When we run pg_dump on our database, our web site becomes completely
unresponsive. I thought pg_dump was runnable while the database was
still being actively used? Anyway, I'm not entirely sure why, but
here's what I'm seeing.
pg_dump -v database_name | gzip > output_file
25% to 50% CPU usage (4 proc machine)
Entire database is cached in memory, so the only I/O is what's being
dumped to the output file
Web site becomes unresponsive almost immediately
Does pg_dump create table locks? It doesn't look like an I/O problem
as far as I can tell...
sdb: tables
sdc: logs
sdd: backup share
----total-cpu-usage----
-dsk/total----dsk/sdc-----dsk/sdb-----dsk/sdd--
------memory-usage----- -net/total- ----swap---
usr sys idl wai hiq siq|_read _writ:_read _writ:_read _writ:_read
_writ|_used _buff _cach _free|_recv _send|_used _free
34 1 64 0 0 0| 0 0 : 0 0 : 0 0 : 0 0
| 525M 96k 14G 1276M| 140B 556B| 132k 7632M
36 1 63 0 0 0| 0 83k: 0 0 : 0 67k: 0 0
| 525M 96k 14G 1272M|1176B 48k| 132k 7632M
34 1 64 0 0 0| 0 0 : 0 0 : 0 0 : 0 0
| 525M 96k 14G 1269M| 140B 556B| 132k 7632M
36 1 63 0 0 0| 0 0 : 0 0 : 0 0 : 0 0
| 525M 96k 14G 1265M| 564B 9824B| 132k 7632M
35 1 64 0 0 0| 0 0 : 0 0 : 0 0 : 0 0
| 525M 96k 14G 1261M| 140B 556B| 132k 7632M
36 1 63 0 0 0| 0 2048B: 0 0 : 0 0 : 0 0
| 525M 96k 14G 1258M| 140B 556B| 132k 7632M
36 4 49 10 0 1| 0 60M: 0 2048B: 0 184k: 0
60M| 535M 96k 14G 1245M|6827B 8682B| 132k 7632M
48 4 39 8 0 1| 0 46M: 0 168k: 0 0 : 0
46M| 542M 96k 14G 1234M| 652k 119M| 132k 7632M
38 2 52 7 0 0| 0 15M: 0 608k: 0 0 : 0
14M| 543M 96k 14G 1230M| 56k 24k| 132k 7632M
Thanks,
Bryan
---------------------------(end of broadcast)---------------------------
TIP 3: Have you checked our extensive FAQ?
http://www.postgresql.org/docs/faq
