Re: [GENERAL] Help setting up warm standby replication
Nico Sabbi wrote: Merlin Moncure wrote: try link mode, not copy mode (-l). make sure you read about the -k switch. merlin replaced -c with -l, but the result was the same. I assume that with -k you mean the socket dir for postgres - since pg_standby doesn't seek to recognize that switch - but I didn't need it: I connected to -p 5433. Anyway, after having stopped the replication I had the same problem; in order to make the slave start I had to manually copy the log files from the archive to pg_xlog/ . Is pg_standby supposed to work with Postgres 8.2 ? or only with 8.1.? update: touching the right trigger makes the replication stop, thus I can connect to the server and see the updated data. Thanks a lot, Nico ---(end of broadcast)--- TIP 2: Don't 'kill -9' the postmaster
Re: [GENERAL] Help setting up warm standby replication
Merlin Moncure wrote: try link mode, not copy mode (-l). make sure you read about the -k switch. merlin replaced -c with -l, but the result was the same. I assume that with -k you mean the socket dir for postgres - since pg_standby doesn't seek to recognize that switch - but I didn't need it: I connected to -p 5433. Anyway, after having stopped the replication I had the same problem; in order to make the slave start I had to manually copy the log files from the archive to pg_xlog/ . Is pg_standby supposed to work with Postgres 8.2 ? or only with 8.1.? ---(end of broadcast)--- TIP 9: In versions below 8.0, the planner will ignore your desire to choose an index scan if your joining column's datatypes do not match
Re: [GENERAL] Help setting up warm standby replication
On 4/16/07, Nico Sabbi <[EMAIL PROTECTED]> wrote: Merlin Moncure wrote: > On 4/16/07, Nico Sabbi <[EMAIL PROTECTED]> wrote: > >> Is there any parameter that I have to pass to the second server to keep >> on requesting WALs? >> I still don't understand what instructs the server to continously >> request the master's logs. > > > google pg_standby. I've set it up and it works. > > merlin > Hi, I installed the second version of pg_standby.tar that you posted here. In recovery.conf I set restore_command = '/usr/local/bin/pg_standby -d -c -s 5 -w 0 /tmp/pg/ %f %p' and I see in the log file that something goes on after every update: Trigger file : (null) Waiting for WAL file : /tmp/pg//00010021 WAL file path: 00010021 Restoring to... : pg_xlog/RECOVERYXLOG Sleep interval : 5 seconds Max wait interval: 0 forever Command for restore : cp /tmp/pg//00010021 pg_xlog/RECOVERYXLOG running restore : success LOG: restored file "00010021" from archivio but "cp /tmp/pg//00010021 pg_xlog/RECOVERYXLOG" looks suspicious to me. Is it expected? should WAL files overwrite pg_xlog/RECOVERYXLOG ? If I stop the replication and start the slave after having removed the recovery.conf the server doesn't come up: LOG: aborting startup due to startup process failure LOG: database system was interrupted while in recovery at log time 2007-04-16 15:29:42 CEST HINT: If this has occurred more than once some data may be corrupted and you may need to choose an earlier recovery target. LOG: impossibile aprire il file "pg_xlog/00010009" (file registro 0, segmento 9): No such file or directory LOG: invalid primary checkpoint record LOG: impossibile aprire il file "pg_xlog/00010021" (file registro 0, segmento 33): No such file or directory LOG: invalid secondary checkpoint record PANIC: impossibile localizzare un checkpoint record valido LOG: startup process (PID 10824) was terminated by signal 6 LOG: aborting startup due to startup process failure infact pg_xlog/ of the slave contains only RECOVERHISTORY: cat pg_xlog/RECOVERYHISTORY START WAL LOCATION: 0/920 (file 00010009) STOP WAL LOCATION: 0/A00 (file 0001000A) CHECKPOINT LOCATION: 0/920 START TIME: 2007-04-16 15:29:42 CEST LABEL: label STOP TIME: 2007-04-16 15:31:39 CEST Did I do something wrong? Thanks for your help, try link mode, not copy mode (-l). make sure you read about the -k switch. merlin ---(end of broadcast)--- TIP 3: Have you checked our extensive FAQ? http://www.postgresql.org/docs/faq
Re: [GENERAL] Help setting up warm standby replication
