Thread: PostgreSQL - Debian - Processes: postgres: writer process, etc.
Hello,
I am running the Debian Lenny PostgreSQL package (v.8.4).
When I issue the following command:
I am wondering what are these four last processes ("postgres: writer",
"postgres: wal writer", "postgres: autovacuum launcher" and "postgres: stats
collector")?
Their PPID is 3069.
What is weird to me is the "COMMAND" column as far as these four processes
are concerned.
In the "COMMAND "column I would expect a real command, not a description.
Can you explain?
(BTW, this is a novice question I'm asking...)
Thanks and best regards.
--
View this message in context:
http://postgresql.1045698.n5.nabble.com/PostgreSQL-Debian-Processes-postgres-writer-process-etc-tp5441003p5441003.html
Sent from the PostgreSQL - novice mailing list archive at Nabble.com.
Re: PostgreSQL - Debian - Processes: postgres: writer process, etc.
From
"Jean-Yves F. Barbier"
Date:
On Mon, 30 Jan 2012 02:24:08 -0800 (PST)
Windybell <lmhelp1@orange.fr> wrote:
> I am running the Debian Lenny PostgreSQL package (v.8.4).
> When I issue the following command:
I guess the command is: ps axuf|grep postgres
> I am wondering what are these four last processes ("postgres: writer",
> "postgres: wal writer", "postgres: autovacuum launcher" and "postgres: stats
> collector")?
Because, as their name suggest it, they take care of Pg different
parts.
> Their PPID is 3069.
That can't be, each running software/process have a different PID.
This PID Nb is in the 2nd column.
> What is weird to me is the "COMMAND" column as far as these four processes
> are concerned.
> In the "COMMAND "column I would expect a real command, not a description.
> Can you explain?
If you use the command I wrote above, you'll see they all depend on
the main process launch by the script in /etc/init.d.
--
Eat shit -- billions of flies can't be wrong.
"Jean-Yves F. Barbier" <12ukwn@gmail.com> writes:
> On Mon, 30 Jan 2012 02:24:08 -0800 (PST)
> Windybell <lmhelp1@orange.fr> wrote:
>> I am wondering what are these four last processes ("postgres: writer",
>> "postgres: wal writer", "postgres: autovacuum launcher" and "postgres: stats
>> collector")?
> Because, as their name suggest it, they take care of Pg different
> parts.
Right.
>> Their PPID is 3069.
> That can't be, each running software/process have a different PID.
> This PID Nb is in the 2nd column.
He said PPID, that is *parent* process ID, so the above is expected ---
they should all be children of the same postmaster process.
>> What is weird to me is the "COMMAND" column as far as these four processes
>> are concerned.
>> In the "COMMAND "column I would expect a real command, not a description.
>> Can you explain?
We intentionally change the command string reported by ps for each
postmaster child process, so that you can tell the various processes
apart. If we did not, every postmaster child would show the exact
same command string as the postmaster itself was launched with.
(To get an idea of how useless that is, try looking at "top" sometime
--- on most systems I've worked with, it ignores the changed command
strings, so you can't tell what's what.)
regards, tom lane
You don't have to guess, you just have to read well ; the command I issued is:Thank you for the additional " f " option which is more illustrative. Again, if you read well, I didn't write "PID" but "PPID". The four processes have the same PPID. Jiff wrote > If you use the command I wrote above, you'll see they all depend on > the main process launch by the script in /etc/init.d. It's what I meant to say by writing that the four processes have the same PPID. But this sentence of yours helps. I guess my question wasn't that clear. I couldn't find a detailled documentation about the PostgreSQL server architecture. Best regards. -- Novice doesn't mean neither stupid nor ignorant but most people confuse the two. -- View this message in context: http://postgresql.1045698.n5.nabble.com/PostgreSQL-Debian-Processes-postgres-writer-process-etc-tp5441003p5441900.html Sent from the PostgreSQL - novice mailing list archive at Nabble.com.
@tom lane Thank you for your explanation. It's much clearer now. Best regards. -- View this message in context: http://postgresql.1045698.n5.nabble.com/PostgreSQL-Debian-Processes-postgres-writer-process-etc-tp5441003p5441908.html Sent from the PostgreSQL - novice mailing list archive at Nabble.com.
Still investigating... -------------------------------------------------------- About the "postgres: autovacuum launcher process" -------------------------------------------------------- The "postgres: autovacuum launcher process" daemon is running because the cluster "autovacuum" parameter default value is "on". Documentation: http://www.postgresql.org/docs/8.4/static/routine-vacuuming.html http://www.postgresql.org/docs/8.4/static/runtime-config-autovacuum.html (If you set this parameter to "false" and restart the server, the daemon isn't launched anymore). BR. -- View this message in context: http://postgresql.1045698.n5.nabble.com/PostgreSQL-Debian-Processes-postgres-writer-process-etc-tp5441003p5447236.html Sent from the PostgreSQL - novice mailing list archive at Nabble.com.
