Test: cts-cli: Update tests for PCMK_VALUE_VERSION as option type

904eb3a

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

Doc: Pacemaker Explained: pcmk_delay_base is a text option

0edaa13

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

API: libcrmcommon: New PCMK_VALUE_DYNAMIC_LIST string constant

8db4f2a

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

API: libcrmcommon: New PCMK_VALUE_STATIC_LIST string constant

834fd59

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

API: libcrmcommon: New PCMK_VALUE_STATUS string constant

52fd4b5

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

API: libcrmcommon: New PCMK_ROLE_STOPPED string constant

534e1c3

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

API: libcrmcommon: New PCMK_ROLE_STARTED string constant

c486999

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

API: libcrmcommon: New PCMK_ROLE_UNPROMOTED string constant

e595495

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

API: libcrmcommon: New PCMK_ROLE_PROMOTED string constant

89b8f20

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

API: libcrmcommon: New PCMK_VALUE_STOP_ONLY string constant

c53719b

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

API: libcrmcommon: New PCMK_VALUE_STOP_START string constant

503e36a

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

API: libcrmcommon: New PCMK_VALUE_STOP_UNEXPECTED string constant

1e8b6aa

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

API: libcrmcommon: New PCMK_VALUE_PORT string constant

3687074

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

Refactor: libcrmcommon: New pcmk__output_primitive_meta()

1ea50d5

Not used yet Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

Refactor: libpacemaker: New pcmk__list_primitive_meta()

a0ea432

Not used yet Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

API: libpacemaker: New pcmk_list_primitive_meta()

52c76d7

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

Refactor: libcrmcommon: New pcmk__opt_primitive enum value

9108504

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

Feature: tools: New crm_resource --list-options=primitive option

7c8e051

This lists all the meta-attributes applicable to a primitive resource as an OCF-like metadata XML string. Bump the CRM_FEATURE_SET. Ref T620 Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

Test: cts-cli: Add tests for crm_resource --list-options=primitive

037e08c

Ref T620 Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

Doc: Pacemaker Explained: Minor primitive meta-attribute fixes

613ef38

Fix formatting of stop_start and note that promote/demote are affected by is-managed/maintenance. Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

Refactor: libcrmcommon: Drop pcmk__valid_script()

6056905

Unused since 73a5d63. Signed-off-by: Reid Wahl <nrwahl@protonmail.com>

nrwahl2 marked this pull request as ready for review April 9, 2024 00:00

nrwahl2 force-pushed the nrwahl2-T620_new branch from 8057c95 to 6056905 Compare April 9, 2024 00:00

kgaillot approved these changes Apr 9, 2024

View reviewed changes

nrwahl2 merged commit fc69389 into ClusterLabs:main Apr 10, 2024
1 check passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Feature: tools: List common fencing resource instance attributes via crm_resource #3399

Feature: tools: List common fencing resource instance attributes via crm_resource #3399

nrwahl2 commented Mar 25, 2024 •

edited

Loading

nrwahl2 commented Mar 25, 2024

nrwahl2 commented Mar 26, 2024

nrwahl2 commented Mar 26, 2024

kgaillot left a comment

kgaillot Apr 4, 2024

nrwahl2 Apr 8, 2024

kgaillot Apr 8, 2024

kgaillot Apr 4, 2024

nrwahl2 Apr 5, 2024 •

edited

Loading

kgaillot Apr 4, 2024

nrwahl2 Apr 5, 2024

kgaillot Apr 4, 2024

nrwahl2 Apr 5, 2024 •

edited

Loading

nrwahl2 commented Apr 5, 2024

nrwahl2 commented Apr 9, 2024

nrwahl2 commented Apr 9, 2024

Feature: tools: List common fencing resource instance attributes via crm_resource #3399

Feature: tools: List common fencing resource instance attributes via crm_resource #3399

Conversation

nrwahl2 commented Mar 25, 2024 • edited Loading

nrwahl2 commented Mar 25, 2024

nrwahl2 commented Mar 26, 2024

nrwahl2 commented Mar 26, 2024

kgaillot left a comment

Choose a reason for hiding this comment

kgaillot Apr 4, 2024

Choose a reason for hiding this comment

nrwahl2 Apr 8, 2024

Choose a reason for hiding this comment

kgaillot Apr 8, 2024

Choose a reason for hiding this comment

kgaillot Apr 4, 2024

Choose a reason for hiding this comment

nrwahl2 Apr 5, 2024 • edited Loading

Choose a reason for hiding this comment

kgaillot Apr 4, 2024

Choose a reason for hiding this comment

nrwahl2 Apr 5, 2024

Choose a reason for hiding this comment

kgaillot Apr 4, 2024

Choose a reason for hiding this comment

nrwahl2 Apr 5, 2024 • edited Loading

Choose a reason for hiding this comment

nrwahl2 commented Apr 5, 2024

nrwahl2 commented Apr 9, 2024

nrwahl2 commented Apr 9, 2024

nrwahl2 commented Mar 25, 2024 •

edited

Loading

nrwahl2 Apr 5, 2024 •

edited

Loading

nrwahl2 Apr 5, 2024 •

edited

Loading