Peter Geoghegan peter.geoghegan86 at gmail.com
Wed Aug 5 01:24:14 PDT 2009
Sorry to repost the same thing, but my earlier message of July 30th
wasn't answered, and was most likely overlooked as it was part of what
was thought to be a finished thread.

It seems I was wrong when I said that the problem was
fixed. I assumed that since the event lag returned to zero when new
paths where stored per Chris Brown's direction, the problem was
corrected. However, the problem persists.

When I restart a slave database (in the following example node 2),
replication works fine (at least as far as can be immediately
observed), but sl_status shows:

1;3;38689;"2009-07-30 12:11:51.796";38688;"2009-07-30
12:12:02.428316";"2009-07-30 12:11:41.859";1;"00:00:14.015"
1;2;38689;"2009-07-30 12:11:51.796";38605;"2009-07-30
11:52:35.119048";"2009-07-30 11:58:05.734";84;"00:13:50.14"

Node 2's event lag grows and grows (until the slon service is
restarted, at which time it returns to zero, just as before).

When I run test_slony_state-dbi.pl while the event lag continues to
grow, it outputs the following:

peter at peter-development-machine:~/slony1-2.0.2/tools>
./test_slony_state-dbi.pl --host=10.0.0.80 --database=lustre
--cluster=lustre_cluster --user=postgres --password=my_password
DSN: dbi:Pg:dbname=lustre;host=10.0.0.80;user=postgres;password=my_password;
===========================
Rummage for DSNs
=============================
Query:

  select p.pa_server, p.pa_conninfo
  from "_lustre_cluster".sl_path p
--   where exists (select * from "_lustre_cluster".sl_subscribe s where
--                          (s.sub_provider = p.pa_server or
s.sub_receiver = p.pa_server) and
--                          sub_active = 't')
  group by pa_server, pa_conninfo;


Tests for node 1 - DSN = dbi:Pg:dbname=lustre host=10.0.0.80
user=postgres password=my_password
========================================
pg_listener info:
Pages: 0
Tuples: 0

Size Tests
================================================
      sl_log_1         0  0.000000
      sl_log_2         0  0.000000
     sl_seqlog         0  0.000000

Listen Path Analysis
===================================================
No problems found with sl_listen

--------------------------------------------------------------------------------
Summary of event info
 Origin  Min SYNC  Max SYNC Min SYNC Age Max SYNC Age
================================================================================
     1     38605     38699     00:00:00     00:15:00    0
     2        20        20     01:08:00     01:08:00    1
     3        30        30     01:02:00     01:02:00    1


---------------------------------------------------------------------------------
Summary of sl_confirm aging
  Origin   Receiver   Min SYNC   Max SYNC  Age of latest SYNC  Age of
eldest SYNC
=================================================================================
       1          2      38605      38605      00:20:00      00:20:00    0
       1          3      38627      38698      00:00:00      00:11:00    0
       2          1         20         20      01:03:00      01:03:00    1
       2          3         20         20      01:02:00      01:02:00    1
       3          1         30         30      01:02:00      01:02:00    1
       3          2         30         30      01:08:00      01:08:00    1


------------------------------------------------------------------------------

Listing of old open connections on node 1
      Database             PID            User    Query Age
    Query
================================================================================


Tests for node 3 - DSN = dbi:Pg:dbname=lustre_slave host=10.0.0.82
user=postgres password=my_password
========================================
pg_listener info:
Pages: 0
Tuples: 0

Size Tests
================================================
      sl_log_1         0  0.000000
      sl_log_2         0  0.000000
     sl_seqlog         0  0.000000

Listen Path Analysis
===================================================
No problems found with sl_listen

--------------------------------------------------------------------------------
Summary of event info
 Origin  Min SYNC  Max SYNC Min SYNC Age Max SYNC Age
================================================================================
     1     38605     38699     00:00:00     00:15:00    0
     2        20        20     01:08:00     01:08:00    1
     3        30        30     01:02:00     01:02:00    1


---------------------------------------------------------------------------------
Summary of sl_confirm aging
  Origin   Receiver   Min SYNC   Max SYNC  Age of latest SYNC  Age of
eldest SYNC
=================================================================================
       1          2      38605      38605      00:21:00      00:21:00    0
       1          3      38629      38699      00:00:00      00:11:00    0
       2          1         20         20      01:03:00      01:03:00    1
       2          3         20         20      01:03:00      01:03:00    1
       3          1         30         30      01:03:00      01:03:00    1
       3          2         30         30      01:08:00      01:08:00    1


------------------------------------------------------------------------------

Listing of old open connections on node 3
      Database             PID            User    Query Age
    Query
================================================================================


Tests for node 2 - DSN = dbi:Pg:dbname=lustre_slave host=10.0.0.81
user=postgres password=my_password
========================================
pg_listener info:
Pages: 0
Tuples: 0

Size Tests
================================================
      sl_log_1         0  0.000000
      sl_log_2         0  0.000000
     sl_seqlog         0  0.000000

Listen Path Analysis
===================================================
No problems found with sl_listen

--------------------------------------------------------------------------------
Summary of event info
 Origin  Min SYNC  Max SYNC Min SYNC Age Max SYNC Age
================================================================================
     1     38573     38699    -00:05:00     00:15:00    0
     2        20        21     00:15:00     01:03:00    0
     3        30        30     00:57:00     00:57:00    1


---------------------------------------------------------------------------------
Summary of sl_confirm aging
  Origin   Receiver   Min SYNC   Max SYNC  Age of latest SYNC  Age of
eldest SYNC
=================================================================================
       1          2      38607      38699      00:00:00      00:15:00    0
       1          3      38573      38698     -00:05:00      00:15:00    0
       2          1         20         20      00:57:00      00:57:00    1
       2          3         20         20      00:57:00      00:57:00    1
       3          1         30         30      00:57:00      00:57:00    1
       3          2         30         30      01:02:00      01:02:00    1


------------------------------------------------------------------------------

Listing of old open connections on node 2
      Database             PID            User    Query Age
    Query
================================================================================

peter at peter-development-machine:~/slony1-2.0.2/tools>

Any further help you could offer is greatly appreciated,

Regards,
Peter Geoghegan


More information about the Slony1-general mailing list