On Mon, Jan 30, 2012 at 11:05:10AM -0500, Tom Lane wrote: > >> What is weird to me is the "COMMAND" column as far as these four processes > >> are concerned. > >> In the "COMMAND "column I would expect a real command, not a description. > >> Can you explain? > > We intentionally change the command string reported by ps for each > postmaster child process, so that you can tell the various processes > apart. If we did not, every postmaster child would show the exact > same command string as the postmaster itself was launched with. > (To get an idea of how useless that is, try looking at "top" sometime > --- on most systems I've worked with, it ignores the changed command > strings, so you can't tell what's what.) I belive the 'c' command in modern versions of top toggles display of the current command string. -- Bruce Momjian <bruce@momjian.us> http://momjian.us EnterpriseDB http://enterprisedb.com + It's impossible for everything to be true. +
On Wed, Feb 15, 2012 at 11:47:58AM +1300, Gavin Flower wrote: > On 15/02/12 06:16, Bruce Momjian wrote: > > On Mon, Jan 30, 2012 at 11:05:10AM -0500, Tom Lane wrote: > > What is weird to me is the "COMMAND" column as far as these four processes > are concerned. > In the "COMMAND "column I would expect a real command, not a description. > Can you explain? > > We intentionally change the command string reported by ps for each > postmaster child process, so that you can tell the various processes > apart. If we did not, every postmaster child would show the exact > same command string as the postmaster itself was launched with. > (To get an idea of how useless that is, try looking at "top" sometime > --- on most systems I've worked with, it ignores the changed command > strings, so you can't tell what's what.) > > I belive the 'c' command in modern versions of top toggles display of > the current command string. > > > Yes, pressing 'c' when top is running gives a fuller description of the process > command string, had done in Fedora for longer than I can remember. So I would > be surprised if any Linux distribution had not had it for several years. What is also interesting is that it remembers your previous 'c' mode when you restart top. -- Bruce Momjian <bruce@momjian.us> http://momjian.us EnterpriseDB http://enterprisedb.com + It's impossible for everything to be true. +
On 15/02/12 06:16, Bruce Momjian wrote:
you can also try
ps aux | grep postgres
at the command line prompt
$ ps aux | grep postgres
postgres 1777 0.0 0.8 2352592 66692 ? S 11:21 0:01 /usr/bin/postgres -D /var/lib/pgsql/data -p 5432
postgres 2979 0.0 0.0 181024 952 ? Ss 11:21 0:00 postgres: logger process
postgres 2989 0.0 0.1 2353620 14532 ? Ss 11:21 0:00 postgres: writer process
postgres 2990 0.0 0.0 2353620 1092 ? Ss 11:21 0:00 postgres: wal writer process
postgres 2991 0.0 0.0 2354356 2380 ? Ss 11:21 0:00 postgres: autovacuum launcher process
postgres 2992 0.0 0.0 183120 1160 ? Ss 11:21 0:00 postgres: stats collector process
gavin 5450 0.0 0.0 109248 848 pts/13 S+ 11:45 0:00 grep --color=auto postgres
$
Yes, pressing 'c' when top is running gives a fuller description of the process command string, had done in Fedora for longer than I can remember. So I would be surprised if any Linux distribution had not had it for several years.On Mon, Jan 30, 2012 at 11:05:10AM -0500, Tom Lane wrote:What is weird to me is the "COMMAND" column as far as these four processes are concerned. In the "COMMAND "column I would expect a real command, not a description. Can you explain?We intentionally change the command string reported by ps for each postmaster child process, so that you can tell the various processes apart. If we did not, every postmaster child would show the exact same command string as the postmaster itself was launched with. (To get an idea of how useless that is, try looking at "top" sometime --- on most systems I've worked with, it ignores the changed command strings, so you can't tell what's what.)I belive the 'c' command in modern versions of top toggles display of the current command string.
you can also try
ps aux | grep postgres
at the command line prompt
$ ps aux | grep postgres
postgres 1777 0.0 0.8 2352592 66692 ? S 11:21 0:01 /usr/bin/postgres -D /var/lib/pgsql/data -p 5432
postgres 2979 0.0 0.0 181024 952 ? Ss 11:21 0:00 postgres: logger process
postgres 2989 0.0 0.1 2353620 14532 ? Ss 11:21 0:00 postgres: writer process
postgres 2990 0.0 0.0 2353620 1092 ? Ss 11:21 0:00 postgres: wal writer process
postgres 2991 0.0 0.0 2354356 2380 ? Ss 11:21 0:00 postgres: autovacuum launcher process
postgres 2992 0.0 0.0 183120 1160 ? Ss 11:21 0:00 postgres: stats collector process
gavin 5450 0.0 0.0 109248 848 pts/13 S+ 11:45 0:00 grep --color=auto postgres
$