Amit Kapila <amit.kapila@huawei.com> writes:
>> AFAIR you can create pg_control from scratch already with pg_resetxlog.
>> The hard part is coming up with values for the counters, such as the
>> next WAL location. Some of them such as next OID are pretty harmless
>> if you don't guess right, but I'm worried that wrong next WAL could
>> make things worse not better.
> I believe if WAL files are proper as mentioned in Alvaro's mail, the
> purposed logic should generate correct values.
I've got a problem with the assumption that, when pg_control is trash,
megabytes or gigabytes of WAL can still be relied on completely.
I'm almost inclined to suggest that we not get next-LSN from WAL, but
by scanning all the pages in the main data store and computing the max
observed LSN. This is clearly not very attractive from a performance
standpoint, but it would avoid the obvious failure mode where you lost
some recent WAL segments along with pg_control.
regards, tom lane
