By Mouhamadou Diaw

PostgreSQL 12 is under development but tests can be done. Steps to install this version can be found in this dbi blog . Many thanks to Daniel.
While reading the documentation I found that there is a new parameter for pg_upgrade. This new parameter is –socketdir.
Why this parameter?
In fact the path name of a UNIX socket is allowed to be maximally 108 chars long. Before PostgreSQL 12, the default directory for the sockets created for the temporary postmasters started by pg_upgrade was the current directory. But depending of the current directory the pathname might be very long for a socket name. In PostgreSQL 12 The default location is still the current working directory, but the parameter socketdir now allows us to specify another location

To better understand I am going to upgrade from PostgreSQL 10 to PostgreSQL 12

1
2
3
4
5
6
7
8
9
10
11
20:59:44 postgres@dbi-pg-essentials:/u02/pgdata/PG12TEST/ [PG12TEST] psql -U postgres -d postgres
psql (12devel dbi services build)
Type "help" for help.
postgres=# select version();
                                                            version
---------------------------------------------------------------------------------------------------------------------
 PostgreSQL 12devel dbi services build on x86_64-pc-linux-gnu, compiled by gcc (GCC) 4.8.5 20150623 (Red Hat 4.8.5-11
(1 row)
postgres=#

Now let’s create a very deeply nested directory

1
mkdir -p ttttttttttttttttttttttttttttttttttttttttt/aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa/aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa/ddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddd/UUUUUUUUUUUUUUUUUUUUUUUUUUUUUuuuuuuuuuuuuuuuuuuuuuuuuuuuu/ddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddd

And let’s do a cd into this new directory and let’s try to do an upgrade

1
2
3
4
$ export PGDATAOLD=/u02/pgdata/PG3
$ export PGBINOLD=/u01/app/postgres/product/10/db_1/bin
$ export PGDATANEW=/u02/pgdata/PG12TEST
$ export PGBINNEW=/u01/app/postgres/product/12dev/db_0/bin

When running the pg_upgrade with the check option, we got following errors

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
$ pg_upgrade -c
Performing Consistency Checks
-----------------------------
Checking cluster versions                                   ok
*failure*
Consult the last few lines of "pg_upgrade_server.log" for
the probable cause of the failure.
connection to database failed: Unix-domain socket path "/home/postgres/ttttttttttttttttttttttttttttttttttttttttt/aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa/aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa/ddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddd/UUUUUUUUUUUUUUUUUUUUUUUUUUUUUuuuuuuuuuuuuuuuuuuuuuuuuuuuu/ddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddd/.s.PGSQL.50432" is too long (maximum 107 bytes)
could not connect to source postmaster started with the command:
"/u01/app/postgres/product/10/db_1/bin/pg_ctl" -w -l "pg_upgrade_server.log" -D "/u02/pgdata/PG3" -o "-p 50432 -b  -c listen_addresses='' -c unix_socket_permissions=0700 -c unix_socket_directories='/home/postgres/ttttttttttttttttttttttttttttttttttttttttt/aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa/aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa/ddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddd/UUUUUUUUUUUUUUUUUUUUUUUUUUUUUuuuuuuuuuuuuuuuuuuuuuuuuuuuu/ddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddd'" start
Failure, exiting
$

Seems that the pathname for the socket is very long.
And if we use this new parameter, we can specify a new location for the sockets. And we can see that the checks are now successful

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
$ pg_upgrade -c --socketdir=/home/postgres/
Performing Consistency Checks
-----------------------------
Checking cluster versions                                   ok
Checking database user is the install user                  ok
Checking database connection settings                       ok
Checking for prepared transactions                          ok
Checking for reg* data types in user tables                 ok
Checking for contrib/isn with bigint-passing mismatch       ok
Checking for tables WITH OIDs                               ok
Checking for presence of required libraries                 ok
Checking database user is the install user                  ok
Checking for prepared transactions                          ok
*Clusters are compatible*

And then we can upgrade the cluster using this new parameter and still staying in this new created directory

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
$ pg_upgrade  --socketdir=/home/postgres/
Performing Consistency Checks
-----------------------------
Checking cluster versions                                   ok
Checking database user is the install user                  ok
Checking database connection settings                       ok
Checking for prepared transactions                          ok
Checking for reg* data types in user tables                 ok
Checking for contrib/isn with bigint-passing mismatch       ok
Checking for tables WITH OIDs                               ok
Creating dump of global objects                             ok
Creating dump of database schemas
                                                            ok
Checking for presence of required libraries                 ok
Checking database user is the install user                  ok
Checking for prepared transactions                          ok
If pg_upgrade fails after this point, you must re-initdb the
new cluster before continuing.
Performing Upgrade
------------------
Analyzing all rows in the new cluster                       ok
Freezing all rows in the new cluster                        ok
Deleting files from new pg_xact                             ok
Copying old pg_xact to new server                           ok
Setting next transaction ID and epoch for new cluster       ok
Deleting files from new pg_multixact/offsets                ok
Copying old pg_multixact/offsets to new server              ok
Deleting files from new pg_multixact/members                ok
Copying old pg_multixact/members to new server              ok
Setting next multixact ID and offset for new cluster        ok
Resetting WAL archives                                      ok
Setting frozenxid and minmxid counters in new cluster       ok
Restoring global objects in the new cluster                 ok
Restoring database schemas in the new cluster
                                                            ok
Copying user relation files
                                                            ok
Setting next OID for new cluster                            ok
Sync data directory to disk                                 ok
Creating script to analyze new cluster                      ok
Creating script to delete old cluster                       ok
Upgrade Complete
----------------
Optimizer statistics are not transferred by pg_upgrade so,
once you start the new server, consider running:
    ./analyze_new_cluster.sh
Running this script will delete the old cluster's data files:
    ./delete_old_cluster.sh
$

Conclusion :

In this blog we have seen the new option –socketdir for pg_upgrade. It’s a good thing to know that this parameter exists, but in most case the current working directory should be ok for an upgrade

Post Views: 1,058
Thumbnail [60x60]
by
Oracle Team