Re: How different is AWS-RDS postgres?

On 5/26/21 1:24 PM, Rob Sargent wrote:
>
>>> My question is:
>>> Should I be surprised that the initial bulk loaded table is nowhere to
>>> be seen, given the thumbs up from the logs?  Is this frailty inherent in
>>> AWS/RDS infrastructure?
>>>
>>> Since this is an academic exorcise, I have minimal AWS support, which
>>> has yet to chime in on this matter. My access to the logs is um, er,
>>> constrained.
>>
>> The big differences I notice are:
>>
>> 1. "postgres" is not a superuser,
>> 2. viewing logs is a hassle.
>>
>> Otherwise, they're really similar.  We've pumped about 6TB of data into
>> an instance, and it's been rock solid.  JBoss is quite happy, and there
>> haven't been any problems.
>>
>> --
>> Angular momentum makes the world go 'round.
> Good to hear.
> I pushed this same data set into our previous server (Version 10) so I was
> surprised it blew up this time, though in that case these separate files
> were more spread out in arrival time. Nothing in the (available) logs
> other that lots of auto-vacuuming  and the suggestion to increase wal size
> (which I thought a little surprising).  I really would like to know just
> what went south but that isn't obvious yet - OOM?, network?, disk full?
> but I can't find any smoking gun on the AWS/RDS pages.

The difficulty in viewing logs needs to be emphasized.

--
Angular momentum makes the world go 'round.