My question is for Francisco who replied regarding xz. I was curious what options he used. Thanks.
On Fri, Oct 16, 2015 at 3:14 PM, Adrian Klaver <adrian.kla...@aklaver.com> wrote: > On 10/16/2015 12:10 PM, anj patnaik wrote: > >> Thanks. what is the recommended command/options for backup and how to >> restore? >> >> I found the below online. let me know if this is better and how to >> restore. Thank you >> >> pg_dump -Fc '<Db-Name>' | xz -3 dump.xz >> > > Again, why would compress an already compressed output? > > Also online: > > http://www.postgresql.org/docs/9.4/interactive/app-pgdump.html > > http://www.postgresql.org/docs/9.4/interactive/app-pgrestore.html > > They step you through the backup and restore process. > >> >> >> On Fri, Oct 16, 2015 at 4:05 AM, Francisco Olarte >> <fola...@peoplecall.com <mailto:fola...@peoplecall.com>> wrote: >> >> On Fri, Oct 16, 2015 at 8:27 AM, Guillaume Lelarge >> <guilla...@lelarge.info <mailto:guilla...@lelarge.info>> wrote: >> > 2015-10-15 23:05 GMT+02:00 Adrian Klaver <adrian.kla...@aklaver.com >> <mailto:adrian.kla...@aklaver.com>>: >> >> On 10/15/2015 01:35 PM, anj patnaik wrote: >> ... >> >>> ./pg_dump -t RECORDER -Fc postgres | gzip > /tmp/dump >> >>> Are there any other options for large tables to run faster and >> occupy >> >>> less disk space? >> >> Yes, do not double compress. -Fc already compresses the file. >> > Right. But I'd say "use custom format but do not compress with >> pg_dump". Use >> > the -Z0 option to disable compression, and use an external >> multi-threaded >> > tool such as pigz or pbzip2 to get faster and better compression. >> >> Actually I would not recommend that, unless you are making a long term >> or offsite copy. Doing it means you need to decompress the dump before >> restoring or even testing it ( via i.e., pg_restore > /dev/null ). >> >> And if you are pressed on disk space you may corner yourself using >> that on a situation where you do NOT have enough disk space for an >> uncompressed dump. Given you normally are nervous enough when >> restoring, for normal operations I think built in compression is >> better. >> >> Also, I'm not current with the compressor Fc uses, I think it still is >> gzip, which is not that bad and is normally quite fast ( In fact I do >> not use that 'pbzip2', but I did some tests about a year ago and I >> found bzip2 was beaten by xz quite easily ( That means on every level >> of bzip2 one of the levels of xz beat it in BOTH size & time, that was >> for my data, YMMV ). >> >> >> Francisco Olarte. >> >> >> > > -- > Adrian Klaver > adrian.kla...@aklaver.com >
