> On Jun 11, 2018, at 1:57 PM, Pavan Teja <pavan.postgresdba@gmail.com> wrote:
>
> So finally there's no script to determine corruption well in advance?? Correct??
It is your responsibility to make sure that the system is solid and all worst cases are covered along with striving for
thefive 9’s. You need to building a system that you can trust which includes making sure you disk subsystem is really
andI mean really reliable. RAID is the most falsely trusted system around; so you really need to know the subsystems
inyour system especially when subsystems that break fsync().
Bit rot is real and if you system doesn’t handle it then that will lead to data corruption. Postgres will only be able
totell you that it occurred if you have data_checksums feature enabled. Your subsystem should handle it and actively
bechecking for it; however, most RAID systems don’t or fail to do a good job at it.
Like I stated already; If I couldn’t entrust the data integrity of Postgres I would not be using it and would be
runningOracle instead.
