Re: Strange path from pgarch_readyXlog() - Mailing list pgsql-hackers

From Bossart, Nathan
Subject Re: Strange path from pgarch_readyXlog()
Date
Msg-id 0678CBBA-8209-41E9-9A53-ACE63D71CE10@amazon.com
Whole thread Raw
In response to Re: Strange path from pgarch_readyXlog()  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: Strange path from pgarch_readyXlog()  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-hackers
On 12/29/21, 1:04 PM, "Tom Lane" <tgl@sss.pgh.pa.us> wrote:
> "Bossart, Nathan" <bossartn@amazon.com> writes:
>> I bet this was a simple mistake in beb4e9b.
>
>> -static char arch_filenames[NUM_FILES_PER_DIRECTORY_SCAN][MAX_XFN_CHARS];
>> +static char arch_filenames[NUM_FILES_PER_DIRECTORY_SCAN][MAX_XFN_CHARS + 1];
>
> Hm, yeah, that looks like a pretty obvious bug.
>
> While we're here, I wonder if we ought to get rid of the static-ness of
> these arrays.  I realize that they're only eating a few kB, but they're
> doing so in every postgres process, when they'll only be used in the
> archiver.

This crossed my mind, too.  I also think one of the arrays can be
eliminated in favor of just using the heap (after rebuilding with a
reversed comparator).  Here is a minimally-tested patch that
demonstrates what I'm thinking.  

Nathan


Attachment

pgsql-hackers by date:

Previous
From: Daniel Gustafsson
Date:
Subject: Re: Adding CI to our tree
Next
From: SATYANARAYANA NARLAPURAM
Date:
Subject: Re: Logging replication state changes