Merlin Moncure wrote: On 4/16/07, Nico Sabbi <[EMAIL PROTECTED]> wrote: Is there any parameter that I have to pass to the second server to keep on requesting WALs? I still don't understand what instructs the server to continously request the master's logs. google pg_standby. I've set it up and it works. merlin Hi, I installed the second version of pg_standby.tar that you posted here. In recovery.conf I set restore_command = '/usr/local/bin/pg_standby -d -c -s 5 -w 0 /tmp/pg/ %f %p' and I see in the log file that something goes on after every update: Trigger file : (null) Waiting for WAL file : /tmp/pg//00010021 WAL file path: 00010021 Restoring to... : pg_xlog/RECOVERYXLOG Sleep interval : 5 seconds Max wait interval: 0 forever Command for restore : cp /tmp/pg//00010021 pg_xlog/RECOVERYXLOG running restore : success LOG: restored file "00010021" from archivio but "cp /tmp/pg//00010021 pg_xlog/RECOVERYXLOG" looks suspicious to me. Is it expected? should WAL files overwrite pg_xlog/RECOVERYXLOG ? If I stop the replication and start the slave after having removed the recovery.conf the server doesn't come up: LOG: aborting startup due to startup process failure LOG: database system was interrupted while in recovery at log time 2007-04-16 15:29:42 CEST HINT: If this has occurred more than once some data may be corrupted and you may need to choose an earlier recovery target. LOG: impossibile aprire il file "pg_xlog/00010009" (file registro 0, segmento 9): No such file or directory LOG: invalid primary checkpoint record LOG: impossibile aprire il file "pg_xlog/00010021" (file registro 0, segmento 33): No such file or directory LOG: invalid secondary checkpoint record PANIC: impossibile localizzare un checkpoint record valido LOG: startup process (PID 10824) was terminated by signal 6 LOG: aborting startup due to startup process failure infact pg_xlog/ of the slave contains only RECOVERHISTORY: cat pg_xlog/RECOVERYHISTORY START WAL LOCATION: 0/920 (file 00010009) STOP WAL LOCATION: 0/A00 (file 0001000A) CHECKPOINT LOCATION: 0/920 START TIME: 2007-04-16 15:29:42 CEST LABEL: label STOP TIME: 2007-04-16 15:31:39 CEST Did I do something wrong? Thanks for your help, Nico ---(end of broadcast)--- TIP 6: explain analyze is your friend
Re: [GENERAL] Help setting up warm standby replication
On 4/16/07, Nico Sabbi <[EMAIL PROTECTED]> wrote: Is there any parameter that I have to pass to the second server to keep on requesting WALs? I still don't understand what instructs the server to continously request the master's logs. google pg_standby. I've set it up and it works. merlin ---(end of broadcast)--- TIP 9: In versions below 8.0, the planner will ignore your desire to choose an index scan if your joining column's datatypes do not match
Re: [GENERAL] Help setting up warm standby replication
Tom Lane wrote: Nico Sabbi <[EMAIL PROTECTED]> writes: To begin with I followed the example of the docs: in the recovery.conf file of the slave instance I set restore_command = 'cp -av /var/lib/pgsql/data/pg_xlog/%f %p' Hm, it looks like you are trying to copy xlog segments straight from the pg_xlog directory of the master? That's not going to work, because the master will rename/overwrite those files as soon as it thinks it's done with them. You need to have an archive_command on the master that is really truly copying the data to somewhere else, and then the restore_command should copy from the somewhere else. after having set up shared directory (one for archive_command and one for recover_command) I could successfully start and sync the slave dbms, but after the initial recovery the file recovery.conf is renamed to recovery.done and no more updates/WAL are asked (the slave's log reads: LOG: archive recovery complete LOG: system database ready ) Is there any parameter that I have to pass to the second server to keep on requesting WALs? I still don't understand what instructs the server to continously request the master's logs. ---(end of broadcast)--- TIP 5: don't forget to increase your free space map settings
Re: [GENERAL] Help setting up warm standby replication
Nico Sabbi <[EMAIL PROTECTED]> writes: > To begin with I followed the example of the docs: in the recovery.conf > file of the > slave instance I set > restore_command = 'cp -av /var/lib/pgsql/data/pg_xlog/%f %p' Hm, it looks like you are trying to copy xlog segments straight from the pg_xlog directory of the master? That's not going to work, because the master will rename/overwrite those files as soon as it thinks it's done with them. You need to have an archive_command on the master that is really truly copying the data to somewhere else, and then the restore_command should copy from the somewhere else. regards, tom lane ---(end of broadcast)--- TIP 4: Have you searched our list archives? http://archives.postgresql.org/