On Mon, Mar 2, 2020 at 3:36 PM Andrew Dunstan <andrew.duns...@2ndquadrant.com> wrote: > > > On 3/2/20 2:05 AM, Tom Lane wrote: > > Andrew Dunstan <andrew.duns...@2ndquadrant.com> writes: > >> On Sat, Feb 29, 2020 at 10:41 AM Tom Lane <t...@sss.pgh.pa.us> wrote: > >>> Did the buildfarm adjustments get made? (I'm assuming Andrew knows) > >> It's in Release 11 of the client and is mentioned in the release > >> notes. The release is dated last September. Making this change would > >> force a flag day update to that version for the buildfarm client, > > Hm, so scraping the buildfarm logs shows that we currently have this many > > animals reporting (on HEAD) for each client script_version: > > > > 42 'REL_11' > > 64 'REL_10' > > 2 'REL_9' > > 7 'REL_8' > > 1 'REL_7' > > 1 'REL_4.15' > > > > Looks like requiring REL_11 would still be a pretty large ask. > > > > > > > > Possibly, although there are a few individuals who account for a > significant number of the lagging instances. I'll send out some emails > and see if we can improve the situation.
It certainly seems worth a try at a push at least. It's kind of sad if our testing framework prevents us from shipping with less insecure defaults. //Magnus