Thread: [HACKERS] pg_upgrade vs. pg_ctl stop -m fast
In 9.5, the default pg_ctl stop mode was changed from "smart" to "fast".
In pg_upgrade, there is this code:
void
stop_postmaster(bool fast)
{ ... exec_prog(SERVER_STOP_LOG_FILE, NULL, !fast, "\"%s/pg_ctl\" -w -D \"%s\" -o \"%s\" %s stop",
cluster->bindir, cluster->pgconfig, cluster->pgopts ? cluster->pgopts : "", fast ? "-m
fast": ""); ...
}
So, when upgrading from 9.5 or later, code that requested a non-fast
shutdown would now always get a fast shutdown.
I think the last line should be changed to something like
fast ? "-m fast" : "-m smart");
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
Peter Eisentraut <peter.eisentraut@2ndquadrant.com> writes:
> In 9.5, the default pg_ctl stop mode was changed from "smart" to "fast".
> In pg_upgrade, there is this code:
> ...
> I think the last line should be changed to something like
> fast ? "-m fast" : "-m smart");
Ugh. Clear oversight.
There is maybe room for a separate discussion about whether pg_upgrade
*should* be using fast mode, but if so we could remove the "bool fast"
argument from this function altogether.
regards, tom lane
On Thu, Jan 12, 2017 at 10:17:52AM -0500, Tom Lane wrote: > Peter Eisentraut <peter.eisentraut@2ndquadrant.com> writes: > > In 9.5, the default pg_ctl stop mode was changed from "smart" to "fast". > > In pg_upgrade, there is this code: > > ... > > I think the last line should be changed to something like > > fast ? "-m fast" : "-m smart"); > > Ugh. Clear oversight. > > There is maybe room for a separate discussion about whether pg_upgrade > *should* be using fast mode, but if so we could remove the "bool fast" > argument from this function altogether. Agreed, it should be remove. Should I do it? -- Bruce Momjian <bruce@momjian.us> http://momjian.us EnterpriseDB http://enterprisedb.com + As you are, so once was I. As I am, so you will be. + + Ancient Roman grave inscription +
On 1/12/17 9:43 PM, Bruce Momjian wrote: > On Thu, Jan 12, 2017 at 10:17:52AM -0500, Tom Lane wrote: >> Peter Eisentraut <peter.eisentraut@2ndquadrant.com> writes: >>> In 9.5, the default pg_ctl stop mode was changed from "smart" to "fast". >>> In pg_upgrade, there is this code: >>> ... >>> I think the last line should be changed to something like >>> fast ? "-m fast" : "-m smart"); >> >> Ugh. Clear oversight. >> >> There is maybe room for a separate discussion about whether pg_upgrade >> *should* be using fast mode, but if so we could remove the "bool fast" >> argument from this function altogether. > > Agreed, it should be remove. Should I do it? For 9.5 and 9.6, I think we should backpatch what I suggested above, to minimize the behavior change. For master we can consider removing the distinction and just use fast shutdown all the time, but I haven't checked all the possible implications of that change. -- Peter Eisentraut http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
Peter Eisentraut <peter.eisentraut@2ndquadrant.com> writes:
> On 1/12/17 9:43 PM, Bruce Momjian wrote:
>> On Thu, Jan 12, 2017 at 10:17:52AM -0500, Tom Lane wrote:
>>> Peter Eisentraut <peter.eisentraut@2ndquadrant.com> writes:
>>>> I think the last line should be changed to something like
>>>> fast ? "-m fast" : "-m smart");
>>> There is maybe room for a separate discussion about whether pg_upgrade
>>> *should* be using fast mode, but if so we could remove the "bool fast"
>>> argument from this function altogether.
>> Agreed, it should be remove. Should I do it?
> For 9.5 and 9.6, I think we should backpatch what I suggested above, to
> minimize the behavior change. For master we can consider removing the
> distinction and just use fast shutdown all the time, but I haven't
> checked all the possible implications of that change.
That sounds sensible to me.
regards, tom lane
On 1/13/17 9:45 AM, Peter Eisentraut wrote: > On 1/12/17 9:43 PM, Bruce Momjian wrote: >> On Thu, Jan 12, 2017 at 10:17:52AM -0500, Tom Lane wrote: >>> Peter Eisentraut <peter.eisentraut@2ndquadrant.com> writes: >>>> In 9.5, the default pg_ctl stop mode was changed from "smart" to "fast". >>>> In pg_upgrade, there is this code: >>>> ... >>>> I think the last line should be changed to something like >>>> fast ? "-m fast" : "-m smart"); >>> >>> Ugh. Clear oversight. >>> >>> There is maybe room for a separate discussion about whether pg_upgrade >>> *should* be using fast mode, but if so we could remove the "bool fast" >>> argument from this function altogether. >> >> Agreed, it should be remove. Should I do it? > > For 9.5 and 9.6, I think we should backpatch what I suggested above, to > minimize the behavior change. I have committed that (including to master). > For master we can consider removing the > distinction and just use fast shutdown all the time, but I haven't > checked all the possible implications of that change. I'm not planning to work on this at this time. -- Peter Eisentraut http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services