[ovs-dev] [PATCH v9] ovsdb-tool: Convert clustered db to standalone db.

Han Zhou zhouhan at gmail.com
Fri Aug 30 05:29:56 UTC 2019


Thanks for the update.

On Thu, Aug 29, 2019 at 6:20 PM <amginwal at gmail.com> wrote:
>
> From: Aliasgar Ginwala <aginwala at ebay.com>
>
> Add support in ovsdb-tool for migrating clustered dbs to standalone dbs.
> E.g. usage to migrate nb/sb db to standalone db from raft:
> ovsdb-tool cluster-to-standalone ovnnb_db.db ovnnb_db_cluster.db
>
> Signed-off-by: Aliasgar Ginwala <aginwala at ebay.com>
> ---
>  Documentation/ref/ovsdb.7.rst |   3 +
>  NEWS                          |   3 +
>  ovsdb/ovsdb-tool.1.in         |   8 +++
>  ovsdb/ovsdb-tool.c            | 101 +++++++++++++++++++++++++++++++++-
>  tests/ovsdb-tool.at           |  43 +++++++++++++++
>  5 files changed, 157 insertions(+), 1 deletion(-)
>
> diff --git a/Documentation/ref/ovsdb.7.rst b/Documentation/ref/ovsdb.7.rst
> index cd1c63d64..b12d8066c 100644
> --- a/Documentation/ref/ovsdb.7.rst
> +++ b/Documentation/ref/ovsdb.7.rst
> @@ -514,6 +514,9 @@ standalone database from the contents of a running
clustered database.
>  When the cluster is down and cannot be revived, ``ovsdb-client backup``
will
>  not work.
>
> +Use ``ovsdb-tool cluster-to-standalone`` to convert clustered database to
> +standalone database when the cluster is down and cannot be revived.
> +
>  Upgrading or Downgrading a Database
>  -----------------------------------
>
> diff --git a/NEWS b/NEWS
> index c5caa13d6..a02f9f1a6 100644
> --- a/NEWS
> +++ b/NEWS
> @@ -49,6 +49,9 @@ v2.12.0 - xx xxx xxxx
>         quickly after a brief disconnection, saving bandwidth and CPU
time.
>         See section 4.1.15 of ovsdb-server(7) for details of related OVSDB
>         protocol extension.
> +     * Support to convert from cluster database to standalone database
is now
> +       available when clustered is down and cannot be revived using
ovsdb-tool
> +       . Check "Database Migration Commands" in ovsdb-tool man section.
>     - OVN:
>       * IPAM/MACAM:
>         - select IPAM mac_prefix in a random manner if not provided by
the user
> diff --git a/ovsdb/ovsdb-tool.1.in b/ovsdb/ovsdb-tool.1.in
> index ec85e14c4..31a918d90 100644
> --- a/ovsdb/ovsdb-tool.1.in
> +++ b/ovsdb/ovsdb-tool.1.in
> @@ -147,6 +147,14 @@ avoid this possibility, specify
\fB\-\-cid=\fIuuid\fR, where
>  \fIuuid\fR is the cluster ID of the cluster to join, as printed by
>  \fBovsdb\-tool get\-cid\fR.
>  .
> +.SS "Database Migration Commands"
> +This commands will convert cluster database to standalone database.
> +.
> +.IP "\fBcluster\-to\-standalone\fI db clusterdb"
> +Use this command to convert to standalone database from clustered
database
> +when the cluster is down and cannot be revived. It creates new standalone
> +\fIdb\fR file from the given cluster \fIdb\fR file.
> +.
>  .SS "Version Management Commands"
>  .so ovsdb/ovsdb-schemas.man
>  .PP
> diff --git a/ovsdb/ovsdb-tool.c b/ovsdb/ovsdb-tool.c
> index 438f97590..3bbf4c8bc 100644
> --- a/ovsdb/ovsdb-tool.c
> +++ b/ovsdb/ovsdb-tool.c
> @@ -173,6 +173,9 @@ usage(void)
>             "  compare-versions A OP B  compare OVSDB schema version
numbers\n"
>             "  query [DB] TRNS         execute read-only transaction on
DB\n"
>             "  transact [DB] TRNS      execute read/write transaction on
DB\n"
> +           "  cluster-to-standalone DB DB    Convert clustered DB to\n"
> +           "      standalone DB when cluster is down and cannot be\n"
> +           "        revived\n"
