Re: pg13.2: invalid memory alloc request size NNNN - Mailing list pgsql-hackers

From Greg Stark
Subject Re: pg13.2: invalid memory alloc request size NNNN
Date
Msg-id CAM-w4HNzCQeq+qij4oGNGXYmNykD249JTht7DPRER3T1MQySgA@mail.gmail.com
Whole thread Raw
In response to Re: pg13.2: invalid memory alloc request size NNNN  (Tomas Vondra <tomas.vondra@enterprisedb.com>)
List pgsql-hackers
I think to get a size of -4 you would be trying to read a varlena
pointer pointing to four nul bytes. I bet if you run dd on the
corresponding block you'll find a chunk of nuls in the page. That
perhaps makes sense with ZFS where if a new page was linked to the
tree but never written it would be an uninitialized page rather than
the old data.

I'm becoming increasingly convinced that there are a lot of storage
systems out there that just lose data whenever they crash or lose
power. Systems that are supposed to be better than that.



pgsql-hackers by date:

Previous
From: Amit Kapila
Date:
Subject: Re: logical replication seems broken
Next
From: "Joel Jacobson"
Date:
Subject: Re: [HACKERS] GSoC 2017: Foreign Key Arrays