>             "  [-m]... show-log [DB]   print DB's log entries\n"
>             "The default DB is %s.\n"
>             "The default SCHEMA is %s.\n",
> @@ -942,6 +945,55 @@ print_raft_record(const struct raft_record *r,
>      }
>  }
>
> +static void
> +raft_header_to_standalone_log(const struct raft_header *h,
> +                              struct ovsdb_log *db_log_data)
> +{
> +    if (h->snap_index) {
> +        if (!h->snap.data || json_array(h->snap.data)->n != 2) {
> +        ovs_fatal(0, "Incorrect raft header data array length");
> +        }
> +
> +        struct json *schema_json = json_array(h->snap.data)->elems[0];
> +        if (schema_json->type != JSON_NULL) {
> +            struct ovsdb_schema *schema;
> +            check_ovsdb_error(ovsdb_schema_from_json(schema_json,
&schema));
> +            ovsdb_schema_destroy(schema);
> +            check_ovsdb_error(ovsdb_log_write_and_free(db_log_data,
> +                                                       schema_json));
> +        }
> +
> +        struct json *data_json = json_array(h->snap.data)->elems[1];
> +        if (!data_json || data_json->type != JSON_OBJECT) {
> +            ovs_fatal(0, "Invalid raft header data");
> +        }
> +        if (data_json->type != JSON_NULL) {
> +            check_ovsdb_error(ovsdb_log_write_and_free(db_log_data,
> +                                                       data_json));
> +        }
> +    }
> +}
> +
> +static void
> +raft_record_to_standalone_log(const struct raft_record *r,
> +                              struct ovsdb_log *db_log_data)
> +{
> +    if (r->type == RAFT_REC_ENTRY) {
> +        if (!r->entry.data) {
> +            return;
> +        }
> +        if (json_array(r->entry.data)->n != 2) {
> +            ovs_fatal(0, "Incorrect raft record array length");
> +        }
> +
> +        struct json *data_json = json_array(r->entry.data)->elems[1];
> +        if (data_json->type != JSON_NULL) {
> +            check_ovsdb_error(ovsdb_log_write_and_free(db_log_data,
> +                                                       data_json));
> +        }
> +    }
> +}
> +
>  static void
>  do_show_log_cluster(struct ovsdb_log *log)
>  {
> @@ -1511,6 +1563,51 @@ do_compare_versions(struct ovs_cmdl_context *ctx)
>      exit(result ? 0 : 2);
>  }
>
> +static void
> +do_convert_to_standalone(struct ovsdb_log *log, struct ovsdb_log
*db_log_data)
> +{
> +    for (unsigned int i = 0; ; i++) {
> +        struct json *json;
> +        check_ovsdb_error(ovsdb_log_read(log, &json));
> +        if (!json) {
> +            break;
> +        }
> +
> +        if (i == 0) {
> +            struct raft_header h;
> +            check_ovsdb_error(raft_header_from_json(&h, json));
> +            raft_header_to_standalone_log(&h, db_log_data);
> +            raft_header_uninit(&h);
> +        } else {
> +            struct raft_record r;
> +            check_ovsdb_error(raft_record_from_json(&r, json));
> +            raft_record_to_standalone_log(&r, db_log_data);
> +            raft_record_uninit(&r);
> +        }
> +    }
> +}
> +
> +static void
> +do_cluster_standalone(struct ovs_cmdl_context *ctx)
> +{
> +    const char *db_file_name = ctx->argv[1];
> +    const char *cluster_db_file_name = ctx->argv[2];
> +    struct ovsdb_log *log;
> +    struct ovsdb_log *db_log_data;
> +
> +    check_ovsdb_error(ovsdb_log_open(cluster_db_file_name,
> +                                     OVSDB_MAGIC"|"RAFT_MAGIC,
> +                                     OVSDB_LOG_READ_ONLY, -1, &log));
> +    check_ovsdb_error(ovsdb_log_open(db_file_name, OVSDB_MAGIC,
> +                                     OVSDB_LOG_CREATE_EXCL, -1,
&db_log_data));
> +    if (strcmp(ovsdb_log_get_magic(log), RAFT_MAGIC) != 0) {
> +        ovs_fatal(0, "Database is not clustered db.\n");
> +    }
> +    do_convert_to_standalone(log, db_log_data);
> +    check_ovsdb_error(ovsdb_log_commit_block(db_log_data));
> +    ovsdb_log_close(db_log_data);
> +    ovsdb_log_close(log);
> +}
>
>  static void
>  do_help(struct ovs_cmdl_context *ctx OVS_UNUSED)
> @@ -1550,7 +1647,9 @@ static const struct ovs_cmdl_command all_commands[]
= {
>      { "compare-versions", "a op b", 3, 3, do_compare_versions, OVS_RO },
>      { "help", NULL, 0, INT_MAX, do_help, OVS_RO },
>      { "list-commands", NULL, 0, INT_MAX, do_list_commands, OVS_RO },
> -    { NULL, NULL, 0, 0, NULL, OVS_RO },
> +    { "cluster-to-standalone", "db clusterdb", 2, 2,
> +    do_cluster_standalone, OVS_RW },
> +    { NULL, NULL, 2, 2, NULL, OVS_RO },
>  };
>
>  static const struct ovs_cmdl_command *get_all_commands(void)
> diff --git a/tests/ovsdb-tool.at b/tests/ovsdb-tool.at
> index 69c5d6afa..953b1ed8f 100644
> --- a/tests/ovsdb-tool.at
> +++ b/tests/ovsdb-tool.at
> @@ -459,3 +459,46 @@ OVS_APP_EXIT_AND_WAIT([ovsdb-server])
>  # Make sure that the clustered data matched the standalone data.
>  AT_CHECK([cat dump2], [0], [expout])
>  AT_CLEANUP
> +
> +AT_SETUP([ovsdb-tool convert-to-standalone])
> +AT_KEYWORDS([ovsdb file positive])
> +ordinal_schema > schema
> +AT_CHECK([ovsdb-tool create-cluster db schema unix:s1.raft], [0],
[stdout], [ignore])
> +AT_CHECK([ovsdb-server --detach --no-chdir --pidfile
--remote=punix:socket --log-file db >/dev/null 2>&1])
> +for txn in m4_foreach([txn], [[[["ordinals",
> +      {"op": "insert",
> +       "table": "ordinals",
> +       "row": {"number": 0, "name": "zero"}},
> +      {"op": "insert",
> +       "table": "ordinals",
> +       "row": {"number": 1, "name": "one"}},
> +      {"op": "insert",
> +       "table": "ordinals",
> +       "row": {"number": 2, "name": "two"}}]]]], ['txn' ]); do
> +AT_CHECK([ovsdb-client transact unix:socket "$txn"], [0], [ignore],
[ignore])
> +done
> +AT_CHECK([ovsdb-client transact unix:socket '[["ordinals"]]'], [0],
> +         [ignore], [ignore])
> +AT_CHECK([ovsdb-client dump unix:socket > clusterdump])
> +AT_CHECK([ovs-appctl -t ovsdb-server -e exit], [0], [ignore], [ignore])
> +
> +# Convert to standalone database from clustered database.
> +AT_CHECK(ovsdb-tool cluster-to-standalone db1 db)
> +
> +# Check its standalone db
> +AT_CHECK([ovsdb-tool db-is-standalone db1])
> +
> +# Dump the standalone db data.
> +AT_CHECK([ovsdb-server -vconsole:off -vfile -vvlog:off --detach
--no-chdir --pidfile --log-file --remote=punix:db.sock db1])
> +AT_CHECK([ovsdb_client_wait ordinals connected])
> +AT_CHECK([ovsdb-client dump  > standalonedump])
> +OVS_APP_EXIT_AND_WAIT([ovsdb-server])
> +
> +# Make sure both standalone and cluster db data matches.
> +AT_CHECK([grep "zero" expout], [0], [ignore])
> +AT_CHECK([grep "zero" standalonedump], [0], [ignore])
> +AT_CHECK([grep "one" clusterdump], [0], [ignore])
> +AT_CHECK([grep "one" standalonedump], [0], [ignore])
> +AT_CHECK([grep "two" clusterdump], [0], [ignore])
> +AT_CHECK([grep "two" standalonedump], [0], [ignore])

Can just compare by:
AT_CHECK([diff standalonedump clusterdump])

Alternatively, if the dumped cluster file name was: expout, then you can do:
AT_CHECK([cat standalonedump], [0], [expout])

This would be a minor fix. So:
Acked-by: Han Zhou <hzhou8 at ebay.com>


More information about the dev mailing list